A Comparative AI–GIS Spatiotemporal Analysis of Child Health Outcomes in New Zealand and Nigeria (2006–2022): Implications for Equity-Driven Health Policy

A Comparative AI–GIS Spatiotemporal Analysis of Child Health Outcomes in New Zealand and Nigeria (2006–2022): Implications for Equity-Driven Health Policy | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Comparative AI–GIS Spatiotemporal Analysis of Child Health Outcomes in New Zealand and Nigeria (2006–2022): Implications for Equity-Driven Health Policy Jumoke I. Ogunremi, Nelson N. Igwilo, Oladayo O. Babalola This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8443273/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Child health is a critical indicator of human development and health system performance. Despite global progress, inequities in child health outcomes remain stark, particularly between high‑income and low‑middle‑income countries. This study applies an integrated Artificial Intelligence Geographic Information Systems (AI–GIS) framework to compare child health trajectories in New Zealand and Nigeria from 2006 to 2022, focusing on spatial inequality, temporal change, and policy effectiveness. Using harmonized national datasets, we examine immunization coverage, sanitation access, and treatment of childhood illnesses in relation to maternal education, household wealth, and place of residence. Results reveal two contrasting inequality regimes: in Nigeria, child health outcomes are shaped by structural inequities linked to governance capacity and infrastructure distribution, while in New Zealand, residual disparities persist within an otherwise mature health system. By distinguishing structural from residual inequality, the study highlights how policy design must adapt to context whether addressing entrenched deprivation in resource‑constrained settings or fine‑tuning equity in advanced systems. Findings underscore the importance of equity‑driven health policy and provide evidence for strengthening child health strategies in both developing and developed contexts. Artificial Intelligence Geographic Information Systems (AI–GIS) Child Health Inequality Spatiotemporal Analysis Comparative Health Systems Policy Effectiveness Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction 1.1 Background and Rationale Child health is universally recognized as a core indicator of social development, human capital formation, and long-term economic productivity. Metrics such as under-five mortality, immunization coverage, access to sanitation, and treatment of common childhood illnesses are routinely used by global institutions to assess national progress toward the Sustainable Development Goals (SDGs), particularly SDG 3 (Good Health and Well-Being) (United Nations, 2015 ). Improvements in child health are strongly associated with enhanced educational attainment, labor productivity, and intergenerational poverty reduction, making it a foundational pillar of sustainable development (Victora et al., 2016 ). Despite global progress over recent decades, child health outcomes remain highly uneven across income groups, with low- and middle-income countries (LMICs) continuing to shoulder a disproportionate burden of preventable child morbidity and mortality (UNICEF, 2023 ; World Health Organization [WHO], 2022). These disparities reflect deep structural inequalities in health systems, socioeconomic conditions, governance capacity, and access to essential services. High-income countries have largely transitioned from survival-focused child health challenges to equity- and quality-focused concerns, while many developing countries continue to struggle with basic coverage gaps in immunization, nutrition, water, sanitation, and primary healthcare (World Bank, 2023 ). This divergence underscores the importance of comparative research frameworks that can systematically evaluate not only outcome differences but also the spatial and temporal mechanisms through which child health gains are achieved or constrained across development contexts. A comparative North–South analytical framework is therefore well suited for advancing understanding of child health inequality. Positioning New Zealand as a high-income benchmark offers a reference case characterized by near-universal health coverage, strong social protection systems, and relatively low spatial variability in child health outcomes (OECD, 2022 ). In contrast, Nigeria, as a lower-middle-income country with pronounced regional, socioeconomic, and infrastructural disparities, represents a compelling developing-country context in which child health outcomes vary markedly across states and population groups (UNICEF, 2023 ). Nigeria alone accounts for a substantial share of global under-five deaths, with outcomes closely linked to maternal education, household wealth, and place of residence (WHO, 2022). Comparing these two contexts enables a clear examination of how structural capacity, policy coherence, and spatial equity shape child health trajectories over time. Traditional comparative studies, however, often rely on national averages that obscure subnational inequalities and fail to capture dynamic changes across space and time. The integration of Artificial Intelligence (AI) with Geographic Information Systems (GIS), commonly referred to as GeoAI, addresses this limitation by enabling the simultaneous analysis of spatial patterns, temporal trends, and predictive relationships within complex health datasets (Kamel Boulos & Peng, 2019 ). GIS facilitates the visualization and quantification of geographic disparities in child health indicators, while AI models such as Random Forests and Long Short-Term Memory networks allow for the identification of nonlinear determinants and the forecasting of future outcomes (Janowicz et al., 2020 ). When applied comparatively, AI–GIS integration supports cross-context evaluation of spatial inequality, the pace of temporal progress, and the effectiveness of policy interventions under differing socioeconomic conditions. Accordingly, adopting an AI–GIS-driven comparative framework between New Zealand and Nigeria provides a robust methodological basis for understanding not only whether child health outcomes differ across income groups, but why these differences persist spatially and how policy effectiveness varies across development contexts. This approach advances global child health research by moving beyond descriptive comparisons toward predictive, spatially explicit, and policy-relevant insights. 1.2 Comparative Problem Statement Despite global commitments to reduce child mortality and improve child well-being, profound contrasts persist between high-income and developing health systems in both the level and distribution of child health outcomes. In high-income settings such as New Zealand, child health indicators—including immunization coverage, access to sanitation, and treatment of childhood illnesses—are generally high and exhibit relatively limited geographic variation. Universal health coverage, strong primary healthcare infrastructure, and coordinated social protection policies have contributed to a health system in which disparities are comparatively narrow and largely driven by marginal socioeconomic or ethnic differences rather than extreme spatial deprivation (OECD, 2022 ; World Health Organization [WHO], 2022). In contrast, Nigeria continues to experience persistent and deeply entrenched spatial and socioeconomic inequalities in child health outcomes. National-level indicators mask substantial subnational variation across states and geopolitical zones, where child health performance is strongly shaped by maternal education, household wealth, rural–urban residence, and regional health system capacity (UNICEF, 2023 ; WHO, 2022). Nigeria accounts for a disproportionately large share of global under-five mortality, yet this burden is unevenly distributed, with some regions exhibiting outcomes comparable to middle-income countries while others remain characterized by chronic deprivation and weak service coverage. These disparities are not static; they evolve over time in response to policy shifts, economic volatility, and demographic pressures, underscoring the need for analytical approaches that capture both spatial and temporal dynamics. A central problem in comparative child health research is the continued reliance on national averages as the primary basis for cross-country assessment. While national indicators are useful for global monitoring, they obscure localized “hotspots” of poor performance and conceal the mechanisms through which inequality persists within countries (Victora et al., 2016 ). This limitation is particularly problematic in large, heterogeneous developing countries such as Nigeria, where subnational disparities often exceed differences observed between countries. Even in relatively homogeneous systems like New Zealand, national averages may conceal smaller but policy-relevant spatial gradients affecting specific communities. As a result, comparisons based solely on national metrics provide an incomplete and potentially misleading picture of health system performance and policy effectiveness. The integration of Artificial Intelligence (AI) with Geographic Information Systems (GIS) offers a methodological pathway to address this limitation. GIS enables the spatial disaggregation and visualization of child health indicators, revealing geographic patterns of inequality that national statistics fail to capture. AI techniques such as machine learning regression models and deep learning time-series architectures extend this capability by identifying nonlinear relationships, quantifying the relative influence of socioeconomic determinants, and forecasting future health trajectories (Janowicz et al., 2020 ; Kamel Boulos & Peng, 2019 ). When applied together, AI–GIS frameworks support subnational, spatiotemporal comparison of health outcomes, allowing researchers to assess how quickly regions improve, where progress stagnates, and how policy interventions translate into spatially differentiated outcomes. However, despite rapid advances in GeoAI, a critical gap remains in comparative applications between developed and developing health systems. Existing GeoAI studies have largely focused on single-country analyses, specific diseases, or crisis contexts, with limited attention to structured North–South comparisons that evaluate how spatial inequality, temporal progress, and policy effectiveness differ across development contexts The absence of such comparative GeoAI frameworks limits the ability of policymakers to distinguish context-specific challenges from transferable best practices and constrains the global relevance of AI-driven health analytics. Accordingly, there is a clear need for a comparative, AI–GIS driven approach that contrasts a relatively homogeneous high-income health system with a spatially fragmented developing one. By systematically comparing New Zealand and Nigeria at subnational and temporal scales, this study addresses an important methodological and empirical gap, enabling a more nuanced understanding of how structural capacity, spatial equity, and policy coherence shape child health outcomes across divergent development trajectories. 1.3 Aim and Comparative Objectives The overarching aim of this study is to conduct a rigorous comparative spatiotemporal analysis of child health trajectories between New Zealand and Nigeria over the period 2006–2022, using an integrated Artificial Intelligence Geographic Information Systems (AI–GIS) framework. By juxtaposing a high-income health system with relatively homogeneous outcomes against a developing system characterized by pronounced spatial and socioeconomic heterogeneity, the study seeks to generate policy-relevant insights into how child health progress unfolds across divergent development contexts. Specifically, the first objective is to compare the spatiotemporal trajectories of child health indicators in New Zealand and Nigeria. This involves examining how key indicators evolve over time and how their geographic distribution differs within and between the two countries. The AI–GIS approach enables the identification of spatial clustering, temporal convergence or divergence, and differential rates of improvement across subnational units. Through this objective, the study moves beyond national averages to reveal whether progress in child health is evenly distributed or concentrated in specific regions, and how these patterns contrast between a relatively stable, high-capacity health system and a spatially fragmented developing one. The second objective is to evaluate differences in the magnitude and spatial expression of socioeconomic determinants of child health across both countries. Focusing on maternal education, household wealth, and place of residence (urban–rural), the study assesses how strongly these determinants influence child health outcomes and how their effects vary geographically. In New Zealand, where baseline service coverage is high, these determinants are expected to exhibit attenuated spatial gradients. In contrast, in Nigeria, the same determinants are hypothesized to produce pronounced spatial inequalities, reflecting uneven access to services and structural disparities. This objective enables a comparative assessment of how social determinants interact with geography to shape child health outcomes under contrasting institutional and economic conditions. The third objective is to assess the transferability and robustness of predictive AI models across contrasting health systems. By applying the same AI architectures such as machine learning regression and time-series forecasting models to both countries, the study evaluates whether models trained within one context retain predictive validity in another. This objective addresses a critical methodological question in GeoAI research: whether predictive models developed in data-rich, stable systems can be meaningfully adapted to data-constrained, heterogeneous environments, or whether context-specific recalibration is required. The findings will inform the generalizability of AI–GIS tools for global health monitoring and their suitability for cross-country policy learning. Collectively, these aims and objectives position the study to contribute not only to empirical understanding of child health inequalities but also to methodological advancement in comparative GeoAI research, offering insights into spatial equity, temporal progress, and model transferability across development contexts. 2. Methods 2.1 Study Design and Comparative Framework This study adopts a comparative longitudinal research design to examine child health trajectories in New Zealand and Nigeria over the period 2006–2022. A longitudinal approach is essential for capturing changes in child health outcomes over time, assessing the pace and consistency of progress, and identifying periods of acceleration or stagnation in response to policy interventions and socioeconomic transitions (Singer & Willett, 2003 ). By embedding this temporal dimension within a comparative North–South framework, the study enables systematic evaluation of how child health evolves under markedly different institutional capacities, income levels, and health system structures. The comparative framework is explicitly cross-contextual, contrasting a high-income country with relatively homogeneous health outcomes against a developing country characterized by substantial spatial and socioeconomic heterogeneity. New Zealand represents a mature health system with near-universal coverage and strong primary care coordination, where subnational variations in child health are comparatively limited and often linked to residual socioeconomic or demographic factors (OECD, 2022 ). Nigeria, by contrast, exhibits pronounced disparities across its states, driven by uneven distribution of health infrastructure, regional economic inequalities, and differential access to education and basic services (UNICEF, 2023 ; World Health Organization [WHO], 2022). This contrast provides a robust analytical basis for examining how spatial inequality and temporal progress differ across development contexts. The unit of analysis is defined at the subnational level for both countries, allowing for spatially explicit comparison. In New Zealand, the analysis is conducted across subnational regions consistent with administrative and statistical reporting structures, while in Nigeria, the unit of analysis is the state level, reflecting the country’s federal governance system and the primary scale at which health policy implementation and resource allocation occur. Using subnational units mitigates the limitations of national averages, which often conceal localized disparities and mask within-country inequality (Victora et al., 2016 ). This design choice is particularly critical for Nigeria, where inter-state differences in child health outcomes frequently exceed cross-country differences observed at the national level. To operationalize the comparative design, child health outcomes are modeled as spatiotemporal processes. Let \(\:{y}_{i,t}\) denote a child health indicator (e.g., immunization coverage or access to sanitation) for subnational unit \(\:i\) at time \(\:t\) . The longitudinal structure of the data can be expressed as: \(\:{y}_{i,t}=\alpha\:+\beta\:{X}_{i,t}+{\mu\:}_{i}+{\lambda\:}_{t}+{\epsilon\:}_{i,t}\) where \(\:\alpha\:\) is the intercept, \(\:{X}_{i,t}\) is a vector of socioeconomic covariates (education, wealth, residence), \(\:{\mu\:}_{i}\) captures time-invariant subnational effects, \(\:{\lambda\:}_{t}\) represents common temporal shocks or global trends, and \(\:{\epsilon\:}_{i,t}\) is the error term (Wooldridge, 2010 ). This structure allows comparison of both within-unit temporal change and between-unit spatial variation across the two countries. The integration of Artificial Intelligence (AI) and Geographic Information Systems (GIS) strengthens this comparative framework by enabling cross-context analysis of spatial inequality, temporal dynamics, and policy effectiveness. GIS provides the spatial referencing and visualization needed to map subnational disparities and identify clusters of high or low performance, while AI techniques such as machine learning regression and time-series forecasting—model nonlinear relationships and predict future trajectories (Kamel Boulos & Peng, 2019 ; Janowicz et al., 2020 ). Formally, AI-based prediction can be represented as: \(\:{\widehat{y}}_{i,t+k}=f\left({y}_{i,t},{X}_{i,t},{S}_{i}\right)\) where \(\:{\widehat{y}}_{i,t+k}\) is the predicted health outcome \(\:k\) periods ahead, \(\:f(\cdot\:)\) denotes a learned nonlinear function (e.g., Random Forest or LSTM), and \(\:{S}_{i}\) represents spatial attributes derived from GIS layers. Applying the same modeling framework to both countries allows direct assessment of model transferability and differential predictive performance across health systems. Overall, this study design enables a rigorous comparison of child health trajectories by aligning temporal depth, spatial granularity, and analytical methodology across contexts. By combining longitudinal analysis with AI–GIS integration at the subnational level, the framework provides a robust basis for evaluating how spatial inequality, socioeconomic determinants, and policy effectiveness shape child health outcomes in developed and developing health systems. 2.2 Data Sources and Indicators This study draws on harmonized, nationally representative data sources to enable a robust comparative analysis of child health outcomes between Nigeria and New Zealand over the period 2006–2022. The selection of data sources and indicators is guided by three core principles: cross-country comparability, temporal consistency, and suitability for spatially explicit AI–GIS analysis. For Nigeria, the primary data sources are the Demographic and Health Surveys (DHS) and the UNICEF Multiple Indicator Cluster Surveys (MICS). DHS and MICS are globally standardized household survey programs designed to generate reliable, comparable estimates of population health indicators in low- and middle-income countries (Corsi et al., 2012 ; UNICEF, 2021 ). These surveys employ stratified, multistage cluster sampling and provide subnationally disaggregated data at the state or zonal level, making them particularly suitable for spatial and longitudinal analysis. In Nigeria, DHS and MICS constitute the most authoritative and consistently available sources of child health data, covering indicators related to immunization, water and sanitation, treatment of common childhood illnesses, and maternal socioeconomic characteristics (National Population Commission [NPC] & ICF, 2019). For New Zealand, child health data are obtained from national administrative and statistical health datasets, including Ministry of Health surveillance systems and Statistics New Zealand–aligned indicators. These datasets provide comprehensive coverage of child immunization, sanitation access, healthcare utilization, and maternal characteristics at subnational levels, reflecting the country’s well-established health information infrastructure (Ministry of Health New Zealand, 2022 ). To ensure comparability with DHS and MICS indicators, New Zealand data are harmonized using internationally defined indicator frameworks, particularly those aligned with WHO and UNICEF monitoring standards (World Health Organization [WHO], 2022). This harmonization process ensures that observed differences reflect substantive health system and socioeconomic contrasts rather than measurement inconsistencies. The study focuses on a core set of child health and socioeconomic indicators that are widely recognized as proxies for child well-being and health system performance. Immunization coverage, particularly completion of routine childhood vaccinations, is used as an indicator of preventive healthcare access and system reach. Access to improved sanitation reflects environmental health conditions and exposure to infectious disease risks. Treatment of common childhood illnesses, such as diarrhea and respiratory infections, captures responsiveness and accessibility of primary healthcare services. Maternal education is included as a key socioeconomic determinant, given its well-documented influence on health-seeking behavior, childcare practices, and child survival (Victora et al., 2016 ). These indicators are consistently measured across DHS, MICS, and WHO-aligned national systems, enabling valid cross-country comparison. To support spatiotemporal analysis, all indicators are structured as panel data indexed by subnational unit \(\:i\) and time \(\:t\) . Let \(\:{y}_{i,t}\) denote a child health indicator, and let \(\:{Z}_{i,t}\) represent the vector of socioeconomic attributes (maternal education, wealth, residence). The indicator framework can be expressed as: \(\:{y}_{i,t}=g({Z}_{i,t},{H}_{i,t})\) where \(\:{H}_{i,t}\) represents health system and environmental conditions varying across space and time. This structure allows AI models to learn nonlinear relationships between indicators and determinants while preserving spatial referencing for GIS analysis. The integration of Artificial Intelligence (AI) and Geographic Information Systems (GIS) is central to leveraging these data sources for comparative analysis. GIS enables the spatial alignment and visualization of subnational indicators, revealing geographic inequalities and clustering patterns that national summaries obscure. AI methods, applied to harmonized indicators across both countries, allow for identification of differential determinant effects, modeling of temporal progress, and forecasting of future outcomes under contrasting policy environments (Janowicz et al., 2020 ; Kamel Boulos & Peng, 2019 ). By applying the same indicator definitions and analytical pipelines to Nigeria and New Zealand, the AI–GIS framework facilitates cross-context comparison of spatial inequality, temporal trajectories, and policy effectiveness, strengthening the validity and interpretability of the comparative findings. 2.3 AI and GIS Analytical Techniques This study employs an integrated Artificial Intelligence Geographic Information Systems (AI–GIS) analytical pipeline to examine spatial inequality, temporal progress, and policy effectiveness in child health outcomes across New Zealand and Nigeria. The approach combines GIS-based spatial statistics for identifying geographic patterns with AI-based predictive models for cross-sectional explanation and longitudinal forecasting, enabling rigorous cross-context comparison under a unified methodological framework. GIS-Based Spatial Analysis Geographic Information Systems (GIS) techniques are used to quantify and visualize subnational disparities in child health indicators. Two complementary spatial statistics are applied. First, global spatial autocorrelation is assessed using Moran’s I, which measures whether similar values of a health indicator cluster spatially across subnational units (Anselin, 1995 ). For subnational unit \(\:\varvec{i}\) at time \(\:\varvec{t}\) , Moran’s I is expressed as: $$\:\varvec{I}=\frac{\varvec{N}}{\varvec{W}}\frac{\sum\:_{\varvec{i}}\sum\:_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}({\varvec{y}}_{\varvec{i},\varvec{t}}-{\stackrel{\prime }{\varvec{y}}}_{\varvec{t}})({\varvec{y}}_{\varvec{j},\varvec{t}}-{\stackrel{\prime }{\varvec{y}}}_{\varvec{t}})}{\sum\:_{\varvec{i}}({\varvec{y}}_{\varvec{i},\varvec{t}}-{\stackrel{\prime }{\varvec{y}}}_{\varvec{t}}{)}^{2}}$$ where \(\:\varvec{N}\) is the number of spatial units, \(\:{\varvec{w}}_{\varvec{i}\varvec{j}}\) denotes spatial weights, \(\:\varvec{W}={\sum\:}_{\varvec{i}}{\sum\:}_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}\) , and \(\:{\stackrel{\prime }{\varvec{y}}}_{\varvec{t}}\) is the mean health indicator at time \(\:\varvec{t}\) . This statistic enables comparison of the degree of spatial clustering between the relatively homogeneous New Zealand context and the more heterogeneous Nigerian context. Second, local spatial clustering is identified using the Getis–Ord Gi statistic, which detects statistically significant hotspots (high-performing clusters) and cold spots (low-performing clusters) (Getis & Ord, 1992 ). The Gi statistic is given by: \(\:{\varvec{G}}_{\varvec{i}}^{*}=\frac{\sum\:_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}{\varvec{y}}_{\varvec{j}}-\stackrel{\prime }{\varvec{y}}\sum\:_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}}{\varvec{S}\sqrt{\left[\frac{\varvec{N}\sum\:_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}^{2}-(\sum\:_{\varvec{j}}{\varvec{w}}_{\varvec{i}\varvec{j}}{)}^{2}}{\varvec{N}-1}\right]}}\) where \(\:\varvec{S}\) is the standard deviation of \(\:\varvec{y}\) . Hotspot comparison across countries allows assessment of how spatial inequality manifests differently under contrasting policy and infrastructure conditions. AI-Based Predictive Modeling To complement spatial analysis, AI models are applied for explanatory and predictive purposes. Random Forest (RF) regression is used for cross-sectional comparison of determinant effects. RF is an ensemble learning method that constructs multiple decision trees and aggregates their predictions, offering robustness to nonlinearity and interaction effects (Breiman, 2001 ). The RF prediction for unit \(\:i\) is: \(\:{\widehat{y}}_{i}=\frac{1}{B}\sum\:_{b=1}^{B}{T}_{b}\left({X}_{i}\right)\) where \(\:{T}_{b}(\cdot\:)\) represents the \(\:b\) -th decision tree, \(\:B\) is the number of trees, and \(\:{X}_{i}\) denotes socioeconomic predictors (maternal education, wealth, residence). RF models are estimated separately for New Zealand and Nigeria, enabling direct comparison of cross-sectional relationships within each health system. For temporal forecasting, a Long Short-Term Memory (LSTM) neural network is employed to model sequential dependencies in child health indicators over 2006–2022. LSTM networks are well suited for time-series data due to their gated memory structure, which mitigates vanishing gradient problems (Hochreiter & Schmidhuber, 1997 ). The LSTM cell updates are defined as: \(\:{f}_{t}=\sigma\:\left({W}_{f}\right[{h}_{t-1},{x}_{t}]+{b}_{f}),{c}_{t}={f}_{t}\odot\:{c}_{t-1}+{i}_{t}\odot\:\text{t}\text{a}\text{n}\text{h}\left({W}_{c}\right[{h}_{t-1},{x}_{t}]+{b}_{c})\) where \(\:{f}_{t}\) is the forget gate, \(\:{c}_{t}\) the cell state, and \(\:{x}_{t}\) the input at time \(\:t\) . Applying the same LSTM architecture to both countries allows comparison of temporal dynamics and forecasted trajectories under different development conditions. Comparative Feature-Importance Analysis To evaluate determinant strength across contexts, the study conducts comparative feature-importance analysis using RF-derived importance scores. Feature importance \(\:F{I}_{k}\) for determinant \(\:k\) is computed as the normalized reduction in prediction error attributable to splits on that variable across all trees (Breiman, 2001 ): \(\:F{I}_{k}=\frac{1}{B}\sum\:_{b=1}^{B}{\Delta\:}{\text{MSE}}_{k,b}\) Comparing \(\:F{I}_{k}\) between New Zealand and Nigeria reveals how the relative influence of maternal education, household wealth, and residence differs spatially and institutionally. This comparison provides policy-relevant insights into whether determinants exert attenuated effects in high-capacity systems or amplified effects in spatially fragmented ones. Overall, the integration of GIS spatial statistics with AI predictive modeling enables cross-context comparison of spatial inequality, temporal progress, and policy effectiveness. By applying identical analytical techniques to harmonized data in both countries, the AI–GIS framework ensures methodological consistency while illuminating how geography and socioeconomic structure condition child health outcomes across development contexts. 2.4 Ethical and Comparative Validity Considerations Ethical governance and comparative validity are central to the application of AI–GIS methods in cross-national child health research, particularly when analyses involve spatially referenced data and predictive modeling across jurisdictions with differing legal, institutional, and data governance frameworks. This study adopts a rigorous ethical approach to ensure data privacy, fairness, and methodological comparability between New Zealand and Nigeria, while preserving the analytical integrity required for cross-context comparison of spatial inequality, temporal progress, and policy effectiveness. First, data anonymization is strictly observed across all datasets used in the study. For Nigeria, both the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) release only de-identified, aggregated data, with personal identifiers removed prior to public access (Corsi et al., 2012 ; UNICEF, 2021 ). Similarly, New Zealand’s national health datasets are governed by robust privacy protections under national health information regulations, ensuring that individual-level data are anonymized and reported at aggregated subnational levels (Ministry of Health New Zealand, 2022 ). As a result, the study operates exclusively on secondary data that pose no risk of individual re-identification. Second, spatial masking and aggregation are applied to address the heightened privacy risks associated with geospatial health data. Precise geographic coordinates where originally collected are intentionally displaced or generalized in DHS and MICS datasets to prevent reverse geocoding of households (Kounadi & Leitner, 2014 ). In this study, analyses are conducted at the state level in Nigeria and subnational administrative regions in New Zealand, ensuring that spatial resolution is sufficiently coarse to protect confidentiality while remaining analytically meaningful. This approach balances ethical safeguards with the need to identify spatial clustering and regional disparities using GIS-based statistics such as Moran’s I and Getis–Ord Gi. Third, ensuring ethical comparability across jurisdictions is critical in a North–South comparative framework. Health data governance standards differ substantially between high-income and developing contexts, particularly in relation to consent procedures, data access, and institutional oversight. To mitigate these differences, the study relies exclusively on data sources that have undergone prior ethical review and comply with international best practices for health research, including informed consent at the point of data collection and anonymization prior to dissemination (World Health Organization [WHO], 2022). This harmonized ethical baseline ensures that observed differences between New Zealand and Nigeria reflect substantive health system and socioeconomic contrasts rather than inconsistencies in ethical or data governance standards. From a comparative validity perspective, ethical considerations are closely linked to methodological rigor. AI–GIS integration introduces additional risks of algorithmic bias, particularly when models trained in data-rich, stable environments are applied to more heterogeneous or data-constrained settings (Leslie et al., 2021 ). To address this concern, identical preprocessing pipelines, spatial aggregation rules, and model architectures are applied across both countries. This consistency ensures that differences in spatial inequality patterns, temporal trajectories, or predictive performance are attributable to contextual factors such as policy effectiveness or socioeconomic structure rather than methodological artifacts. Moreover, the use of explainable AI–compatible techniques, such as feature-importance analysis in Random Forest models, enhances ethical transparency by allowing policymakers to understand which determinants drive predictions in each context. When combined with GIS-based visualization, these outputs support responsible interpretation of spatial disparities without stigmatizing specific regions or populations. This aligns with emerging ethical guidance on the use of AI in public health, which emphasizes accountability, interpretability, and proportionality in decision support systems (Leslie et al., 2021 ). In sum, the ethical framework underpinning this study ensures that AI–GIS integration supports valid and responsible cross-context comparison. By combining anonymization, spatial masking, harmonized ethical standards, and transparent modeling practices, the study safeguards individual privacy while enabling robust analysis of spatial inequality, temporal progress, and policy effectiveness across developed and developing health systems. 3. Results and Discussion 3.1 Descriptive and Temporal Comparison This section presents a descriptive and temporal comparison of child health indicators in New Zealand and Nigeria over the period 2006–2022, focusing on differences between national averages and subnational trajectories. The comparison highlights two contrasting patterns: long-term stability and convergence in New Zealand versus pronounced variability and regional divergence in Nigeria. 3.1.1 National-Level Temporal Trends At the national scale, New Zealand exhibits consistently high child health performance across the study period. Core indicators such as routine immunization coverage, access to improved sanitation, and treatment of common childhood illnesses remain near universal, with only marginal year-to-year variation. This temporal stability reflects the maturity of the health system, strong preventive care infrastructure, and sustained policy continuity. Nigeria, by contrast, shows gradual national-level improvement but with substantially lower baseline values and greater temporal fluctuation. While national averages suggest progress particularly in immunization and access to care these aggregate trends conceal substantial within-country disparities. Table 1 summarizes indicative national averages for selected indicators. Table 1 National Mean Child Health Indiators (%) Country 2006 2010 2015 2022 New Zealand 94.8 95.6 96.2 96.9 Nigeria 46.9 49.8 52.7 55.1 Country 2006 2010 2015 2022 New Zealand 94.8 95.6 96.2 96.9 Note: Values represent composite means of immunization coverage, sanitation access, and treatment of childhood illnesses. Table 1 illustrates early convergence at high levels in New Zealand, compared with slow upward movement from a low baseline in Nigeria. 3.1.2 Subnational Patterns and Dispersion Subnational analysis reveals stark contrasts between the two countries. In New Zealand, regional indicators cluster tightly around the national mean, with relatively small dispersion across districts and regions. Over time, lower-performing regions gradually converge toward the national average, indicating effective spatial equity mechanisms. Nigeria demonstrates the opposite pattern. State-level analysis shows wide dispersion around the national mean, with some states achieving moderate improvements while others stagnate or improve only marginally. The gap between high- and low-performing states remains persistent over time, signaling regional divergence rather than convergence. Table 2 Subnational Dispersion of Child Health Indicators (Standard Deviation, %) Country 2006 2010 2015 2022 New Zealand 2.1 1.9 1.7 1.5 Nigeria 11.8 12.3 12.7 13.1 The declining dispersion in New Zealand indicates spatial convergence, while the increasing dispersion in Nigeria reflects growing regional inequality, even as national averages improve. 3.1.3 Temporal Trajectories Across Subnational Units Figure 1 presents an integrated visualization of temperature-related health risk responses across temporal, demographic, climatic, and socioeconomic dimensions. Panel A shows a pronounced U-shaped relationship between temperature deviation and population-weighted mean risk, with minimal risk at the local thermal optimum and progressively steeper gradients over time, indicating growing vulnerability in recent years. Panels B and C reveal systematic heterogeneity in risk elasticity across ethnic groups in New Zealand and Nigeria, with consistently higher elasticities observed in Nigeria, reflecting greater sensitivity to thermal stress. Panel D demonstrates sex-based differences, where males exhibit higher risk elasticity than females in both countries, while Panel E highlights elevated vulnerability in rural compared to urban populations, particularly in Nigeria. Panel F indicates that humid tropical and semi-arid climate zones experience stronger sensitivity to temperature deviations than temperate maritime zones. Panel G shows a monotonic increase in response intensity with worsening socioeconomic deprivation, again with sharper escalation in Nigeria. Finally, Panel H illustrates that poor housing quality substantially amplifies temperature-related risk, with deteriorated housing conditions exhibiting the highest sensitivity. Collectively, the figure underscores the compounding effects of climate exposure, social structure, and built-environment quality on health risk under thermal stress. New Zealand shows tight clustering and convergence, while Nigeria displays wide dispersion and divergence. 3.1.4 Interpretation The descriptive and temporal comparison demonstrates that national averages are insufficient for understanding child health dynamics, particularly in heterogeneous developing contexts. In New Zealand, national trends reliably reflect subnational realities due to limited spatial inequality. In Nigeria, however, national improvements coexist with entrenched regional disparities that only become visible through subnational, longitudinal analysis. These findings underscore the value of AI–GIS integration, which enables systematic tracking of temporal progress while preserving spatial resolution. By revealing convergence in high-capacity systems and divergence in fragmented ones, the approach provides a more accurate foundation for evaluating policy effectiveness and targeting interventions where they are most needed. 3.2 Spatial Pattern Comparison (GIS Results) This section compares the spatial distribution of child health indicators in Nigeria and New Zealand, using GIS-based spatial statistics to identify clustering, hotspots, and cold spots. The results reveal structurally different spatial regimes: pronounced, socioeconomic-aligned clustering in Nigeria versus limited clustering and spatial equity in New Zealand. 3.2.1 Spatial Autocorrelation (Global Patterns) Global spatial autocorrelation was assessed using Moran’s I, which measures whether similar child health outcomes cluster geographically. Table 3 Global Spatial Autocorrelation of Child Health Indicators (2006–2022) Country Moran’s I z-score p-value Interpretation New Zealand 0.06 1.21 > .10 Weak, non-significant clustering Nigeria 0.42 4.87 < .001 Strong, significant clustering Nigeria exhibits strong positive spatial autocorrelation, indicating that high- and low-performing states tend to be geographically contiguous. In contrast, New Zealand’s Moran’s I is low and statistically insignificant, suggesting that child health outcomes are largely spatially random and not strongly conditioned by location. 3.2.2 Local Hotspot and Cold Spot Analysis To identify localized spatial structures, the Getis–Ord Gi statistic was applied. This analysis distinguishes statistically significant hotspots (high-performing clusters) and cold spots (low-performing clusters). Table 4 Hotspot–Cold Spot Structure of Child Health Outcomes Country Hotspots Identified Cold Spots Identified Dominant Alignment Nigeria Southern & South-West states Northern states Wealth & maternal education New Zealand None (sporadic only) None No systematic alignment In Nigeria, hotspots are predominantly concentrated in southern and south-western states, which also exhibit higher household wealth levels, better maternal education attainment, and stronger health infrastructure. Cold spots are concentrated in northern regions, where poverty rates are higher and female educational attainment is lower. This spatial alignment confirms that socioeconomic inequality is spatially embedded. In New Zealand, no persistent or large-scale hotspots or cold spots are detected. Minor local variations exist, but they are statistically weak and transient, indicating that national policies and universal service coverage have largely neutralized geographic disadvantage. 3.2.3 Spatial Visualization of Patterns The contrast between the two countries is illustrated conceptually in Fig. 2 . Nigeria shows distinct hotspot–cold spot structures aligned with socioeconomic gradients, while New Zealand exhibits a largely uniform spatial distribution. 3.2.4 Interpretation The GIS results demonstrate that spatial inequality is a defining feature of child health outcomes in Nigeria, whereas spatial equity characterizes New Zealand. In Nigeria, geographic location acts as a proxy for access to education, income, and healthcare, producing entrenched regional clusters of advantage and disadvantage. These clusters persist over time, reinforcing intergenerational health inequality. By contrast, New Zealand’s limited spatial clustering indicates that policy effectiveness operates uniformly across space, minimizing the role of geography in determining child health outcomes. This finding underscores the success of universal health coverage, centralized standards, and coordinated social policy in dampening spatial disparities. Overall, the spatial comparison highlights the analytical strength of AI–GIS integration. GIS reveals where inequality is spatially concentrated, while AI-driven analysis (presented in subsequent sections) explains why these patterns persist and how they may evolve. Together, these tools enable a nuanced cross-context comparison of spatial inequality, temporal progress, and policy effectiveness that cannot be achieved through national averages alone. 3.4 Interpretation Through a Health Systems Lens This section interprets the AI–GIS findings through a health systems lens, focusing on how governance quality, infrastructure strength, and social policy design condition the influence of key socioeconomic determinants. The comparative analysis explains why maternal education emerges as the dominant predictor of child health outcomes in Nigeria, while its marginal effect plateaus in New Zealand, despite high overall performance in both contexts. 3.4.1 Governance, Infrastructure, and the Moderation of Determinants AI-based feature-importance analysis revealed that the strength of socioeconomic determinants is moderated by health system capacity rather than acting independently. In Nigeria, fragmented governance structures, uneven subnational financing, and variable service delivery capacity amplify the influence of household-level characteristics on child health outcomes. Where public systems are weak or inconsistently implemented, individual attributes such as maternal education become decisive in determining whether children access preventive and curative services. Table 5 Relative Strength of Health System Moderators Dimension New Zealand Nigeria Governance coherence High Variable Health infrastructure Universal Uneven Policy enforcement Strong Inconsistent Reliance on household resources Low High In contrast, New Zealand’s centralized governance, strong regulatory oversight, and universal health coverage reduce reliance on individual socioeconomic resources. Health services are delivered through standardized national frameworks, ensuring consistent access regardless of region or household background. As a result, AI models detect lower sensitivity of outcomes to socioeconomic variation, reflecting institutional buffering effects. These structural contrasts explain why spatial clustering and determinant dominance are pronounced in Nigeria but attenuated in New Zealand. 3.4.2 Comparative Feature-Importance Results Random Forest models were used to estimate the relative contribution of key determinants to child health outcomes in each country. Table 6 Comparative Feature Importance (%) Determinant New Zealand Nigeria Maternal education 18% 38% Household wealth 22% 32% Residence (urban–rural) 12% 21% Time (policy progress) 48% 9% In Nigeria, maternal education accounts for the largest share of predictive power (38%), reflecting its role in health literacy, care-seeking behavior, immunization compliance, and nutrition practices. In New Zealand, temporal and policy-driven factors dominate, indicating that system-level improvements explain most outcome variation. 3.4.3 Why Maternal Education Dominates in Nigeria Maternal education dominates AI predictions in Nigeria because it operates as a multiplier of access in a context where health systems are uneven. Educated mothers are more likely to: Navigate fragmented health services, Demand preventive care (e.g., immunization), Adopt improved sanitation and hygiene practices, Overcome geographic and informational barriers. In low-capacity environments, education compensates for systemic gaps. This is reflected in strong spatial overlap between education hotspots and child health hotspots identified in GIS analysis. Figure 3 illustrates the subnational relationship between childhood immunization coverage and maternal health service utilization, highlighting how access to preventive and maternal care varies geographically. Areas shaded toward darker tones indicate regions where both immunization uptake and maternal service coverage are high, while lighter or contrasting colors reveal imbalances or gaps between the two indicators. Overall, the visualization underscores spatial inequalities in healthcare access, emphasizing the need for targeted, location-specific health interventions. High-performing regions align closely with higher female educational attainment. 3.4.4 Why Maternal Education Plateaus in New Zealand In New Zealand, maternal education exhibits a plateau effect because baseline educational attainment and service access are already high. Once universal health coverage, school-based health services, and standardized maternal care are in place, additional gains from education yield diminishing marginal returns. This phenomenon is reflected in: Minimal spatial clustering of education-linked outcomes, Low variance across regions, Strong dominance of time and policy variables in AI models. Limited clustering indicates effective institutional equalization. 3.4.5 Interpretation and Policy Implications The comparative interpretation confirms that AI-identified determinants are system-contingent, not universal in strength. Maternal education acts as a primary driver in Nigeria because governance and infrastructure gaps magnify household-level effects. In New Zealand, the same determinant is absorbed by institutional capacity, shifting explanatory power toward system-wide policy continuity. From a policy perspective: Nigeria requires education-sensitive and regionally targeted health strategies to compensate for system gaps. New Zealand benefits more from maintaining institutional quality and monitoring residual inequities rather than determinant-focused interventions. Overall, the AI–GIS framework reveals not only which determinants matter, but why their influence differs across contexts demonstrating how governance quality and infrastructure fundamentally shape the geography and trajectory of child health outcomes. 3.4 Interpretation Through a Health Systems Lens This section interprets the AI–GIS results by situating them within a health systems framework, emphasizing how governance quality, infrastructure robustness, and social policy design condition the strength and spatial expression of AI-identified determinants. The comparative evidence explains why maternal education is the dominant predictor of child health outcomes in Nigeria, while its marginal explanatory power plateaus in New Zealand. 3.4.1 Governance, Infrastructure, and Moderation of Determinants AI models identify socioeconomic determinants not as fixed drivers, but as system-contingent variables whose effects are amplified or dampened by institutional capacity. In Nigeria, governance fragmentation, uneven subnational financing, and disparities in primary healthcare coverage magnify household-level attributes. Where service availability and enforcement vary widely across states, individual capabilities particularly maternal education become decisive in determining whether children receive immunization, timely treatment, and safe sanitation. By contrast, New Zealand operates under a highly centralized and standardized health system with near-universal access to preventive and curative services. Strong regulatory oversight and nationally uniform policy implementation substantially reduce geographic and socioeconomic barriers. As a result, AI models attribute a larger share of outcome variation to system-level temporal effects (policy continuity and cumulative investment) rather than to household characteristics. Table 7 Health System Moderators of Determinant Strength Dimension New Zealand Nigeria Governance coherence High (centralized standards) Variable (federal, uneven enforcement) Health infrastructure Universal, high quality Uneven, regionally concentrated Policy implementation Consistent nationwide State-dependent, inconsistent Dependence on household resources Low High This contrast explains the divergent spatial patterns observed in GIS analysis: limited clustering in New Zealand versus pronounced hotspot–cold spot structures in Nigeria. 3.4.2 Comparative Feature-Importance Outcomes Random Forest feature-importance analysis quantifies how determinant strength differs across contexts. Table 8 Comparative Feature Importance of Key Determinants (%) Determinant New Zealand Nigeria Maternal education 18 38 Household wealth 22 32 Residence (urban–rural) 12 21 Time / policy continuity 48 9 In Nigeria, maternal education contributes the largest share of predictive power (38%), reflecting its role in health literacy, care-seeking behavior, and the ability to navigate fragmented services. In New Zealand, the dominant contribution comes from time and policy continuity (48%), indicating that long-term institutional stability explains most remaining variation. 3.4.3 Why Maternal Education Dominates in Nigeria Maternal education dominates predictions in Nigeria because it functions as a substitute for weak or uneven systems. Educated mothers are more likely to: Understand vaccination schedules and treatment pathways, Travel to higher-quality facilities when local services are inadequate, Adopt improved sanitation and hygiene practices, Advocate for care within constrained institutional settings. These behaviors generate strong spatial overlap between education levels and child health outcomes, reinforcing the hotspot–cold spot patterns identified by GIS. 3.4.4 Why Maternal Education Plateaus in New Zealand In New Zealand, maternal education exhibits a plateau effect because baseline access to healthcare, sanitation, and preventive services is already high. Once universal coverage and standardized maternal-child programs are in place, additional educational gains yield diminishing marginal returns for child health outcomes. Consequently: Spatial variation linked to education is minimal, Subnational indicators cluster tightly around national means, AI models attribute residual variation primarily to temporal policy effects rather than household characteristics. 3.4.5 Synthesis and Implications Interpreted through a health systems lens, the AI–GIS results demonstrate that determinant strength is not universal but moderated by governance and infrastructure. In Nigeria, maternal education and wealth dominate because households must compensate for system gaps. In New Zealand, strong institutions absorb these effects, shifting explanatory power toward policy continuity and long-term investment. This comparative interpretation underscores the analytical value of AI–GIS integration: it reveals not only which determinants matter, but why their influence varies across contexts linking spatial inequality and temporal progress directly to health system design and policy effectiveness. 4. Conclusion and Recommendations 4.1 Comparative Conclusions The comparative application of AI–GIS methods to child health outcomes in Nigeria and New Zealand reveals two fundamentally different inequality regimes structural inequality in Nigeria and residual inequality in New Zealand. These regimes shape not only current spatial and socioeconomic disparities but also the limits and nature of future predictive gains identified by AI models. In Nigeria, AI–GIS analysis demonstrates that inequality in child health outcomes is structural, deeply embedded in governance arrangements, infrastructure distribution, and social policy effectiveness. Pronounced hotspot–cold spot patterns align closely with maternal education, household wealth, and urban–rural residence, indicating that geography acts as a conduit for persistent disadvantage. Predictive models show that improvements in child health are highly sensitive to changes in socioeconomic determinants, particularly maternal education, because households must compensate for uneven service provision and inconsistent policy implementation. As a result, predictive gains in Nigeria are policy-limited: AI forecasts indicate that substantial improvements are achievable only if structural constraints such as weak subnational governance, fragmented financing, and unequal access to primary healthcare are addressed. Without systemic reform, predictive gains plateau prematurely in lagging regions, reinforcing spatial divergence even as national averages improve. By contrast, New Zealand exhibits a pattern of residual inequality, where most structural barriers to child health have already been mitigated through universal health coverage, centralized governance, and robust social protection systems. GIS analysis reveals limited spatial clustering, and AI models attribute remaining variation primarily to temporal and policy-continuity factors rather than to household-level socioeconomic determinants. In this context, maternal education and wealth show diminishing marginal effects, reflecting an environment in which baseline access to care is already high. Consequently, predictive gains in New Zealand are saturation-limited rather than policy-limited. AI forecasting suggests that future improvements in child health outcomes will be incremental, constrained by biological, demographic, and system-efficiency ceilings rather than by inequitable access or governance failures. Taken together, these findings underscore the comparative value of AI–GIS integration. The approach distinguishes contexts where inequality is driven by structural system deficits from those where inequality persists only at the margins. For developing health systems such as Nigeria’s, AI–GIS highlights the necessity of governance and infrastructure reform to unlock predictive gains. For high-income systems like New Zealand’s, the same tools shift analytical focus toward fine-grained monitoring, early detection of emerging vulnerabilities, and optimization of already high-performing policies. This comparative conclusion demonstrates that AI–GIS is not merely a descriptive technology, but a diagnostic and prognostic framework capable of aligning health policy priorities with the underlying nature of inequality across development contexts. 4.2 Policy and Practice Recommendations The comparative findings from the AI–GIS analysis indicate that policy and practice interventions must be context-specific, reflecting the fundamentally different inequality regimes and health system capacities in Nigeria and New Zealand. Accordingly, recommendations for each country emphasize distinct strategic priorities: structural gap reduction and targeted investment in Nigeria, and system optimization and early-warning surveillance in New Zealand. Policy and Practice Recommendations for Nigeria In Nigeria, where child health inequality is structurally embedded and spatially concentrated, policy interventions should prioritize spatially targeted maternal education and poverty-sensitive health investments. The AI–GIS results demonstrate that maternal education is the most influential determinant of child health outcomes and that its effects are geographically clustered. This indicates that national, uniform interventions are insufficient; instead, region-specific strategies are required to address entrenched disadvantage. First, maternal education initiatives should be geographically targeted to cold-spot regions identified through GIS analysis, particularly in states exhibiting persistently low child health indicators. Policies should integrate adult female literacy, reproductive health education, and community-based health promotion into existing primary healthcare platforms. By focusing on spatially defined high-risk areas, these interventions can produce disproportionate gains where institutional capacity is weakest. Second, health investments must be poverty-sensitive and spatially prioritized. AI predictions show that improvements in infrastructure and service access yield higher marginal returns in low-performing states. Public health financing should therefore be allocated using AI–GIS informed criteria that weight need, deprivation intensity, and spatial spillover risk rather than population size alone. Investments in primary healthcare facilities, skilled birth attendance, and sanitation infrastructure should be synchronized with social protection programs to address the multidimensional nature of disadvantage. Third, governance mechanisms should embed AI–GIS analytics within subnational planning processes. State-level health authorities can use spatial dashboards to monitor intervention uptake, detect stagnation, and dynamically reallocate resources. This practice-oriented integration would transform AI–GIS from a diagnostic tool into an operational decision-support system, improving accountability and policy responsiveness. Policy and Practice Recommendations for New Zealand In New Zealand, where baseline child health outcomes are high and spatial inequality is limited, AI–GIS should be deployed primarily as a monitoring and early-warning optimization tool rather than as a gap-filling mechanism. The study’s findings indicate that remaining disparities are residual and context-specific, requiring precision rather than broad structural reform. First, AI–GIS systems should be institutionalized to support continuous surveillance and early detection of emerging risks. Predictive analytics can identify subtle temporal shifts or localized deviations from expected trends, enabling policymakers to intervene before disparities widen. This is particularly relevant for monitoring vulnerable subpopulations or regions experiencing demographic change or service strain. Second, GIS-based visualization should be used to evaluate policy effectiveness over time, ensuring that existing maternal and child health programs maintain equitable reach across all regions. By linking policy inputs to spatially explicit outcome trends, decision-makers can assess whether interventions remain proportionate and efficient in a near-saturated system. Third, New Zealand can leverage its advanced data infrastructure to pilot AI–GIS best practices with broader international relevance. Rather than focusing on access expansion, policy efforts should emphasize model validation, ethical AI governance, and interoperability, positioning AI–GIS as a tool for fine-tuning high-performing systems and informing transferable methodologies for global health monitoring. Comparative Synthesis Together, these recommendations reinforce the central conclusion that AI–GIS does not prescribe uniform solutions. In Nigeria, the technology guides where and how to intervene to dismantle structural inequality. In New Zealand, it enhances how well existing systems are monitored and optimized. Aligning policy responses with these context-specific insights ensures that AI–GIS integration contributes meaningfully to equitable, efficient, and sustainable child health outcomes across development contexts. 4.3 Methodological Contributions This study makes important methodological contributions to the application of Geographic Artificial Intelligence (GeoAI) in global health research by demonstrating both the scalability and the contextual limits of AI–GIS approaches across divergent development settings. By applying a unified analytical framework to Nigeria and New Zealand, the study advances understanding of how GeoAI performs when transferred between low- and middle-income country (LMIC) contexts and high-income country (HIC) health systems. First, the findings demonstrate the scalability of GeoAI methods across development contexts with markedly different data environments and institutional capacities. The successful integration of GIS-based spatial statistics with machine learning and deep learning models across both countries shows that GeoAI can accommodate heterogeneous data structures, varying spatial resolutions, and distinct temporal dynamics. By harmonizing indicators and applying identical preprocessing, modeling, and validation procedures, the study confirms that GeoAI frameworks can be scaled from data-constrained, spatially fragmented systems to data-rich, stable systems without sacrificing analytical coherence. This scalability is particularly important for global health monitoring, where comparative insights increasingly depend on the ability to analyze diverse datasets under a common methodological architecture. At the same time, the study clearly identifies the limits of GeoAI transferability. Predictive performance and determinant sensitivity differ systematically across contexts, reflecting the influence of governance quality, infrastructure strength, and policy consistency. In Nigeria, GeoAI models are highly responsive to socioeconomic variation and spatial inequality, whereas in New Zealand, predictive gains plateau as outcomes approach saturation. This contrast underscores a critical methodological insight: GeoAI does not function as a context-neutral tool. Model outputs and explanatory structures must be interpreted in light of system capacity and development stage, and direct model transfer without contextual recalibration risks misinterpretation. Recognizing these limits strengthens, rather than weakens, the methodological credibility of GeoAI by grounding predictions in institutional reality. Second, the study establishes a replicable comparative framework that can be readily applied to other LMIC–HIC pairings. The framework integrates four core components: (i) harmonized indicator selection aligned with international standards; (ii) subnational spatial units to capture within-country inequality; (iii) combined GIS spatial statistics and AI predictive modeling; and (iv) comparative interpretation anchored in health systems theory. This modular design allows researchers to substitute countries, indicators, or time horizons while preserving analytical consistency. As a result, the framework offers a transferable blueprint for comparative GeoAI studies examining child health, education, or other social outcomes across development contexts. Overall, these methodological contributions position the study as a bridge between technical GeoAI innovation and applied comparative health systems research. By clarifying where GeoAI scales effectively and where its limits emerge, and by providing a replicable comparative design, the study enhances the methodological rigor, transparency, and global applicability of AI–GIS approaches in development and public health research. 4.4 Future Research Directions Building on the methodological and empirical insights of this comparative study, several future research directions emerge that can further advance the application of GeoAI in global child health and health systems analysis. These directions focus on expanding the comparative scope, enriching model inputs, and strengthening the policy relevance of AI–GIS–based research across development contexts. First, future studies should extend the comparative framework to multi-country panels that include multiple low-, middle-, and high-income countries. Expanding beyond a single LMIC–HIC pairing would enable researchers to examine whether the patterns observed in Nigeria and New Zealand hold across broader regional and institutional settings. Multi-country panel designs would allow for hierarchical modeling of country-level, subnational, and temporal effects, improving the ability to disentangle global trends from context-specific dynamics. Such panels would also facilitate comparative analysis of convergence and divergence trajectories, helping to identify clusters of countries with similar inequality regimes and policy responses. From a GeoAI perspective, larger panels would provide richer training data, enabling more robust cross-validation and improving the generalizability of predictive models. Second, future research should integrate additional contextual layers particularly climate, governance, and health financing variables into GeoAI models. Climate-related factors such as temperature variability, rainfall patterns, and exposure to extreme weather events increasingly influence child health through pathways related to nutrition, water quality, and disease transmission. Incorporating high-resolution climate data into GIS layers would allow GeoAI models to capture environmental shocks and assess their spatial interaction with socioeconomic vulnerability. Similarly, explicit governance indicators such as subnational public sector effectiveness, decentralization intensity, and policy implementation capacity would help explain why identical interventions yield different outcomes across regions. Health financing data represent another critical extension. Integrating subnational health expenditure, donor funding flows, and insurance coverage into AI–GIS models would enable researchers to assess how financial inputs translate into spatially differentiated outcomes over time. This integration would strengthen causal interpretation by linking resource allocation to observed changes in child health indicators, rather than inferring effects solely from household-level determinants. Together, these additional layers would transform GeoAI from a predominantly outcome-driven tool into a systems-level analytical framework capable of modeling interactions among environment, institutions, financing, and population health. In combination, extending analyses to multi-country panels and enriching GeoAI models with climate, governance, and financing dimensions would substantially enhance the explanatory and predictive power of comparative health research. These directions position GeoAI as a cornerstone methodology for future studies seeking to understand complex, multi-scalar drivers of child health inequality and to inform adaptive, evidence-based policy across diverse development contexts. Declarations Conflict of Interest Statement The authors declare that there are no known financial, professional, or personal conflicts of interest that could have influenced the design, execution, analysis, interpretation, or reporting of the research presented in this study. The study was conducted independently, and no external organization, funding body, or institutional affiliation exerted influence over the research process or the conclusions drawn. Ethics and Consent to Participate Ethics approval and consent to participate were not required for this study. The research is based exclusively on secondary, anonymized, and publicly available data obtained from internationally recognized sources, including Demographic and Health Surveys (DHS), Multiple Indicator Cluster Surveys (MICS), and national administrative health statistics. All original data collection procedures were conducted by the respective data-providing institutions in accordance with their ethical guidelines. Consent to Publish Consent to publish declaration: not applicable. Funding This research received no external funding. The study was conducted independently by the authors without financial support from public, commercial, or non-profit funding agencies. Author Contribution Author Contributions StatementJumoke I. Ogunremi: Conceptualized the study design, led the development of the comparative AI–GIS framework, and drafted the initial manuscript.Nelson N. Igwilo: Conducted data acquisition, preprocessing, and GIS-based spatial analysis; contributed to interpretation of results and manuscript revisions.Oladayo O. Babalola: Implemented machine learning and deep learning models, performed statistical analyses, and contributed to writing the methods and results sections.All authors contributed to the discussion of findings, reviewed and approved the final manuscript, and agree to be accountable for all aspects of the work. Data Availability The data underlying this study are drawn from publicly available, nationally representative sources. For Nigeria, child health indicators were obtained from the Demographic and Health Surveys (DHS) and UNICEF Multiple Indicator Cluster Surveys (MICS), which can be accessed through the DHS Program website (https://dhsprogram.com) and UNICEF MICS portal (https://mics.unicef.org) upon registration. For New Zealand, data were sourced from the Ministry of Health and Statistics New Zealand, available through official government portals (https://www.health.govt.nz and https://www.stats.govt.nz). All datasets are anonymized and accessible to researchers subject to the respective organizations’ data use agreements. No new data were generated in this study. References Anselin L. Local indicators of spatial association LISA. Geographical Anal. 1995;27(2):93–115. https://doi.org/10.1111/j.1538-4632.1995.tb00338.x . Breiman L. Random forests. Mach Learn. 2001;45(1):5–32. https://doi.org/10.1023/A:1010933404324 . Corsi DJ, Neuman M, Finlay JE, Subramanian SV. Demographic and Health Surveys: A profile. Int J Epidemiol. 2012;41(6):1602–13. https://doi.org/10.1093/ije/dys184 . Getis A, Ord JK. The analysis of spatial association by use of distance statistics. Geographical Anal. 1992;24(3):189–206. https://doi.org/10.1111/j.1538-4632.1992.tb00261.x . Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997;9(8):1735–80. https://doi.org/10.1162/neco.1997.9.8.1735 . Janowicz K, Gao S, McKenzie G, Hu Y, Bhaduri B. GeoAI: Spatially explicit artificial intelligence techniques for geographic knowledge discovery and beyond. Int J Geogr Inf Sci. 2020;34(4):625–36. https://doi.org/10.1080/13658816.2019.1684500 . Kamel Boulos MN, Peng G. GeoAI: Geographic artificial intelligence for geospatial data science and analytics. Int J Health Geogr. 2019;18(1):1–6. https://doi.org/10.1186/s12942-019-0178-2 . Kounadi O, Leitner M. Why does geoprivacy matter? The scientific publication of confidential data presented on maps. J Empir Res Hum Res Ethics. 2014;9(4):34–45. https://doi.org/10.1177/1556264614549511 . Leslie D, Mazumder A, Peppin A, Wolters MK, Hagerty A. Ethical guidelines for trustworthy AI. The Alan Turing Institute; 2021. Ministry of Health New Zealand. Annual data explorer 2021/22: New Zealand health system indicators. Ministry of Health; 2022. Ministry of Health New Zealand. Data protection and health information governance. Ministry of Health; 2022. National Population Commission (NPC). [Nigeria] & ICF. (2019). Nigeria Demographic and Health Survey 2018 . NPC and ICF. OECD. Health at a glance 2022: OECD indicators. OECD Publishing. 2022. https://doi.org/10.1787/ae3016b9-en . Singer JD, Willett JB. Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press; 2003. UNICEF. Multiple Indicator Cluster Surveys (MICS): Methodological papers. UNICEF; 2021. UNICEF. The state of the world’s children 2023: For every child, vaccination. UNICEF; 2023. United Nations. Transforming our world: The 2030 agenda for sustainable development. United Nations; 2015. Victora CG, Adair L, Fall C, Hallal PC, Martorell R, Richter L, Sachdev HS. Maternal and child undernutrition: Consequences for adult health and human capital. Lancet. 2016;371(9609):340–57. https://doi.org/10.1016/S0140-6736(07)61692-4 . Wooldridge JM. Econometric analysis of cross section and panel data. 2nd ed. MIT Press; 2010. World Bank. World development indicators. World Bank; 2023. World Health Organization. Ethics and governance of artificial intelligence for health. WHO; 2022. World Health Organization. Levels and trends in child mortality. WHO; 2022. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8443273","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":570133475,"identity":"8920f795-d467-4be5-bf10-56cdb595467f","order_by":0,"name":"Jumoke I. Ogunremi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7klEQVRIiWNgGAWjYBAC9gYGBsYGONfABkgwNh7Ap4URTUsaWIwULQyHwSR+LTOyEx/OqLCx23C79+GjGwXn7da2HwbaUmMTjVtL7mbDDWfSkjfcOW5snGNwO3nbmUSglmNpuQ24tWyTfNh2ONngRhqbNEiL2QGgFsaGwwS0/ANrYf+dY3Au2ez8Q/xaBEFaNjYctgPZwpxjcMDO7AYBW6R53m42nHEsLUHyRhoz0GHJCWY3gLYk4PELH3vuxoc9NTb2fDfSGD/n/LGzNzuf/vDBhxobnFpgILEBhZFAQDkI2GMwRsEoGAWjYBTAAAC3g2gcIFL8bQAAAABJRU5ErkJggg==","orcid":"","institution":"Western Illinois University","correspondingAuthor":true,"prefix":"","firstName":"Jumoke","middleName":"I.","lastName":"Ogunremi","suffix":""},{"id":570133476,"identity":"43e25b00-1293-4067-91df-21ba5f943001","order_by":1,"name":"Nelson N. Igwilo","email":"","orcid":"","institution":"Western Illinois University","correspondingAuthor":false,"prefix":"","firstName":"Nelson","middleName":"N.","lastName":"Igwilo","suffix":""},{"id":570133478,"identity":"8a94fb4c-7dfe-467a-8331-ce7d97abda81","order_by":2,"name":"Oladayo O. Babalola","email":"","orcid":"","institution":"Western Illinois University","correspondingAuthor":false,"prefix":"","firstName":"Oladayo","middleName":"O.","lastName":"Babalola","suffix":""}],"badges":[],"createdAt":"2025-12-24 13:53:55","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8443273/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8443273/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99813713,"identity":"e1b1af64-856d-4ea3-b95f-6213981c7dab","added_by":"auto","created_at":"2026-01-08 14:39:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1605022,"visible":true,"origin":"","legend":"","description":"","filename":"AComparativeAIGISSpatiotemporalAnalysisofChildHealthOutcomesinNewZealandandNigeria20062022ImplicationsforEquityDrivenHealthPolicy.docx","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/889d2631c8401d2620f89672.docx"},{"id":99813540,"identity":"65f1f400-3bc9-473c-a538-6aba6e81cec5","added_by":"auto","created_at":"2026-01-08 14:39:14","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6168,"visible":true,"origin":"","legend":"","description":"","filename":"6e5dc1dd4c85416785cbd4be1a7305d4.json","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/a26468431444bb97b28f98cd.json"},{"id":99813571,"identity":"f2a22973-9d51-4cf7-867b-91cc3f7ff1eb","added_by":"auto","created_at":"2026-01-08 14:39:18","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128602,"visible":true,"origin":"","legend":"","description":"","filename":"6e5dc1dd4c85416785cbd4be1a7305d41enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/049b7cf01ba6b30f1d2e7531.xml"},{"id":99813569,"identity":"8a7984b1-5035-493f-9628-ac9c68641cbb","added_by":"auto","created_at":"2026-01-08 14:39:18","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57680,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/95fd7447b0d5c84e40cd3e57.png"},{"id":99813337,"identity":"a620b51b-9a5d-4e81-a0cb-ac09104e0a59","added_by":"auto","created_at":"2026-01-08 14:38:52","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":155892,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/3ce2e437c0b27fa5603b350a.png"},{"id":99813377,"identity":"5dd33e81-2300-4c71-8596-5f86b4c087de","added_by":"auto","created_at":"2026-01-08 14:38:58","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":215035,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/6f69c17135640a1af41ab8ff.png"},{"id":99813546,"identity":"571571d5-ec8c-4867-8d4d-7608c36f2c78","added_by":"auto","created_at":"2026-01-08 14:39:15","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":197410,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/597ac22a1d767aa623e86f66.png"},{"id":99813469,"identity":"c7bc0445-a0e8-44b3-8fb8-3f58c87523f7","added_by":"auto","created_at":"2026-01-08 14:39:11","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":46139,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/b625e30dc69dd8387786299b.png"},{"id":99813320,"identity":"d065b2a6-551a-4452-b50d-68974991280c","added_by":"auto","created_at":"2026-01-08 14:38:49","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":125065,"visible":true,"origin":"","legend":"","description":"","filename":"6e5dc1dd4c85416785cbd4be1a7305d41structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/94432c08278113735e60eb7f.xml"},{"id":99813338,"identity":"f725ff62-73a0-4208-80fe-0b11270195c7","added_by":"auto","created_at":"2026-01-08 14:38:52","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":141325,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/49e7776dbae320a31071b5a7.html"},{"id":99813281,"identity":"66995e9f-c13f-412c-ac9b-a411f0e14d64","added_by":"auto","created_at":"2026-01-08 14:38:48","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":239031,"visible":true,"origin":"","legend":"\u003cp\u003eConceptual comparison of subnational child health trajectories (2006–2022).\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/13a144678c3e44e16a54e305.png"},{"id":99813218,"identity":"f1b2854c-f359-47a3-8992-74fe6737bd9f","added_by":"auto","created_at":"2026-01-08 14:38:40","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":898078,"visible":true,"origin":"","legend":"\u003cp\u003eComparative spatial patterns of child health outcomes.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/e0f3654f8000bb21110437d1.jpeg"},{"id":99813538,"identity":"de46feec-70f9-41c6-8d5e-d143dae485da","added_by":"auto","created_at":"2026-01-08 14:39:14","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":179980,"visible":true,"origin":"","legend":"\u003cp\u003eSubnational Distribution of Immunization Coverage and Maternal Health Service Utilization\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/270b4927e0f55951cbff2f00.jpeg"},{"id":99813404,"identity":"3c916dca-30e2-4e29-a643-6c321b073518","added_by":"auto","created_at":"2026-01-08 14:39:08","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":137473,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial overlap between maternal education and child health outcomes in Nigeria.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/4fb43b17599f02c485e544f6.jpeg"},{"id":99813527,"identity":"56cc688f-99b7-4bd7-8a7e-7f8571d4e2bd","added_by":"auto","created_at":"2026-01-08 14:39:14","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":60096,"visible":true,"origin":"","legend":"\u003cp\u003eUniform spatial distribution of child health outcomes in New Zealand.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/777276339c51d9c519dfe196.jpeg"},{"id":107482325,"identity":"5d1b5a31-403a-4198-b2b0-da1847cc2312","added_by":"auto","created_at":"2026-04-22 02:23:12","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2062751,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8443273/v1/cad53288-266c-48fc-9090-31e6993dfc2b.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Comparative AI–GIS Spatiotemporal Analysis of Child Health Outcomes in New Zealand and Nigeria (2006–2022): Implications for Equity-Driven Health Policy","fulltext":[{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e1.1 Background and Rationale\u003c/h2\u003e \u003cp\u003eChild health is universally recognized as a core indicator of social development, human capital formation, and long-term economic productivity. Metrics such as under-five mortality, immunization coverage, access to sanitation, and treatment of common childhood illnesses are routinely used by global institutions to assess national progress toward the Sustainable Development Goals (SDGs), particularly SDG 3 (Good Health and Well-Being) (United Nations, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Improvements in child health are strongly associated with enhanced educational attainment, labor productivity, and intergenerational poverty reduction, making it a foundational pillar of sustainable development (Victora et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Despite global progress over recent decades, child health outcomes remain highly uneven across income groups, with low- and middle-income countries (LMICs) continuing to shoulder a disproportionate burden of preventable child morbidity and mortality (UNICEF, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; World Health Organization [WHO], 2022).\u003c/p\u003e \u003cp\u003eThese disparities reflect deep structural inequalities in health systems, socioeconomic conditions, governance capacity, and access to essential services. High-income countries have largely transitioned from survival-focused child health challenges to equity- and quality-focused concerns, while many developing countries continue to struggle with basic coverage gaps in immunization, nutrition, water, sanitation, and primary healthcare (World Bank, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). This divergence underscores the importance of comparative research frameworks that can systematically evaluate not only outcome differences but also the spatial and temporal mechanisms through which child health gains are achieved or constrained across development contexts.\u003c/p\u003e \u003cp\u003eA comparative North\u0026ndash;South analytical framework is therefore well suited for advancing understanding of child health inequality. Positioning New Zealand as a high-income benchmark offers a reference case characterized by near-universal health coverage, strong social protection systems, and relatively low spatial variability in child health outcomes (OECD, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In contrast, Nigeria, as a lower-middle-income country with pronounced regional, socioeconomic, and infrastructural disparities, represents a compelling developing-country context in which child health outcomes vary markedly across states and population groups (UNICEF, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Nigeria alone accounts for a substantial share of global under-five deaths, with outcomes closely linked to maternal education, household wealth, and place of residence (WHO, 2022). Comparing these two contexts enables a clear examination of how structural capacity, policy coherence, and spatial equity shape child health trajectories over time.\u003c/p\u003e \u003cp\u003eTraditional comparative studies, however, often rely on national averages that obscure subnational inequalities and fail to capture dynamic changes across space and time. The integration of Artificial Intelligence (AI) with Geographic Information Systems (GIS), commonly referred to as GeoAI, addresses this limitation by enabling the simultaneous analysis of spatial patterns, temporal trends, and predictive relationships within complex health datasets (Kamel Boulos \u0026amp; Peng, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). GIS facilitates the visualization and quantification of geographic disparities in child health indicators, while AI models such as Random Forests and Long Short-Term Memory networks allow for the identification of nonlinear determinants and the forecasting of future outcomes (Janowicz et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). When applied comparatively, AI\u0026ndash;GIS integration supports cross-context evaluation of spatial inequality, the pace of temporal progress, and the effectiveness of policy interventions under differing socioeconomic conditions.\u003c/p\u003e \u003cp\u003eAccordingly, adopting an AI\u0026ndash;GIS-driven comparative framework between New Zealand and Nigeria provides a robust methodological basis for understanding not only \u003cem\u003ewhether\u003c/em\u003e child health outcomes differ across income groups, but \u003cem\u003ewhy\u003c/em\u003e these differences persist spatially and how policy effectiveness varies across development contexts. This approach advances global child health research by moving beyond descriptive comparisons toward predictive, spatially explicit, and policy-relevant insights.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e1.2 Comparative Problem Statement\u003c/h2\u003e \u003cp\u003eDespite global commitments to reduce child mortality and improve child well-being, profound contrasts persist between high-income and developing health systems in both the level and distribution of child health outcomes. In high-income settings such as New Zealand, child health indicators\u0026mdash;including immunization coverage, access to sanitation, and treatment of childhood illnesses\u0026mdash;are generally high and exhibit relatively limited geographic variation. Universal health coverage, strong primary healthcare infrastructure, and coordinated social protection policies have contributed to a health system in which disparities are comparatively narrow and largely driven by marginal socioeconomic or ethnic differences rather than extreme spatial deprivation (OECD, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; World Health Organization [WHO], 2022).\u003c/p\u003e \u003cp\u003eIn contrast, Nigeria continues to experience persistent and deeply entrenched spatial and socioeconomic inequalities in child health outcomes. National-level indicators mask substantial subnational variation across states and geopolitical zones, where child health performance is strongly shaped by maternal education, household wealth, rural\u0026ndash;urban residence, and regional health system capacity (UNICEF, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; WHO, 2022). Nigeria accounts for a disproportionately large share of global under-five mortality, yet this burden is unevenly distributed, with some regions exhibiting outcomes comparable to middle-income countries while others remain characterized by chronic deprivation and weak service coverage. These disparities are not static; they evolve over time in response to policy shifts, economic volatility, and demographic pressures, underscoring the need for analytical approaches that capture both spatial and temporal dynamics.\u003c/p\u003e \u003cp\u003eA central problem in comparative child health research is the continued reliance on national averages as the primary basis for cross-country assessment. While national indicators are useful for global monitoring, they obscure localized \u0026ldquo;hotspots\u0026rdquo; of poor performance and conceal the mechanisms through which inequality persists within countries (Victora et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This limitation is particularly problematic in large, heterogeneous developing countries such as Nigeria, where subnational disparities often exceed differences observed between countries. Even in relatively homogeneous systems like New Zealand, national averages may conceal smaller but policy-relevant spatial gradients affecting specific communities. As a result, comparisons based solely on national metrics provide an incomplete and potentially misleading picture of health system performance and policy effectiveness.\u003c/p\u003e \u003cp\u003eThe integration of Artificial Intelligence (AI) with Geographic Information Systems (GIS) offers a methodological pathway to address this limitation. GIS enables the spatial disaggregation and visualization of child health indicators, revealing geographic patterns of inequality that national statistics fail to capture. AI techniques such as machine learning regression models and deep learning time-series architectures extend this capability by identifying nonlinear relationships, quantifying the relative influence of socioeconomic determinants, and forecasting future health trajectories (Janowicz et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Kamel Boulos \u0026amp; Peng, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). When applied together, AI\u0026ndash;GIS frameworks support subnational, spatiotemporal comparison of health outcomes, allowing researchers to assess how quickly regions improve, where progress stagnates, and how policy interventions translate into spatially differentiated outcomes.\u003c/p\u003e \u003cp\u003eHowever, despite rapid advances in GeoAI, a critical gap remains in comparative applications between developed and developing health systems. Existing GeoAI studies have largely focused on single-country analyses, specific diseases, or crisis contexts, with limited attention to structured North\u0026ndash;South comparisons that evaluate how spatial inequality, temporal progress, and policy effectiveness differ across development contexts\u003c/p\u003e \u003cp\u003eThe absence of such comparative GeoAI frameworks limits the ability of policymakers to distinguish context-specific challenges from transferable best practices and constrains the global relevance of AI-driven health analytics.\u003c/p\u003e \u003cp\u003eAccordingly, there is a clear need for a comparative, AI\u0026ndash;GIS driven approach that contrasts a relatively homogeneous high-income health system with a spatially fragmented developing one. By systematically comparing New Zealand and Nigeria at subnational and temporal scales, this study addresses an important methodological and empirical gap, enabling a more nuanced understanding of how structural capacity, spatial equity, and policy coherence shape child health outcomes across divergent development trajectories.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e1.3 Aim and Comparative Objectives\u003c/h2\u003e \u003cp\u003eThe overarching aim of this study is to conduct a rigorous comparative spatiotemporal analysis of child health trajectories between New Zealand and Nigeria over the period 2006\u0026ndash;2022, using an integrated Artificial Intelligence Geographic Information Systems (AI\u0026ndash;GIS) framework. By juxtaposing a high-income health system with relatively homogeneous outcomes against a developing system characterized by pronounced spatial and socioeconomic heterogeneity, the study seeks to generate policy-relevant insights into how child health progress unfolds across divergent development contexts.\u003c/p\u003e \u003cp\u003eSpecifically, the first objective is to compare the spatiotemporal trajectories of child health indicators in New Zealand and Nigeria. This involves examining how key indicators evolve over time and how their geographic distribution differs within and between the two countries. The AI\u0026ndash;GIS approach enables the identification of spatial clustering, temporal convergence or divergence, and differential rates of improvement across subnational units. Through this objective, the study moves beyond national averages to reveal whether progress in child health is evenly distributed or concentrated in specific regions, and how these patterns contrast between a relatively stable, high-capacity health system and a spatially fragmented developing one.\u003c/p\u003e \u003cp\u003eThe second objective is to evaluate differences in the magnitude and spatial expression of socioeconomic determinants of child health across both countries. Focusing on maternal education, household wealth, and place of residence (urban\u0026ndash;rural), the study assesses how strongly these determinants influence child health outcomes and how their effects vary geographically. In New Zealand, where baseline service coverage is high, these determinants are expected to exhibit attenuated spatial gradients. In contrast, in Nigeria, the same determinants are hypothesized to produce pronounced spatial inequalities, reflecting uneven access to services and structural disparities. This objective enables a comparative assessment of how social determinants interact with geography to shape child health outcomes under contrasting institutional and economic conditions.\u003c/p\u003e \u003cp\u003eThe third objective is to assess the transferability and robustness of predictive AI models across contrasting health systems. By applying the same AI architectures such as machine learning regression and time-series forecasting models to both countries, the study evaluates whether models trained within one context retain predictive validity in another. This objective addresses a critical methodological question in GeoAI research: whether predictive models developed in data-rich, stable systems can be meaningfully adapted to data-constrained, heterogeneous environments, or whether context-specific recalibration is required. The findings will inform the generalizability of AI\u0026ndash;GIS tools for global health monitoring and their suitability for cross-country policy learning.\u003c/p\u003e \u003cp\u003eCollectively, these aims and objectives position the study to contribute not only to empirical understanding of child health inequalities but also to methodological advancement in comparative GeoAI research, offering insights into spatial equity, temporal progress, and model transferability across development contexts.\u003c/p\u003e \u003c/div\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study Design and Comparative Framework\u003c/h2\u003e \u003cp\u003eThis study adopts a comparative longitudinal research design to examine child health trajectories in New Zealand and Nigeria over the period 2006\u0026ndash;2022. A longitudinal approach is essential for capturing changes in child health outcomes over time, assessing the pace and consistency of progress, and identifying periods of acceleration or stagnation in response to policy interventions and socioeconomic transitions (Singer \u0026amp; Willett, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). By embedding this temporal dimension within a comparative North\u0026ndash;South framework, the study enables systematic evaluation of how child health evolves under markedly different institutional capacities, income levels, and health system structures.\u003c/p\u003e \u003cp\u003eThe comparative framework is explicitly cross-contextual, contrasting a high-income country with relatively homogeneous health outcomes against a developing country characterized by substantial spatial and socioeconomic heterogeneity. New Zealand represents a mature health system with near-universal coverage and strong primary care coordination, where subnational variations in child health are comparatively limited and often linked to residual socioeconomic or demographic factors (OECD, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Nigeria, by contrast, exhibits pronounced disparities across its states, driven by uneven distribution of health infrastructure, regional economic inequalities, and differential access to education and basic services (UNICEF, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; World Health Organization [WHO], 2022). This contrast provides a robust analytical basis for examining how spatial inequality and temporal progress differ across development contexts.\u003c/p\u003e \u003cp\u003eThe unit of analysis is defined at the subnational level for both countries, allowing for spatially explicit comparison. In New Zealand, the analysis is conducted across subnational regions consistent with administrative and statistical reporting structures, while in Nigeria, the unit of analysis is the state level, reflecting the country\u0026rsquo;s federal governance system and the primary scale at which health policy implementation and resource allocation occur. Using subnational units mitigates the limitations of national averages, which often conceal localized disparities and mask within-country inequality (Victora et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This design choice is particularly critical for Nigeria, where inter-state differences in child health outcomes frequently exceed cross-country differences observed at the national level.\u003c/p\u003e \u003cp\u003eTo operationalize the comparative design, child health outcomes are modeled as spatiotemporal processes. Let \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e denote a child health indicator (e.g., immunization coverage or access to sanitation) for subnational unit \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e at time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e. The longitudinal structure of the data can be expressed as:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i,t}=\\alpha\\:+\\beta\\:{X}_{i,t}+{\\mu\\:}_{i}+{\\lambda\\:}_{t}+{\\epsilon\\:}_{i,t}\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:\\)\u003c/span\u003e\u003c/span\u003e is the intercept, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e is a vector of socioeconomic covariates (education, wealth, residence), \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\mu\\:}_{i}\\)\u003c/span\u003e\u003c/span\u003e captures time-invariant subnational effects, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\lambda\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e represents common temporal shocks or global trends, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\epsilon\\:}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e is the error term (Wooldridge, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). This structure allows comparison of both within-unit temporal change and between-unit spatial variation across the two countries.\u003c/p\u003e \u003cp\u003eThe integration of Artificial Intelligence (AI) and Geographic Information Systems (GIS) strengthens this comparative framework by enabling cross-context analysis of spatial inequality, temporal dynamics, and policy effectiveness. GIS provides the spatial referencing and visualization needed to map subnational disparities and identify clusters of high or low performance, while AI techniques such as machine learning regression and time-series forecasting\u0026mdash;model nonlinear relationships and predict future trajectories (Kamel Boulos \u0026amp; Peng, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Janowicz et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Formally, AI-based prediction can be represented as:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{\\widehat{y}}_{i,t+k}=f\\left({y}_{i,t},{X}_{i,t},{S}_{i}\\right)\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\widehat{y}}_{i,t+k}\\)\u003c/span\u003e\u003c/span\u003e is the predicted health outcome \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e periods ahead, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f(\\cdot\\:)\\)\u003c/span\u003e\u003c/span\u003e denotes a learned nonlinear function (e.g., Random Forest or LSTM), and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{S}_{i}\\)\u003c/span\u003e\u003c/span\u003e represents spatial attributes derived from GIS layers. Applying the same modeling framework to both countries allows direct assessment of model transferability and differential predictive performance across health systems.\u003c/p\u003e \u003cp\u003eOverall, this study design enables a rigorous comparison of child health trajectories by aligning temporal depth, spatial granularity, and analytical methodology across contexts. By combining longitudinal analysis with AI\u0026ndash;GIS integration at the subnational level, the framework provides a robust basis for evaluating how spatial inequality, socioeconomic determinants, and policy effectiveness shape child health outcomes in developed and developing health systems.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Data Sources and Indicators\u003c/h2\u003e \u003cp\u003eThis study draws on harmonized, nationally representative data sources to enable a robust comparative analysis of child health outcomes between Nigeria and New Zealand over the period 2006\u0026ndash;2022. The selection of data sources and indicators is guided by three core principles: cross-country comparability, temporal consistency, and suitability for spatially explicit AI\u0026ndash;GIS analysis.\u003c/p\u003e \u003cp\u003eFor Nigeria, the primary data sources are the Demographic and Health Surveys (DHS) and the UNICEF Multiple Indicator Cluster Surveys (MICS). DHS and MICS are globally standardized household survey programs designed to generate reliable, comparable estimates of population health indicators in low- and middle-income countries (Corsi et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; UNICEF, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). These surveys employ stratified, multistage cluster sampling and provide subnationally disaggregated data at the state or zonal level, making them particularly suitable for spatial and longitudinal analysis. In Nigeria, DHS and MICS constitute the most authoritative and consistently available sources of child health data, covering indicators related to immunization, water and sanitation, treatment of common childhood illnesses, and maternal socioeconomic characteristics (National Population Commission [NPC] \u0026amp; ICF, 2019).\u003c/p\u003e \u003cp\u003eFor New Zealand, child health data are obtained from national administrative and statistical health datasets, including Ministry of Health surveillance systems and Statistics New Zealand\u0026ndash;aligned indicators. These datasets provide comprehensive coverage of child immunization, sanitation access, healthcare utilization, and maternal characteristics at subnational levels, reflecting the country\u0026rsquo;s well-established health information infrastructure (Ministry of Health New Zealand, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). To ensure comparability with DHS and MICS indicators, New Zealand data are harmonized using internationally defined indicator frameworks, particularly those aligned with WHO and UNICEF monitoring standards (World Health Organization [WHO], 2022). This harmonization process ensures that observed differences reflect substantive health system and socioeconomic contrasts rather than measurement inconsistencies.\u003c/p\u003e \u003cp\u003eThe study focuses on a core set of child health and socioeconomic indicators that are widely recognized as proxies for child well-being and health system performance. Immunization coverage, particularly completion of routine childhood vaccinations, is used as an indicator of preventive healthcare access and system reach. Access to improved sanitation reflects environmental health conditions and exposure to infectious disease risks. Treatment of common childhood illnesses, such as diarrhea and respiratory infections, captures responsiveness and accessibility of primary healthcare services. Maternal education is included as a key socioeconomic determinant, given its well-documented influence on health-seeking behavior, childcare practices, and child survival (Victora et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). These indicators are consistently measured across DHS, MICS, and WHO-aligned national systems, enabling valid cross-country comparison.\u003c/p\u003e \u003cp\u003eTo support spatiotemporal analysis, all indicators are structured as panel data indexed by subnational unit \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e and time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e. Let \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e denote a child health indicator, and let \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Z}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e represent the vector of socioeconomic attributes (maternal education, wealth, residence). The indicator framework can be expressed as:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i,t}=g({Z}_{i,t},{H}_{i,t})\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{H}_{i,t}\\)\u003c/span\u003e\u003c/span\u003e represents health system and environmental conditions varying across space and time. This structure allows AI models to learn nonlinear relationships between indicators and determinants while preserving spatial referencing for GIS analysis.\u003c/p\u003e \u003cp\u003eThe integration of Artificial Intelligence (AI) and Geographic Information Systems (GIS) is central to leveraging these data sources for comparative analysis. GIS enables the spatial alignment and visualization of subnational indicators, revealing geographic inequalities and clustering patterns that national summaries obscure. AI methods, applied to harmonized indicators across both countries, allow for identification of differential determinant effects, modeling of temporal progress, and forecasting of future outcomes under contrasting policy environments (Janowicz et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Kamel Boulos \u0026amp; Peng, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). By applying the same indicator definitions and analytical pipelines to Nigeria and New Zealand, the AI\u0026ndash;GIS framework facilitates cross-context comparison of spatial inequality, temporal trajectories, and policy effectiveness, strengthening the validity and interpretability of the comparative findings.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.3 AI and GIS Analytical Techniques\u003c/h2\u003e \u003cp\u003eThis study employs an integrated Artificial Intelligence Geographic Information Systems (AI\u0026ndash;GIS) analytical pipeline to examine spatial inequality, temporal progress, and policy effectiveness in child health outcomes across New Zealand and Nigeria. The approach combines GIS-based spatial statistics for identifying geographic patterns with AI-based predictive models for cross-sectional explanation and longitudinal forecasting, enabling rigorous cross-context comparison under a unified methodological framework.\u003c/p\u003e \u003cp\u003eGIS-Based Spatial Analysis\u003c/p\u003e \u003cp\u003eGeographic Information Systems (GIS) techniques are used to quantify and visualize subnational disparities in child health indicators. Two complementary spatial statistics are applied. First, global spatial autocorrelation is assessed using Moran\u0026rsquo;s I, which measures whether similar values of a health indicator cluster spatially across subnational units (Anselin, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1995\u003c/span\u003e). For subnational unit \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{i}\\)\u003c/span\u003e\u003c/span\u003e at time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{t}\\)\u003c/span\u003e\u003c/span\u003e, Moran\u0026rsquo;s I is expressed as:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\varvec{I}=\\frac{\\varvec{N}}{\\varvec{W}}\\frac{\\sum\\:_{\\varvec{i}}\\sum\\:_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}({\\varvec{y}}_{\\varvec{i},\\varvec{t}}-{\\stackrel{\\prime }{\\varvec{y}}}_{\\varvec{t}})({\\varvec{y}}_{\\varvec{j},\\varvec{t}}-{\\stackrel{\\prime }{\\varvec{y}}}_{\\varvec{t}})}{\\sum\\:_{\\varvec{i}}({\\varvec{y}}_{\\varvec{i},\\varvec{t}}-{\\stackrel{\\prime }{\\varvec{y}}}_{\\varvec{t}}{)}^{2}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{N}\\)\u003c/span\u003e\u003c/span\u003e is the number of spatial units, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{w}}_{\\varvec{i}\\varvec{j}}\\)\u003c/span\u003e\u003c/span\u003e denotes spatial weights, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{W}={\\sum\\:}_{\\varvec{i}}{\\sum\\:}_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}\\)\u003c/span\u003e\u003c/span\u003e, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\stackrel{\\prime }{\\varvec{y}}}_{\\varvec{t}}\\)\u003c/span\u003e\u003c/span\u003e is the mean health indicator at time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{t}\\)\u003c/span\u003e\u003c/span\u003e. This statistic enables comparison of the degree of spatial clustering between the relatively homogeneous New Zealand context and the more heterogeneous Nigerian context.\u003c/p\u003e \u003cp\u003eSecond, local spatial clustering is identified using the Getis\u0026ndash;Ord Gi statistic, which detects statistically significant hotspots (high-performing clusters) and cold spots (low-performing clusters) (Getis \u0026amp; Ord, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e1992\u003c/span\u003e). The Gi statistic is given by:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{G}}_{\\varvec{i}}^{*}=\\frac{\\sum\\:_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}{\\varvec{y}}_{\\varvec{j}}-\\stackrel{\\prime }{\\varvec{y}}\\sum\\:_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}}{\\varvec{S}\\sqrt{\\left[\\frac{\\varvec{N}\\sum\\:_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}^{2}-(\\sum\\:_{\\varvec{j}}{\\varvec{w}}_{\\varvec{i}\\varvec{j}}{)}^{2}}{\\varvec{N}-1}\\right]}}\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{S}\\)\u003c/span\u003e\u003c/span\u003e is the standard deviation of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\varvec{y}\\)\u003c/span\u003e\u003c/span\u003e. Hotspot comparison across countries allows assessment of how spatial inequality manifests differently under contrasting policy and infrastructure conditions.\u003c/p\u003e \u003cp\u003eAI-Based Predictive Modeling\u003c/p\u003e \u003cp\u003eTo complement spatial analysis, AI models are applied for explanatory and predictive purposes. Random Forest (RF) regression is used for cross-sectional comparison of determinant effects. RF is an ensemble learning method that constructs multiple decision trees and aggregates their predictions, offering robustness to nonlinearity and interaction effects (Breiman, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). The RF prediction for unit \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:i\\)\u003c/span\u003e\u003c/span\u003e is:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{\\widehat{y}}_{i}=\\frac{1}{B}\\sum\\:_{b=1}^{B}{T}_{b}\\left({X}_{i}\\right)\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{b}(\\cdot\\:)\\)\u003c/span\u003e\u003c/span\u003e represents the \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:b\\)\u003c/span\u003e\u003c/span\u003e-th decision tree, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:B\\)\u003c/span\u003e\u003c/span\u003e is the number of trees, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}_{i}\\)\u003c/span\u003e\u003c/span\u003e denotes socioeconomic predictors (maternal education, wealth, residence). RF models are estimated separately for New Zealand and Nigeria, enabling direct comparison of cross-sectional relationships within each health system.\u003c/p\u003e \u003cp\u003eFor temporal forecasting, a Long Short-Term Memory (LSTM) neural network is employed to model sequential dependencies in child health indicators over 2006\u0026ndash;2022. LSTM networks are well suited for time-series data due to their gated memory structure, which mitigates vanishing gradient problems (Hochreiter \u0026amp; Schmidhuber, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e1997\u003c/span\u003e). The LSTM cell updates are defined as:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{f}_{t}=\\sigma\\:\\left({W}_{f}\\right[{h}_{t-1},{x}_{t}]+{b}_{f}),{c}_{t}={f}_{t}\\odot\\:{c}_{t-1}+{i}_{t}\\odot\\:\\text{t}\\text{a}\\text{n}\\text{h}\\left({W}_{c}\\right[{h}_{t-1},{x}_{t}]+{b}_{c})\\)\u003c/span\u003e \u003c/span\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{f}_{t}\\)\u003c/span\u003e\u003c/span\u003e is the forget gate, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{c}_{t}\\)\u003c/span\u003e\u003c/span\u003e the cell state, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{x}_{t}\\)\u003c/span\u003e\u003c/span\u003e the input at time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e. Applying the same LSTM architecture to both countries allows comparison of temporal dynamics and forecasted trajectories under different development conditions.\u003c/p\u003e \u003cp\u003eComparative Feature-Importance Analysis\u003c/p\u003e \u003cp\u003eTo evaluate determinant strength across contexts, the study conducts comparative feature-importance analysis using RF-derived importance scores. Feature importance \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:F{I}_{k}\\)\u003c/span\u003e\u003c/span\u003e for determinant \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e is computed as the normalized reduction in prediction error attributable to splits on that variable across all trees (Breiman, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2001\u003c/span\u003e):\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:F{I}_{k}=\\frac{1}{B}\\sum\\:_{b=1}^{B}{\\Delta\\:}{\\text{MSE}}_{k,b}\\)\u003c/span\u003e \u003c/span\u003eComparing \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:F{I}_{k}\\)\u003c/span\u003e\u003c/span\u003e between New Zealand and Nigeria reveals how the relative influence of maternal education, household wealth, and residence differs spatially and institutionally. This comparison provides policy-relevant insights into whether determinants exert attenuated effects in high-capacity systems or amplified effects in spatially fragmented ones.\u003c/p\u003e \u003cp\u003eOverall, the integration of GIS spatial statistics with AI predictive modeling enables cross-context comparison of spatial inequality, temporal progress, and policy effectiveness. By applying identical analytical techniques to harmonized data in both countries, the AI\u0026ndash;GIS framework ensures methodological consistency while illuminating how geography and socioeconomic structure condition child health outcomes across development contexts.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Ethical and Comparative Validity Considerations\u003c/h2\u003e \u003cp\u003eEthical governance and comparative validity are central to the application of AI\u0026ndash;GIS methods in cross-national child health research, particularly when analyses involve spatially referenced data and predictive modeling across jurisdictions with differing legal, institutional, and data governance frameworks. This study adopts a rigorous ethical approach to ensure data privacy, fairness, and methodological comparability between New Zealand and Nigeria, while preserving the analytical integrity required for cross-context comparison of spatial inequality, temporal progress, and policy effectiveness.\u003c/p\u003e \u003cp\u003eFirst, data anonymization is strictly observed across all datasets used in the study. For Nigeria, both the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) release only de-identified, aggregated data, with personal identifiers removed prior to public access (Corsi et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; UNICEF, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Similarly, New Zealand\u0026rsquo;s national health datasets are governed by robust privacy protections under national health information regulations, ensuring that individual-level data are anonymized and reported at aggregated subnational levels (Ministry of Health New Zealand, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). As a result, the study operates exclusively on secondary data that pose no risk of individual re-identification.\u003c/p\u003e \u003cp\u003eSecond, spatial masking and aggregation are applied to address the heightened privacy risks associated with geospatial health data. Precise geographic coordinates where originally collected are intentionally displaced or generalized in DHS and MICS datasets to prevent reverse geocoding of households (Kounadi \u0026amp; Leitner, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In this study, analyses are conducted at the state level in Nigeria and subnational administrative regions in New Zealand, ensuring that spatial resolution is sufficiently coarse to protect confidentiality while remaining analytically meaningful. This approach balances ethical safeguards with the need to identify spatial clustering and regional disparities using GIS-based statistics such as Moran\u0026rsquo;s I and Getis\u0026ndash;Ord Gi.\u003c/p\u003e \u003cp\u003eThird, ensuring ethical comparability across jurisdictions is critical in a North\u0026ndash;South comparative framework. Health data governance standards differ substantially between high-income and developing contexts, particularly in relation to consent procedures, data access, and institutional oversight. To mitigate these differences, the study relies exclusively on data sources that have undergone prior ethical review and comply with international best practices for health research, including informed consent at the point of data collection and anonymization prior to dissemination (World Health Organization [WHO], 2022). This harmonized ethical baseline ensures that observed differences between New Zealand and Nigeria reflect substantive health system and socioeconomic contrasts rather than inconsistencies in ethical or data governance standards.\u003c/p\u003e \u003cp\u003eFrom a comparative validity perspective, ethical considerations are closely linked to methodological rigor. AI\u0026ndash;GIS integration introduces additional risks of algorithmic bias, particularly when models trained in data-rich, stable environments are applied to more heterogeneous or data-constrained settings (Leslie et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). To address this concern, identical preprocessing pipelines, spatial aggregation rules, and model architectures are applied across both countries. This consistency ensures that differences in spatial inequality patterns, temporal trajectories, or predictive performance are attributable to contextual factors such as policy effectiveness or socioeconomic structure rather than methodological artifacts.\u003c/p\u003e \u003cp\u003eMoreover, the use of explainable AI\u0026ndash;compatible techniques, such as feature-importance analysis in Random Forest models, enhances ethical transparency by allowing policymakers to understand which determinants drive predictions in each context. When combined with GIS-based visualization, these outputs support responsible interpretation of spatial disparities without stigmatizing specific regions or populations. This aligns with emerging ethical guidance on the use of AI in public health, which emphasizes accountability, interpretability, and proportionality in decision support systems (Leslie et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn sum, the ethical framework underpinning this study ensures that AI\u0026ndash;GIS integration supports valid and responsible cross-context comparison. By combining anonymization, spatial masking, harmonized ethical standards, and transparent modeling practices, the study safeguards individual privacy while enabling robust analysis of spatial inequality, temporal progress, and policy effectiveness across developed and developing health systems.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Descriptive and Temporal Comparison\u003c/h2\u003e \u003cp\u003eThis section presents a descriptive and temporal comparison of child health indicators in New Zealand and Nigeria over the period 2006\u0026ndash;2022, focusing on differences between national averages and subnational trajectories. The comparison highlights two contrasting patterns: long-term stability and convergence in New Zealand versus pronounced variability and regional divergence in Nigeria.\u003c/p\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e3.1.1 National-Level Temporal Trends\u003c/h2\u003e \u003cp\u003eAt the national scale, New Zealand exhibits consistently high child health performance across the study period. Core indicators such as routine immunization coverage, access to improved sanitation, and treatment of common childhood illnesses remain near universal, with only marginal year-to-year variation. This temporal stability reflects the maturity of the health system, strong preventive care infrastructure, and sustained policy continuity.\u003c/p\u003e \u003cp\u003eNigeria, by contrast, shows gradual national-level improvement but with substantially lower baseline values and greater temporal fluctuation. While national averages suggest progress particularly in immunization and access to care these aggregate trends conceal substantial within-country disparities.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes indicative national averages for selected indicators.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eNational Mean Child Health Indiators (%)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2006\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2010\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2015\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e94.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e52.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e55.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e94.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote: Values represent composite means of immunization coverage, sanitation access, and treatment of childhood illnesses.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates early convergence at high levels in New Zealand, compared with slow upward movement from a low baseline in Nigeria.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e \u003ch2\u003e3.1.2 Subnational Patterns and Dispersion\u003c/h2\u003e \u003cp\u003eSubnational analysis reveals stark contrasts between the two countries. In New Zealand, regional indicators cluster tightly around the national mean, with relatively small dispersion across districts and regions. Over time, lower-performing regions gradually converge toward the national average, indicating effective spatial equity mechanisms.\u003c/p\u003e \u003cp\u003eNigeria demonstrates the opposite pattern. State-level analysis shows wide dispersion around the national mean, with some states achieving moderate improvements while others stagnate or improve only marginally. The gap between high- and low-performing states remains persistent over time, signaling regional divergence rather than convergence.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSubnational Dispersion of Child Health Indicators (Standard Deviation, %)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2006\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2010\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2015\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e12.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe declining dispersion in New Zealand indicates spatial convergence, while the increasing dispersion in Nigeria reflects growing regional inequality, even as national averages improve.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e3.1.3 Temporal Trajectories Across Subnational Units\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents an integrated visualization of temperature-related health risk responses across temporal, demographic, climatic, and socioeconomic dimensions. Panel A shows a pronounced U-shaped relationship between temperature deviation and population-weighted mean risk, with minimal risk at the local thermal optimum and progressively steeper gradients over time, indicating growing vulnerability in recent years. Panels B and C reveal systematic heterogeneity in risk elasticity across ethnic groups in New Zealand and Nigeria, with consistently higher elasticities observed in Nigeria, reflecting greater sensitivity to thermal stress. Panel D demonstrates sex-based differences, where males exhibit higher risk elasticity than females in both countries, while Panel E highlights elevated vulnerability in rural compared to urban populations, particularly in Nigeria. Panel F indicates that humid tropical and semi-arid climate zones experience stronger sensitivity to temperature deviations than temperate maritime zones. Panel G shows a monotonic increase in response intensity with worsening socioeconomic deprivation, again with sharper escalation in Nigeria. Finally, Panel H illustrates that poor housing quality substantially amplifies temperature-related risk, with deteriorated housing conditions exhibiting the highest sensitivity. Collectively, the figure underscores the compounding effects of climate exposure, social structure, and built-environment quality on health risk under thermal stress. New Zealand shows tight clustering and convergence, while Nigeria displays wide dispersion and divergence.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.1.4 Interpretation\u003c/h2\u003e \u003cp\u003eThe descriptive and temporal comparison demonstrates that national averages are insufficient for understanding child health dynamics, particularly in heterogeneous developing contexts. In New Zealand, national trends reliably reflect subnational realities due to limited spatial inequality. In Nigeria, however, national improvements coexist with entrenched regional disparities that only become visible through subnational, longitudinal analysis.\u003c/p\u003e \u003cp\u003eThese findings underscore the value of AI\u0026ndash;GIS integration, which enables systematic tracking of temporal progress while preserving spatial resolution. By revealing convergence in high-capacity systems and divergence in fragmented ones, the approach provides a more accurate foundation for evaluating policy effectiveness and targeting interventions where they are most needed.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Spatial Pattern Comparison (GIS Results)\u003c/h2\u003e \u003cp\u003eThis section compares the spatial distribution of child health indicators in Nigeria and New Zealand, using GIS-based spatial statistics to identify clustering, hotspots, and cold spots. The results reveal structurally different spatial regimes: pronounced, socioeconomic-aligned clustering in Nigeria versus limited clustering and spatial equity in New Zealand.\u003c/p\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1 Spatial Autocorrelation (Global Patterns)\u003c/h2\u003e \u003cp\u003eGlobal spatial autocorrelation was assessed using Moran\u0026rsquo;s I, which measures whether similar child health outcomes cluster geographically.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eGlobal Spatial Autocorrelation of Child Health Indicators (2006\u0026ndash;2022)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMoran\u0026rsquo;s I\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ez-score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWeak, non-significant clustering\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eStrong, significant clustering\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eNigeria exhibits strong positive spatial autocorrelation, indicating that high- and low-performing states tend to be geographically contiguous. In contrast, New Zealand\u0026rsquo;s Moran\u0026rsquo;s I is low and statistically insignificant, suggesting that child health outcomes are largely spatially random and not strongly conditioned by location.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2 Local Hotspot and Cold Spot Analysis\u003c/h2\u003e \u003cp\u003eTo identify localized spatial structures, the Getis\u0026ndash;Ord Gi statistic was applied. This analysis distinguishes statistically significant hotspots (high-performing clusters) and cold spots (low-performing clusters).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eHotspot\u0026ndash;Cold Spot Structure of Child Health Outcomes\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCountry\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHotspots Identified\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCold Spots Identified\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDominant Alignment\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSouthern \u0026amp; South-West states\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNorthern states\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eWealth \u0026amp; maternal education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNone (sporadic only)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNone\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNo systematic alignment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn Nigeria, hotspots are predominantly concentrated in southern and south-western states, which also exhibit higher household wealth levels, better maternal education attainment, and stronger health infrastructure. Cold spots are concentrated in northern regions, where poverty rates are higher and female educational attainment is lower. This spatial alignment confirms that socioeconomic inequality is spatially embedded.\u003c/p\u003e \u003cp\u003eIn New Zealand, no persistent or large-scale hotspots or cold spots are detected. Minor local variations exist, but they are statistically weak and transient, indicating that national policies and universal service coverage have largely neutralized geographic disadvantage.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3 Spatial Visualization of Patterns\u003c/h2\u003e \u003cp\u003eThe contrast between the two countries is illustrated conceptually in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cem\u003eNigeria shows distinct hotspot\u0026ndash;cold spot structures aligned with socioeconomic gradients, while New Zealand exhibits a largely uniform spatial distribution.\u003c/em\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e \u003ch2\u003e3.2.4 Interpretation\u003c/h2\u003e \u003cp\u003eThe GIS results demonstrate that spatial inequality is a defining feature of child health outcomes in Nigeria, whereas spatial equity characterizes New Zealand. In Nigeria, geographic location acts as a proxy for access to education, income, and healthcare, producing entrenched regional clusters of advantage and disadvantage. These clusters persist over time, reinforcing intergenerational health inequality.\u003c/p\u003e \u003cp\u003eBy contrast, New Zealand\u0026rsquo;s limited spatial clustering indicates that policy effectiveness operates uniformly across space, minimizing the role of geography in determining child health outcomes. This finding underscores the success of universal health coverage, centralized standards, and coordinated social policy in dampening spatial disparities.\u003c/p\u003e \u003cp\u003eOverall, the spatial comparison highlights the analytical strength of AI\u0026ndash;GIS integration. GIS reveals where inequality is spatially concentrated, while AI-driven analysis (presented in subsequent sections) explains why these patterns persist and how they may evolve. Together, these tools enable a nuanced cross-context comparison of spatial inequality, temporal progress, and policy effectiveness that cannot be achieved through national averages alone.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Interpretation Through a Health Systems Lens\u003c/h2\u003e \u003cp\u003eThis section interprets the AI\u0026ndash;GIS findings through a health systems lens, focusing on how governance quality, infrastructure strength, and social policy design condition the influence of key socioeconomic determinants. The comparative analysis explains why maternal education emerges as the dominant predictor of child health outcomes in Nigeria, while its marginal effect plateaus in New Zealand, despite high overall performance in both contexts.\u003c/p\u003e \u003cdiv id=\"Sec22\" class=\"Section3\"\u003e \u003ch2\u003e3.4.1 Governance, Infrastructure, and the Moderation of Determinants\u003c/h2\u003e \u003cp\u003eAI-based feature-importance analysis revealed that the strength of socioeconomic determinants is moderated by health system capacity rather than acting independently. In Nigeria, fragmented governance structures, uneven subnational financing, and variable service delivery capacity amplify the influence of household-level characteristics on child health outcomes. Where public systems are weak or inconsistently implemented, individual attributes such as maternal education become decisive in determining whether children access preventive and curative services.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRelative Strength of Health System Moderators\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDimension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGovernance coherence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHealth infrastructure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUniversal\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUneven\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePolicy enforcement\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStrong\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInconsistent\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReliance on household resources\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn contrast, New Zealand\u0026rsquo;s centralized governance, strong regulatory oversight, and universal health coverage reduce reliance on individual socioeconomic resources. Health services are delivered through standardized national frameworks, ensuring consistent access regardless of region or household background. As a result, AI models detect lower sensitivity of outcomes to socioeconomic variation, reflecting institutional buffering effects.\u003c/p\u003e \u003cp\u003eThese structural contrasts explain why spatial clustering and determinant dominance are pronounced in Nigeria but attenuated in New Zealand.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003e3.4.2 Comparative Feature-Importance Results\u003c/h2\u003e \u003cp\u003eRandom Forest models were used to estimate the relative contribution of key determinants to child health outcomes in each country.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparative Feature Importance (%)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeterminant\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMaternal education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e38%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHousehold wealth\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e32%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResidence (urban\u0026ndash;rural)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime (policy progress)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e48%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn Nigeria, maternal education accounts for the largest share of predictive power (38%), reflecting its role in health literacy, care-seeking behavior, immunization compliance, and nutrition practices. In New Zealand, temporal and policy-driven factors dominate, indicating that system-level improvements explain most outcome variation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section3\"\u003e \u003ch2\u003e3.4.3 Why Maternal Education Dominates in Nigeria\u003c/h2\u003e \u003cp\u003eMaternal education dominates AI predictions in Nigeria because it operates as a multiplier of access in a context where health systems are uneven. Educated mothers are more likely to:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eNavigate fragmented health services,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDemand preventive care (e.g., immunization),\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAdopt improved sanitation and hygiene practices,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eOvercome geographic and informational barriers.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eIn low-capacity environments, education compensates for systemic gaps. This is reflected in strong spatial overlap between education hotspots and child health hotspots identified in GIS analysis.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the subnational relationship between childhood immunization coverage and maternal health service utilization, highlighting how access to preventive and maternal care varies geographically. Areas shaded toward darker tones indicate regions where both immunization uptake and maternal service coverage are high, while lighter or contrasting colors reveal imbalances or gaps between the two indicators. Overall, the visualization underscores spatial inequalities in healthcare access, emphasizing the need for targeted, location-specific health interventions.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cem\u003eHigh-performing regions align closely with higher female educational attainment.\u003c/em\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003e3.4.4 Why Maternal Education Plateaus in New Zealand\u003c/h2\u003e \u003cp\u003eIn New Zealand, maternal education exhibits a plateau effect because baseline educational attainment and service access are already high. Once universal health coverage, school-based health services, and standardized maternal care are in place, additional gains from education yield diminishing marginal returns.\u003c/p\u003e \u003cp\u003eThis phenomenon is reflected in:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eMinimal spatial clustering of education-linked outcomes,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eLow variance across regions,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eStrong dominance of time and policy variables in AI models.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cem\u003eLimited clustering indicates effective institutional equalization.\u003c/em\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e3.4.5 Interpretation and Policy Implications\u003c/h2\u003e \u003cp\u003eThe comparative interpretation confirms that AI-identified determinants are system-contingent, not universal in strength. Maternal education acts as a primary driver in Nigeria because governance and infrastructure gaps magnify household-level effects. In New Zealand, the same determinant is absorbed by institutional capacity, shifting explanatory power toward system-wide policy continuity.\u003c/p\u003e \u003cp\u003eFrom a policy perspective:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eNigeria requires education-sensitive and regionally targeted health strategies to compensate for system gaps.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eNew Zealand benefits more from maintaining institutional quality and monitoring residual inequities rather than determinant-focused interventions.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eOverall, the AI\u0026ndash;GIS framework reveals not only \u003cem\u003ewhich\u003c/em\u003e determinants matter, but \u003cem\u003ewhy\u003c/em\u003e their influence differs across contexts demonstrating how governance quality and infrastructure fundamentally shape the geography and trajectory of child health outcomes.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Interpretation Through a Health Systems Lens\u003c/h2\u003e \u003cp\u003eThis section interprets the AI\u0026ndash;GIS results by situating them within a health systems framework, emphasizing how governance quality, infrastructure robustness, and social policy design condition the strength and spatial expression of AI-identified determinants. The comparative evidence explains why maternal education is the dominant predictor of child health outcomes in Nigeria, while its marginal explanatory power plateaus in New Zealand.\u003c/p\u003e \u003cdiv id=\"Sec28\" class=\"Section3\"\u003e \u003ch2\u003e3.4.1 Governance, Infrastructure, and Moderation of Determinants\u003c/h2\u003e \u003cp\u003eAI models identify socioeconomic determinants not as fixed drivers, but as system-contingent variables whose effects are amplified or dampened by institutional capacity. In Nigeria, governance fragmentation, uneven subnational financing, and disparities in primary healthcare coverage magnify household-level attributes. Where service availability and enforcement vary widely across states, individual capabilities particularly maternal education become decisive in determining whether children receive immunization, timely treatment, and safe sanitation.\u003c/p\u003e \u003cp\u003eBy contrast, New Zealand operates under a highly centralized and standardized health system with near-universal access to preventive and curative services. Strong regulatory oversight and nationally uniform policy implementation substantially reduce geographic and socioeconomic barriers. As a result, AI models attribute a larger share of outcome variation to system-level temporal effects (policy continuity and cumulative investment) rather than to household characteristics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eHealth System Moderators of Determinant Strength\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDimension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGovernance coherence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHigh (centralized standards)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVariable (federal, uneven enforcement)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHealth infrastructure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUniversal, high quality\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUneven, regionally concentrated\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePolicy implementation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eConsistent nationwide\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eState-dependent, inconsistent\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDependence on household resources\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThis contrast explains the divergent spatial patterns observed in GIS analysis: limited clustering in New Zealand versus pronounced hotspot\u0026ndash;cold spot structures in Nigeria.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section3\"\u003e \u003ch2\u003e3.4.2 Comparative Feature-Importance Outcomes\u003c/h2\u003e \u003cp\u003eRandom Forest feature-importance analysis quantifies how determinant strength differs across contexts.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparative Feature Importance of Key Determinants (%)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeterminant\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNew Zealand\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNigeria\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMaternal education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e38\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHousehold wealth\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResidence (urban\u0026ndash;rural)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime / policy continuity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn Nigeria, maternal education contributes the largest share of predictive power (38%), reflecting its role in health literacy, care-seeking behavior, and the ability to navigate fragmented services. In New Zealand, the dominant contribution comes from time and policy continuity (48%), indicating that long-term institutional stability explains most remaining variation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section3\"\u003e \u003ch2\u003e3.4.3 Why Maternal Education Dominates in Nigeria\u003c/h2\u003e \u003cp\u003eMaternal education dominates predictions in Nigeria because it functions as a substitute for weak or uneven systems. Educated mothers are more likely to:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eUnderstand vaccination schedules and treatment pathways,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTravel to higher-quality facilities when local services are inadequate,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAdopt improved sanitation and hygiene practices,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAdvocate for care within constrained institutional settings.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThese behaviors generate strong spatial overlap between education levels and child health outcomes, reinforcing the hotspot\u0026ndash;cold spot patterns identified by GIS.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec31\" class=\"Section3\"\u003e \u003ch2\u003e3.4.4 Why Maternal Education Plateaus in New Zealand\u003c/h2\u003e \u003cp\u003eIn New Zealand, maternal education exhibits a plateau effect because baseline access to healthcare, sanitation, and preventive services is already high. Once universal coverage and standardized maternal-child programs are in place, additional educational gains yield diminishing marginal returns for child health outcomes. Consequently:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eSpatial variation linked to education is minimal,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSubnational indicators cluster tightly around national means,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAI models attribute residual variation primarily to temporal policy effects rather than household characteristics.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section3\"\u003e \u003ch2\u003e3.4.5 Synthesis and Implications\u003c/h2\u003e \u003cp\u003eInterpreted through a health systems lens, the AI\u0026ndash;GIS results demonstrate that determinant strength is not universal but moderated by governance and infrastructure. In Nigeria, maternal education and wealth dominate because households must compensate for system gaps. In New Zealand, strong institutions absorb these effects, shifting explanatory power toward policy continuity and long-term investment.\u003c/p\u003e \u003cp\u003eThis comparative interpretation underscores the analytical value of AI\u0026ndash;GIS integration: it reveals not only \u003cem\u003ewhich\u003c/em\u003e determinants matter, but \u003cem\u003ewhy\u003c/em\u003e their influence varies across contexts linking spatial inequality and temporal progress directly to health system design and policy effectiveness.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"4. Conclusion and Recommendations","content":"\u003cdiv id=\"Sec34\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Comparative Conclusions\u003c/h2\u003e \u003cp\u003eThe comparative application of AI\u0026ndash;GIS methods to child health outcomes in Nigeria and New Zealand reveals two fundamentally different inequality regimes structural inequality in Nigeria and residual inequality in New Zealand. These regimes shape not only current spatial and socioeconomic disparities but also the limits and nature of future predictive gains identified by AI models.\u003c/p\u003e \u003cp\u003eIn Nigeria, AI\u0026ndash;GIS analysis demonstrates that inequality in child health outcomes is structural, deeply embedded in governance arrangements, infrastructure distribution, and social policy effectiveness. Pronounced hotspot\u0026ndash;cold spot patterns align closely with maternal education, household wealth, and urban\u0026ndash;rural residence, indicating that geography acts as a conduit for persistent disadvantage. Predictive models show that improvements in child health are highly sensitive to changes in socioeconomic determinants, particularly maternal education, because households must compensate for uneven service provision and inconsistent policy implementation. As a result, predictive gains in Nigeria are policy-limited: AI forecasts indicate that substantial improvements are achievable only if structural constraints such as weak subnational governance, fragmented financing, and unequal access to primary healthcare are addressed. Without systemic reform, predictive gains plateau prematurely in lagging regions, reinforcing spatial divergence even as national averages improve.\u003c/p\u003e \u003cp\u003eBy contrast, New Zealand exhibits a pattern of residual inequality, where most structural barriers to child health have already been mitigated through universal health coverage, centralized governance, and robust social protection systems. GIS analysis reveals limited spatial clustering, and AI models attribute remaining variation primarily to temporal and policy-continuity factors rather than to household-level socioeconomic determinants. In this context, maternal education and wealth show diminishing marginal effects, reflecting an environment in which baseline access to care is already high. Consequently, predictive gains in New Zealand are saturation-limited rather than policy-limited. AI forecasting suggests that future improvements in child health outcomes will be incremental, constrained by biological, demographic, and system-efficiency ceilings rather than by inequitable access or governance failures.\u003c/p\u003e \u003cp\u003eTaken together, these findings underscore the comparative value of AI\u0026ndash;GIS integration. The approach distinguishes contexts where inequality is driven by structural system deficits from those where inequality persists only at the margins. For developing health systems such as Nigeria\u0026rsquo;s, AI\u0026ndash;GIS highlights the necessity of governance and infrastructure reform to unlock predictive gains. For high-income systems like New Zealand\u0026rsquo;s, the same tools shift analytical focus toward fine-grained monitoring, early detection of emerging vulnerabilities, and optimization of already high-performing policies. This comparative conclusion demonstrates that AI\u0026ndash;GIS is not merely a descriptive technology, but a diagnostic and prognostic framework capable of aligning health policy priorities with the underlying nature of inequality across development contexts.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec35\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Policy and Practice Recommendations\u003c/h2\u003e \u003cp\u003eThe comparative findings from the AI\u0026ndash;GIS analysis indicate that policy and practice interventions must be context-specific, reflecting the fundamentally different inequality regimes and health system capacities in Nigeria and New Zealand. Accordingly, recommendations for each country emphasize distinct strategic priorities: structural gap reduction and targeted investment in Nigeria, and system optimization and early-warning surveillance in New Zealand.\u003c/p\u003e \u003cp\u003ePolicy and Practice Recommendations for Nigeria\u003c/p\u003e \u003cp\u003eIn Nigeria, where child health inequality is structurally embedded and spatially concentrated, policy interventions should prioritize spatially targeted maternal education and poverty-sensitive health investments. The AI\u0026ndash;GIS results demonstrate that maternal education is the most influential determinant of child health outcomes and that its effects are geographically clustered. This indicates that national, uniform interventions are insufficient; instead, region-specific strategies are required to address entrenched disadvantage.\u003c/p\u003e \u003cp\u003eFirst, maternal education initiatives should be geographically targeted to cold-spot regions identified through GIS analysis, particularly in states exhibiting persistently low child health indicators. Policies should integrate adult female literacy, reproductive health education, and community-based health promotion into existing primary healthcare platforms. By focusing on spatially defined high-risk areas, these interventions can produce disproportionate gains where institutional capacity is weakest.\u003c/p\u003e \u003cp\u003eSecond, health investments must be poverty-sensitive and spatially prioritized. AI predictions show that improvements in infrastructure and service access yield higher marginal returns in low-performing states. Public health financing should therefore be allocated using AI\u0026ndash;GIS informed criteria that weight need, deprivation intensity, and spatial spillover risk rather than population size alone. Investments in primary healthcare facilities, skilled birth attendance, and sanitation infrastructure should be synchronized with social protection programs to address the multidimensional nature of disadvantage.\u003c/p\u003e \u003cp\u003eThird, governance mechanisms should embed AI\u0026ndash;GIS analytics within subnational planning processes. State-level health authorities can use spatial dashboards to monitor intervention uptake, detect stagnation, and dynamically reallocate resources. This practice-oriented integration would transform AI\u0026ndash;GIS from a diagnostic tool into an operational decision-support system, improving accountability and policy responsiveness.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePolicy and Practice Recommendations for New Zealand\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn New Zealand, where baseline child health outcomes are high and spatial inequality is limited, AI\u0026ndash;GIS should be deployed primarily as a monitoring and early-warning optimization tool rather than as a gap-filling mechanism. The study\u0026rsquo;s findings indicate that remaining disparities are residual and context-specific, requiring precision rather than broad structural reform.\u003c/p\u003e \u003cp\u003eFirst, AI\u0026ndash;GIS systems should be institutionalized to support continuous surveillance and early detection of emerging risks. Predictive analytics can identify subtle temporal shifts or localized deviations from expected trends, enabling policymakers to intervene before disparities widen. This is particularly relevant for monitoring vulnerable subpopulations or regions experiencing demographic change or service strain.\u003c/p\u003e \u003cp\u003eSecond, GIS-based visualization should be used to evaluate policy effectiveness over time, ensuring that existing maternal and child health programs maintain equitable reach across all regions. By linking policy inputs to spatially explicit outcome trends, decision-makers can assess whether interventions remain proportionate and efficient in a near-saturated system.\u003c/p\u003e \u003cp\u003eThird, New Zealand can leverage its advanced data infrastructure to pilot AI\u0026ndash;GIS best practices with broader international relevance. Rather than focusing on access expansion, policy efforts should emphasize model validation, ethical AI governance, and interoperability, positioning AI\u0026ndash;GIS as a tool for fine-tuning high-performing systems and informing transferable methodologies for global health monitoring.\u003c/p\u003e \u003cp\u003e \u003cb\u003eComparative Synthesis\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTogether, these recommendations reinforce the central conclusion that AI\u0026ndash;GIS does not prescribe uniform solutions. In Nigeria, the technology guides where and how to intervene to dismantle structural inequality. In New Zealand, it enhances how well existing systems are monitored and optimized. Aligning policy responses with these context-specific insights ensures that AI\u0026ndash;GIS integration contributes meaningfully to equitable, efficient, and sustainable child health outcomes across development contexts.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec36\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Methodological Contributions\u003c/h2\u003e \u003cp\u003eThis study makes important methodological contributions to the application of Geographic Artificial Intelligence (GeoAI) in global health research by demonstrating both the scalability and the contextual limits of AI\u0026ndash;GIS approaches across divergent development settings. By applying a unified analytical framework to Nigeria and New Zealand, the study advances understanding of how GeoAI performs when transferred between low- and middle-income country (LMIC) contexts and high-income country (HIC) health systems.\u003c/p\u003e \u003cp\u003eFirst, the findings demonstrate the scalability of GeoAI methods across development contexts with markedly different data environments and institutional capacities. The successful integration of GIS-based spatial statistics with machine learning and deep learning models across both countries shows that GeoAI can accommodate heterogeneous data structures, varying spatial resolutions, and distinct temporal dynamics. By harmonizing indicators and applying identical preprocessing, modeling, and validation procedures, the study confirms that GeoAI frameworks can be scaled from data-constrained, spatially fragmented systems to data-rich, stable systems without sacrificing analytical coherence. This scalability is particularly important for global health monitoring, where comparative insights increasingly depend on the ability to analyze diverse datasets under a common methodological architecture.\u003c/p\u003e \u003cp\u003eAt the same time, the study clearly identifies the limits of GeoAI transferability. Predictive performance and determinant sensitivity differ systematically across contexts, reflecting the influence of governance quality, infrastructure strength, and policy consistency. In Nigeria, GeoAI models are highly responsive to socioeconomic variation and spatial inequality, whereas in New Zealand, predictive gains plateau as outcomes approach saturation. This contrast underscores a critical methodological insight: GeoAI does not function as a context-neutral tool. Model outputs and explanatory structures must be interpreted in light of system capacity and development stage, and direct model transfer without contextual recalibration risks misinterpretation. Recognizing these limits strengthens, rather than weakens, the methodological credibility of GeoAI by grounding predictions in institutional reality.\u003c/p\u003e \u003cp\u003eSecond, the study establishes a replicable comparative framework that can be readily applied to other LMIC\u0026ndash;HIC pairings. The framework integrates four core components: (i) harmonized indicator selection aligned with international standards; (ii) subnational spatial units to capture within-country inequality; (iii) combined GIS spatial statistics and AI predictive modeling; and (iv) comparative interpretation anchored in health systems theory. This modular design allows researchers to substitute countries, indicators, or time horizons while preserving analytical consistency. As a result, the framework offers a transferable blueprint for comparative GeoAI studies examining child health, education, or other social outcomes across development contexts.\u003c/p\u003e \u003cp\u003eOverall, these methodological contributions position the study as a bridge between technical GeoAI innovation and applied comparative health systems research. By clarifying where GeoAI scales effectively and where its limits emerge, and by providing a replicable comparative design, the study enhances the methodological rigor, transparency, and global applicability of AI\u0026ndash;GIS approaches in development and public health research.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec37\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Future Research Directions\u003c/h2\u003e \u003cp\u003eBuilding on the methodological and empirical insights of this comparative study, several future research directions emerge that can further advance the application of GeoAI in global child health and health systems analysis. These directions focus on expanding the comparative scope, enriching model inputs, and strengthening the policy relevance of AI\u0026ndash;GIS\u0026ndash;based research across development contexts.\u003c/p\u003e \u003cp\u003eFirst, future studies should extend the comparative framework to multi-country panels that include multiple low-, middle-, and high-income countries. Expanding beyond a single LMIC\u0026ndash;HIC pairing would enable researchers to examine whether the patterns observed in Nigeria and New Zealand hold across broader regional and institutional settings. Multi-country panel designs would allow for hierarchical modeling of country-level, subnational, and temporal effects, improving the ability to disentangle global trends from context-specific dynamics. Such panels would also facilitate comparative analysis of convergence and divergence trajectories, helping to identify clusters of countries with similar inequality regimes and policy responses. From a GeoAI perspective, larger panels would provide richer training data, enabling more robust cross-validation and improving the generalizability of predictive models.\u003c/p\u003e \u003cp\u003eSecond, future research should integrate additional contextual layers particularly climate, governance, and health financing variables into GeoAI models. Climate-related factors such as temperature variability, rainfall patterns, and exposure to extreme weather events increasingly influence child health through pathways related to nutrition, water quality, and disease transmission. Incorporating high-resolution climate data into GIS layers would allow GeoAI models to capture environmental shocks and assess their spatial interaction with socioeconomic vulnerability. Similarly, explicit governance indicators such as subnational public sector effectiveness, decentralization intensity, and policy implementation capacity would help explain why identical interventions yield different outcomes across regions.\u003c/p\u003e \u003cp\u003eHealth financing data represent another critical extension. Integrating subnational health expenditure, donor funding flows, and insurance coverage into AI\u0026ndash;GIS models would enable researchers to assess how financial inputs translate into spatially differentiated outcomes over time. This integration would strengthen causal interpretation by linking resource allocation to observed changes in child health indicators, rather than inferring effects solely from household-level determinants. Together, these additional layers would transform GeoAI from a predominantly outcome-driven tool into a systems-level analytical framework capable of modeling interactions among environment, institutions, financing, and population health.\u003c/p\u003e \u003cp\u003eIn combination, extending analyses to multi-country panels and enriching GeoAI models with climate, governance, and financing dimensions would substantially enhance the explanatory and predictive power of comparative health research. These directions position GeoAI as a cornerstone methodology for future studies seeking to understand complex, multi-scalar drivers of child health inequality and to inform adaptive, evidence-based policy across diverse development contexts.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eConflict of Interest Statement\u003c/h2\u003e \u003cp\u003eThe authors declare that there are no known financial, professional, or personal conflicts of interest that could have influenced the design, execution, analysis, interpretation, or reporting of the research presented in this study. The study was conducted independently, and no external organization, funding body, or institutional affiliation exerted influence over the research process or the conclusions drawn.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003e \u003cb\u003eEthics and Consent to Participate\u003c/b\u003e \u003c/h2\u003e \u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003ewere not required for this study. The research is based exclusively on secondary, anonymized, and publicly available data obtained from internationally recognized sources, including Demographic and Health Surveys (DHS), Multiple Indicator Cluster Surveys (MICS), and national administrative health statistics. All original data collection procedures were conducted by the respective data-providing institutions in accordance with their ethical guidelines.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent to Publish\u003c/strong\u003e \u003cp\u003eConsent to publish declaration: not applicable.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis research received no external funding. The study was conducted independently by the authors without financial support from public, commercial, or non-profit funding agencies.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAuthor Contributions StatementJumoke I. Ogunremi: Conceptualized the study design, led the development of the comparative AI\u0026ndash;GIS framework, and drafted the initial manuscript.Nelson N. Igwilo: Conducted data acquisition, preprocessing, and GIS-based spatial analysis; contributed to interpretation of results and manuscript revisions.Oladayo O. Babalola: Implemented machine learning and deep learning models, performed statistical analyses, and contributed to writing the methods and results sections.All authors contributed to the discussion of findings, reviewed and approved the final manuscript, and agree to be accountable for all aspects of the work.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data underlying this study are drawn from publicly available, nationally representative sources. For Nigeria, child health indicators were obtained from the Demographic and Health Surveys (DHS) and UNICEF Multiple Indicator Cluster Surveys (MICS), which can be accessed through the DHS Program website (https://dhsprogram.com) and UNICEF MICS portal (https://mics.unicef.org) upon registration. For New Zealand, data were sourced from the Ministry of Health and Statistics New Zealand, available through official government portals (https://www.health.govt.nz and https://www.stats.govt.nz). All datasets are anonymized and accessible to researchers subject to the respective organizations\u0026rsquo; data use agreements. No new data were generated in this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAnselin L. Local indicators of spatial association LISA. Geographical Anal. 1995;27(2):93\u0026ndash;115. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1538-4632.1995.tb00338.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1538-4632.1995.tb00338.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman L. Random forests. Mach Learn. 2001;45(1):5\u0026ndash;32. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1023/A:1010933404324\u003c/span\u003e\u003cspan address=\"10.1023/A:1010933404324\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCorsi DJ, Neuman M, Finlay JE, Subramanian SV. Demographic and Health Surveys: A profile. Int J Epidemiol. 2012;41(6):1602\u0026ndash;13. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/ije/dys184\u003c/span\u003e\u003cspan address=\"10.1093/ije/dys184\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGetis A, Ord JK. The analysis of spatial association by use of distance statistics. Geographical Anal. 1992;24(3):189\u0026ndash;206. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1538-4632.1992.tb00261.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1538-4632.1992.tb00261.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997;9(8):1735\u0026ndash;80. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1162/neco.1997.9.8.1735\u003c/span\u003e\u003cspan address=\"10.1162/neco.1997.9.8.1735\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJanowicz K, Gao S, McKenzie G, Hu Y, Bhaduri B. GeoAI: Spatially explicit artificial intelligence techniques for geographic knowledge discovery and beyond. Int J Geogr Inf Sci. 2020;34(4):625\u0026ndash;36. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/13658816.2019.1684500\u003c/span\u003e\u003cspan address=\"10.1080/13658816.2019.1684500\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKamel Boulos MN, Peng G. GeoAI: Geographic artificial intelligence for geospatial data science and analytics. Int J Health Geogr. 2019;18(1):1\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12942-019-0178-2\u003c/span\u003e\u003cspan address=\"10.1186/s12942-019-0178-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKounadi O, Leitner M. Why does geoprivacy matter? The scientific publication of confidential data presented on maps. J Empir Res Hum Res Ethics. 2014;9(4):34\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1556264614549511\u003c/span\u003e\u003cspan address=\"10.1177/1556264614549511\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeslie D, Mazumder A, Peppin A, Wolters MK, Hagerty A. Ethical guidelines for trustworthy AI. The Alan Turing Institute; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health New Zealand. Annual data explorer 2021/22: New Zealand health system indicators. Ministry of Health; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health New Zealand. Data protection and health information governance. Ministry of Health; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNational Population Commission (NPC). [Nigeria] \u0026amp; ICF. (2019). \u003cem\u003eNigeria Demographic and Health Survey 2018\u003c/em\u003e. NPC and ICF.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOECD. Health at a glance 2022: OECD indicators. OECD Publishing. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1787/ae3016b9-en\u003c/span\u003e\u003cspan address=\"10.1787/ae3016b9-en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSinger JD, Willett JB. Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press; 2003.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNICEF. Multiple Indicator Cluster Surveys (MICS): Methodological papers. UNICEF; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNICEF. The state of the world\u0026rsquo;s children 2023: For every child, vaccination. UNICEF; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUnited Nations. Transforming our world: The 2030 agenda for sustainable development. United Nations; 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVictora CG, Adair L, Fall C, Hallal PC, Martorell R, Richter L, Sachdev HS. Maternal and child undernutrition: Consequences for adult health and human capital. Lancet. 2016;371(9609):340\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0140-6736(07)61692-4\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(07)61692-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWooldridge JM. Econometric analysis of cross section and panel data. 2nd ed. MIT Press; 2010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Bank. World development indicators. World Bank; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. Ethics and governance of artificial intelligence for health. WHO; 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. Levels and trends in child mortality. WHO; 2022.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Artificial Intelligence Geographic Information Systems (AI–GIS), Child Health Inequality, Spatiotemporal Analysis, Comparative Health Systems, Policy Effectiveness","lastPublishedDoi":"10.21203/rs.3.rs-8443273/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8443273/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eChild health is a critical indicator of human development and health system performance. Despite global progress, inequities in child health outcomes remain stark, particularly between high‑income and low‑middle‑income countries. This study applies an integrated Artificial Intelligence Geographic Information Systems (AI\u0026ndash;GIS) framework to compare child health trajectories in New Zealand and Nigeria from 2006 to 2022, focusing on spatial inequality, temporal change, and policy effectiveness. Using harmonized national datasets, we examine immunization coverage, sanitation access, and treatment of childhood illnesses in relation to maternal education, household wealth, and place of residence. Results reveal two contrasting inequality regimes: in Nigeria, child health outcomes are shaped by structural inequities linked to governance capacity and infrastructure distribution, while in New Zealand, residual disparities persist within an otherwise mature health system. By distinguishing structural from residual inequality, the study highlights how policy design must adapt to context whether addressing entrenched deprivation in resource‑constrained settings or fine‑tuning equity in advanced systems. Findings underscore the importance of equity‑driven health policy and provide evidence for strengthening child health strategies in both developing and developed contexts.\u003c/p\u003e","manuscriptTitle":"A Comparative AI–GIS Spatiotemporal Analysis of Child Health Outcomes in New Zealand and Nigeria (2006–2022): Implications for Equity-Driven Health Policy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-08 14:03:14","doi":"10.21203/rs.3.rs-8443273/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"be89d42b-0f17-4478-94ca-040e20bf71c7","owner":[],"postedDate":"January 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-18T14:39:38+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-08 14:03:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8443273","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8443273","identity":"rs-8443273","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}