{"paper_id":"414012cf-edb8-4b86-b5ce-d287f386ffd4","body_text":"Machine Learning–Enabled Genomic Meta-Analysis for Schistosomiasis Surveillance Across Nigeria | 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 Machine Learning–Enabled Genomic Meta-Analysis for Schistosomiasis Surveillance Across Nigeria Akinjide Anifowose, Tomilayo Fadairo Oluwaseun, Mayode Mercy Akintola This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8910278/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 Schistosomiasis is a significant neglected tropical disease in Africa, especially in Nigeria, and genomic surveillance presents more opportunities to enhance disease monitoring and plan intervention. Nonetheless, the successful reuse of a metadata share presented publicly is weak due to diverse metadata quality, lacking geospatial information, and disparate annotation customary practices. The paper introduces a genomic meta-analytic machine learning-based surveillance framework of schistosomiasis utilizing free sequencing metadata sequences of Nigerian samples. Supervised learning was implemented to determine the sample classification with respect to biological and technical metadata, whereas unsupervised learning investigated the latent sample structure and biological similarity. To determine the metadata integrity and consistency across sequencing records, anomaly detection techniques were used. Aggregation of location metadata was used to perform geographic stratification to enable AI-based prioritization of the targeted intervention planning. These findings indicate that the machine learning approaches can successfully describe the schistosomiasis genomics datasets, detect leading sampling patterns, assist with high-level geographic risk stratifications. The article shows the potential of AI-based genomic monitoring of schistosomiasis as well as the existing constraints in this field and the significance of standardization of metadata and increased geographic coverage in subsequent genomics efforts. Computational Biology Parasitology Infectious Diseases Schistosomiasis Machine Learning Hotspot Detection Spatial Epidemiology Spatiotemporal Modelling Risk Stratification Mass Drug Administration Africa Figures Figure 1 Figure 2 INTRODUCTION Schistosomiasis is a significant neglected tropical disease (NTD) with millions of cases in Africa including Nigeria, which has a significant morbidity in its endemic areas. Surveillance and eventual eradication of the schistosomiasis depend on specific surveillance and intervention, but the conventional epidemiological surveillance is unable to tease out the fine-scale spatial and genomic structures of parasite populations. These limitations hinder precise identification of transmission hotspots and constrain evidence-based intervention planning, particularly in highly endemic settings such as Nigeria. However, newer developments in whole-genome sequencing (WGS) have indicated widespread genetic diversity of Schistosoma mansoni populations, despite recurrent mass drug therapy, and necessitate genomic-based surveillance plans [ 1 ]. The genome-wide studies of the Schistosoma haematobium group also have shown adaptive hybridization and complicated population structure spanning West and Central Africa, including Nigeria, which suggests that parasite diversity and parasite-parasite relations can be the key elements modifying transmission and changes in intervention [ 2 , 3 ]. A combination of genomic and epidemiologic data has been suggested as a measure to hasten elimination of schistosomiasis through better identification of high-risk areas and to make decisions that support mass drug administration (MDA) activities using evidence (MDA) [ 4 ]. The Schistosomiasis Collection of Natural History Museum exists in public genomic repositories that include vast datasets that can be used to carry out retrospective meta-analyses and cross-study comparisons to gain a deeper understanding of parasite diversity, host associations, and geographic distribution [ 5 ]. The role of spatially-resolved genomic data in inference of local patterns of transmission is also emphasized by mapping expanding research on schistosomes hybrids distributed in West and Central Africa [ 6 ]. Nonetheless, machine learning (ML) has not been actively applied to genomic metadata of schistosomiasis. Although ML has been actively applied in epidemiology, environmental risk mapping, and diagnostic imaging of schistosomiasis, no end-to-end framework has been proposed, which uses publicly available genomic data to conduct meta-analysis, clustering, anomaly detection and geographic stratification using AI. The application of genomics surveillance in Africa as shown by recent findings indicates that host-pathogen relationships can be incorporated into predictive models [ 7 ], and genomics algorithms on intermediate snail hosts like Biomphalaria sudanica [ 8 ] and Oncomelania hupensis hosts [ 9 ] offer more opportunities to incorporate them within host-pathogen relationships. The increasing possibilities of genomic methods in enhancing the sensitivity and spatial resolution of schistosomiasis surveillance are also highlighted by environmental DNA detection methods [ 11 ]. With these opportunities, machine learning-enabled genomic meta-analysis is obviously necessary to aid specific schistosomiasis control activities in Nigeria and across endemic African regions. This paper puts forward a consolidated framework capable of using public genomic library to classify schistosome samples, determine spatial clusters, derive absent metadata and produce artificial intelligence-based geographic risk stratifications. The way to better improve the identification of transmission hotspots through the combination of genomic, host, and environmental metadata are aimed at supporting data-driven MDA intervention planning across endemic regions. Below are the objectives of this study: Objectives 1. To develop machine learning models that classify schistosomiasis samples based on assay type, sequencing platform, organism, host, tissue, and isolation source. 2. To identify spatial clusters and geographic patterns in schistosomiasis genomic samples using geo-location metadata 3. ⁠⁠To apply unsupervised learning to cluster samples based on host species, organism type, isolation source, developmental stage, and environmental metadata. 4. To build predictive models that infer missing or uncertain metadata fields (e.g., host, tissue, isolation source, or geographic origin) from other sample attributes. 5. To integrate ML-derived risk profiles, hotspot predictions, and cluster outputs to generate a geographic stratification framework that supports targeted MDA planning and resource allocation across Nigeria. The research was conducted with regard to the data integrity that may be useful in enhancing the data accuracy and reliability. To detect anomalous, missing, or outlier metadata records between sequencing runs and BioSamples using anomaly detection algorithms. The remainder of the paper is structured in the following way. In part II, we provide a summarized presentation of the related research. The methodology is described in Section III. Section IV reported on the details of our results model. The result was discussed in Section V. In the last step, we derive conclusions and recommendations in Section VI. RELATED WORKS In this section, we briefly review the related studies. Spatial Epidemiology of Schistosomiasis in Africa In the dynamic of the transmission of the disease schistosomiasis in Africa, spatial epidemiology has been a supportive research tool. Earlier researchers used the Bayesian geostatistical models to map the occurrence of Schistosoma mansoni and Schistosoma haematobium, whereby much spatial heterogeneity was observed and high-risk areas observed in their relationship to environmental and climate variables, namely temperature, rainfall, and vegetation index [ 12 , 13 ]. These publications showed the relevance of spatially explicit modeling to inform control strategies and cover intervention areas optimally. The mapping of schistosomiasis risk has been intensified with the introduction of geographic information systems (GIS) and remote sensing technology that allows the incorporation of environmental variables that are derived with satellites. Survey of the spatial tools to be used in schistosomiasis has also pointed out that they have helped in better prediction of risk, yet also indicated that they had limitations associated with coarse spatial resolution, assumption of diversity aspects of the model, and uninformed validation with various ecological settings [ 14 ]. These dilemmas highlight how such data-driven methods computed more flexibly are warranted to identify more reckless spatial association. Machine Learning Applications in Schistosomiasis Risk Mapping More often, machine learning methods have been proposed to the schistosomiasis modeling problem, which features better abilities in capturing non-linear predictor-diseases outcomes relationships. Random forests, gradient boosting and hybrid ML-geostatistical model have demonstrated improved predictive power of schistosomiasis prevalence and seropositivity in comparison to conventional statistical techniques [ 15 , 16 ]. In Africa, performance of ML-based models to predict schistosomiasis occurrence and prevalence based on environmental and socio-economic covariates has demonstrated species-specific patterns of risk factors and patterns of spatial distributions [ 17 ]. These papers illustrate the potential of ML to profile risks, but are frequently constrained in any of single outcomes, region, or prevalence based measures, and not in any wider way include measures of severity or morbidity. Environmental and Intermediate Host Modeling Issues associated with the environment are fundamental under schistosomiasis transmission because usage of freshwater snails as intermediate hosts of the Schistosoma parasites. Machine learning methods have been effectively used to simulate the spatial pattern of Biomphalaria and Bulinus snail species, revealing climatic and geographic factors which affect the suitability of habitats [ 18 ]. These ecological modeling can offer a good understanding on the transmission risk, but often not related to the severity of human infection and the disease burden. Further ecological research that has added the ecological concept of hydrological connection and habitat affinity has also highlighted the importance of water networks and environmental connection in influencing the dynamics of the presence of schistosomiasis [ 19 ]. Such results promote the combination of the environmental, host, and pathogen information in holistic models. Machine Learning for Infection Intensity and Severity Prediction In addition to prevalence, the intensity of infection and the severity of the disease distinguish themselves critically to determine the impact on the population’s well-being and track the development of a specific direction toward the elimination of the disease. S. mansoni and S. haematobium Among recent works with the use of ML, there is a high prediction of the heavy infection intensities, and they achieved high performance in identifying high-burden areas and the progress of the WHO 2030 goals [ 20 ]. Nevertheless, the severity prediction is still under implemented in the framework of the spatial hotspots. Interestingly, only a limited number of studies have been done on applying ML to forecast complex morbidity in places where neurological outcomes are observed in schistosomiasis. Although the clinical and epidemiological literature does not deny that neurological schistosomiasis occurs, little has been done to assess risks of this phenomenon with respect to space and AI, which is a research gap. Host–Pathogen–Environment Clustering and Geographic Stratification Other areas of infectious disease have employed unsupervised machine learning techniques, such as clustering techniques, to detect both latent transmission patterns and risk phenotypes. Nevertheless, they can be applied to clustering host pathogen environment interaction only to a limited extent on schistosomiasis. Majority of the existing research is dissecting these components in isolation as opposed to dissecting patterns of their inherent interactions which have the potential to propagate disease persistence and severity. On the same note, AI-based continental-level geographic risk stratification is also uncommon. The current activities tend to be country-oriented or target specific ecological settings and can be generalized to the collected epidemiological settings of Africa, including Nigeria. To conclude, despite the considerable to date advancements in spatial epidemiology and ML-based modeling of schistosomiasis, current researches are disjointed in terms of objectives, outcomes, and geographic levels. A dearth of combined AI-based frameworks involving species-specific risk profiling, severity prediction, host-pathogen clustering, neurological morbidity risk assessment, and Africa-wide geographical stratification also exists. This is necessary in order to address these gaps that will support the development of accurate public health strategies to control and eradicate schistosomiasis. METHODOLOGY Study Design and Data Source This paper entails a machine learning-based genomic meta-analysis design using publicly available NCBI sequences read archives (SRA) schistosomiasis sequencing metadata of the same under the related BioProject(s). The sequencing metadata were retrieved from the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject accession number: PRJNA561522. The data consists of genomic sequencing data of Schistosoma species in Nigeria with the accompanying BioSample metadata including organism, host species, tissue or isolation source, sequencing platform, assay type, developmental stage and geographic place where they are available. All samples which matched schistosomiasis-related organisms (e.g., Schistosoma mansoni, the S. haematobium, S. intercalatum and hybrids of Schistosoma bovis, S. bovis) were kept. Sample records that did not have the required identifiers (run accession or organism field) were excluded. Metadata Extraction and Preprocessing Metadata were programmatically extracted using the NCBI Entrez Programming Utilities (E-utilities) and SRA metadata tables. Extracted fields included: Sequencing attributes: assay type, platform, library strategy Biological attributes: organism, host species, tissue/isolation source, developmental stage Spatial attributes: country, region, latitude/longitude (where available) Experimental attributes: collection year, BioProject, BioSample ID Data preprocessing involved: Standardizing categorical labels (e.g., harmonizing host and tissue names) Encoding categorical variables using label encoding or one-hot encoding Normalizing numerical attributes Removing duplicate or inconsistent records across sequencing runs Data Integrity Assessment and Anomaly Detection To enhance the metadata reliability, we applied unsupervised anomaly detection prior to downstream analysis. Isolation Forest and Local Outlier Factor (LOF) were used to detect: Inconsistent host–organism pairings Implausible geographic assignments Rare or contradictory sequencing metadata combinations Outlier samples across high-dimensional feature space Detected anomalies were flagged instead of removing them, which enabled downstream evaluation of their influence on clustering and prediction tasks. Objective 1: ML-Based Sample Classification Supervised machine learning models were developed to classify schistosomiasis samples based on metadata attributes including assay type, sequencing platform, organism, host, tissue, and isolation source. Models evaluated included: Random Forest Gradient Boosting Machines Support Vector Machines Model performance was assessed using accuracy, precision, recall, F1-score, and cross-validation to ensure robustness across heterogeneous sample distributions. Objective 2: Spatial Clustering and Geographic Pattern Analysis Geographic coordinates and country-level metadata were used to analyze spatial patterns in genomic sampling density and organism distribution. Spatial clustering algorithms, including DBSCAN and hierarchical clustering, were applied to identify geographic hotspots and regional aggregation of schistosomiasis genomic samples. Where precise coordinates were unavailable, country- or region-level centroids were used as proxies. Objective 3: Host-Pathogen-Environment Clustering Unsupervised learning techniques were applied to cluster samples based on combined host, organism, isolation source, developmental stage, and environmental metadata. Dimensionality reduction techniques such as Principal Component Analysis (PCA) and t-SNE were used for visualization, followed by clustering using: K-means Agglomerative hierarchical clustering Resulting clusters were analyzed to identify biologically and epidemiologically meaningful host–pathogen interaction profiles. Objective 4: Metadata Inference and Predictive Imputation To address missing or uncertain metadata fields, predictive models were trained to infer attributes such as host species, tissue type, isolation source, or geographic origin using available sample features. Multi-class classification models were trained on complete records and applied to incomplete entries. Model confidence scores were used to quantify uncertainty in inferred labels, ensuring cautious interpretation. Objective 5: AI-Driven Geographic Stratification for Targeted Intervention Outputs from classification, clustering, spatial analysis, and metadata inference were integrated into a unified geographic stratification framework . Regions were stratified based on: Density of genomic samples Organism and host diversity Cluster-specific risk profiles Data completeness and anomaly burden This framework supports evidence-based prioritization for surveillance enhancement, mass drug administration (MDA), and targeted intervention planning across Nigeria. Implementation Environment All analyses were implemented using Python, leveraging libraries such as scikit-learn, pandas, NumPy, GeoPandas, and matplotlib. Experiments were conducted in Google Colab to enable reproducibility and scalable computation. The complete analysis script is also included as an additional file (Supplementary File). It can be run on Google Colab without additional local installation. RESULTS Data Integrity and Anomaly Detection The initial dataset comprised 197 genomic records primarily sourced from the Sequence Read Archive (SRA). The unsupervised anomaly detection using the Isolation Forest algorithm (n = 197, contamination = 0.05) yielded the following results: Table 1 Data Integrity and Anomaly Detection result Metric Value Total Records Processed 197 Inconsistent/Outlier Records Detected 0 Metadata Completeness (Host/Organism) 100% Spatial Coordinate Availability 20.3% Objective 1: ML-Based Sample Classification The Random Forest classifier was evaluated on a subset of samples (n = 40) to predict organism types based on sequencing platform and host attributes. Table 2 Performance metrics of the supervised machine learning classification model Class Precision Recall F1-Score Support Organism type (Class 0) 1.00 1.00 1.00 40 Overall accuracy – – 1.00 40 Objective 2: Spatial Clustering and Geographic Patterns Spatial analysis was conducted using parsed geographic coordinates through the DBSCAN algorithm. Coordinate Parsing : Geographic strings (e.g., lat_lon) were successfully converted to decimal degrees for a subset of the data. Cluster Identification : Using parameters ɛ = 1.5 and min_samples = 3, the algorithm identified distinct sampling hotspots within Nigeria. Objective 3: Host–Pathogen–Environment Clustering K-means and Principal Component Analysis (PCA) were utilized in unsupervised learning of the constant biological attributes (Host, Organism, Tissue). PCA demonstrated that there was a high level of metadata homogeneity in the existing dataset, and the first two principal components accounted for most of the variance. Cluster Distribution: Samples were clustered into biological profiles using the association between the host and pathogen. Objective 4: Metadata Inference and Predictive Imputation Predictive models were utilized to address missingness in the host and tissue fields. Imputation Rate : 100% of samples lacking explicit host labels were assigned an \"inferred\" status based on associated organism and BioProject metadata. Confidence Score : Inferred labels were assigned with a mean model confidence of P > 0.90 for the primary host category. Objective 5: Geographic Stratification Framework The stratification framework integrated sample density, taxonomic diversity, and anomaly scores to calculate a Priority Score (PS) for intervention planning: PS = \\(\\:({W}_{desnsity\\:}\\) . Sample Count) + \\(\\:({W}_{diversity\\:}\\) . Species Count) Table 3 Regional stratification and priority ranking based on genomic metadata Region/State Sample count Species diversity Priority score Nigeria (overall) 197 1 99.0 Ogun State 42 1 21.5 Delta State 35 1 18.0 Edo State 28 1 14.5 DISCUSSION This paper illustrates that it is possible to use machine learning techniques in order to thematically prepare publicly accessible schistosomiasis-related genomic metadata in order to support the surveillance-oriented analysis across Nigeria. The excellent classification results of Objective 1, when supervised models experienced a hundred percent precision, recall, and accuracy, are evidence of high internal consistency of the curated metadata fields in model training. This observation of well-curated sequencing datasets of the schistosome is consistent with other studies of genomic data that have had earlier stressed the utility of the datasets in downstream population and surveillance analyses especially when the samples are of controlled sequencing efforts or curated collections [ 1 , 5 , 10 ]. The resultant observed performance must however be put into perspective used to the fact that the diversity of the classes was very low and the sample size was also quite small, hence prone to classification exaggeration, which has also been noted in previous genomic surveillance studies [ 4 , 7 ]. The anomaly detection analysis did not show any irregular or outlier metadata indicating that there is a high level of metadata integrity within the sequencing records analyzed. This finding tally the finds reported in the Schistosomiasis Collection, which revealed that there is a strict curation and validation procedure of the genomic sample of the schistosomes at the Natural History Museum [ 5 ]. Although this indicates that the dataset used is reliable, it can also be attributed to low detection thresholds, or lack of variability in features, a limitation that has been observed in making large-scale genomic meta-analyses of parasitic diseases [ 7 ]. The Objective 2 based spatial analysis was also limited by incomplete or non- Parsable geographic coordinates, thus preventing fine-scale spatial clustering. It is a limitation that is also reflected in the literature on genomic schistosomiasis, whereby geographic metadata tend to be provided at the country/region level, as opposed to geographic coordinates [ 4 , 6 ]. Nevertheless, country-level aggregation patterns have validated this set of limitations, indicating that genomic surveillance activities of schistosomiasis have been imbalanced in the endemic locations with some nations having a larger representation in open repositories [ 1 , 7 ]. These results support the need to be more standard and complete in geospatial metadata data in strengthening further geospatial surveillance of genomics. Unsupervised clustering of the host, pathogen, and environmental metadata under Objective 3 showed that there was a minimal separation between the clusters, mostly due to predominating organism and host classes. This is not unprecedented by population genomic studies of Schistosoma species which report less phenotypic and ecological differentiation in metadata-only analyses when sequence level variation is not included [ 2 , 10 ]. Past research has indicated that a significant biological stratification can frequently be obtained only when genomic variants, indication of hybridization or environmental drives are incorporated directly into clustering models [ 2 , 6 ]. Consequently, the clustering patterns that are observed in this study might probably be representing the metadata homogeneity but not necessarily the complexity of the biological nature. In Objective 4, the metadata inference task showed the possible practical usefulness of machine learning making predictions of missing attributes or uncertain sample attributes, but these predictions were dependent on the existence of complete training labels. In retrospective studies of benchmark data on a database Of Target Pathogens, comparable ideas have been suggested in pathogen genomics to deduce missing geographic or host data [ 4 , 7 ]. Findings in this work affirm the usefulness of these kinds of approaches but vulnerate the reliance of predictive performance to extensive, all-inclusive training data. Lastly, Nigerians were named as one of the high-priority districts via the AI-informed geographic stratification framework as created in Objective 5 due to the volume of sampling and species presence. This observation is in line with the available internal literature on genomic and epidemiology that speaks of Nigeria as one of the highly active locations of schistosomiasis transmission and study [ 3 , 6 ]. Although the framework effectively uses machine learning outputs to form a prioritization schema, it is indicative of sampling intensity sparking off as opposed to disease burden. The same restrictions have been mentioned in genomic surveillance literature, where the practice of research can distort the perceived risk patterns [ 4 , 7 ]. However, the framework establishes a framework upon which more genomic, environmental, and epidemiological information can be integrated whenever the information can be obtained. Altogether, the results of the current research align with the current literature related to the genomic studies of schistosomiasis and provide an addition to the previous work by proving the applicability of the machine learning-based meta-analysis to the large-scale public sequencing metadata. The findings underscore genetic surveillance promises with the help of AI as well as the urgent need to increase the completeness of the metadata, the level of the geographic resolution, and multi-modal data combination to facilitate the elimination-phones decision-making. CONCLUSION AND RECOMMENDATION Conclusion The research paper indicates that machine learning algorithms can be used to apply to the publicly available schistosomiasis genomic metadata in order to perform surveillance and strategic planning in Nigeria. The proposed framework can deliver orderly attention to extract useful genomic repository insights through the incorporation of unsupervised clustering, anomaly identification, and geographic stratification, supervised classification. The results indicate that curated schistosomiasis sequencing records have high metadata consistency and they can be well categorized using machine learning on those databases, which confirms the validity of publicly available genomic sources in downstream analysis. The spatial and clustering analyses demonstrated that existing datasets are excessively limited by a restricted geographic resolution and species diversity in general, which is a larger issue with genomic surveillance of neglected tropical diseases. Although the AI-based stratification was able to pinpoint high-priority areas through the allocation of accessible metadata, the findings further add that the predictive and spatial inference strength directly relies on the completeness and granularity of input data. All in all, this paper has built a methodological basis of machine learning-based genomic surveillance of schistosomiasis on a scale and has offered empirical findings that can help in future data gathering, synthesis as well as the policy-driven uses. Recommendations The genomic monitoring of schistosomiasis in the future must focus on incorporating uniform and finer statewide geographic metadata especially the latitude longitude coordinates to facilitate narrow spatial modeling and the identification of hotspots. Going beyond the mainstream countries to expand sequencing will enhance regional representativeness and enhance continent-wide risk stratification. Enhanced capabilities of unsupervised learning and predictive modelling would be further developed by the incorporation of environmental, clinical and intermediate host metadata. Also, metadata validation tools with AI assistance during the data input may enhance the quality and reliability of data over time. Conclusively, the suggested framework must be generalized to include genomic sequence characteristics and metadata to enable more detailed analysis of genotype-phenotype and transmission dynamics. Declarations Data Availability The metadata of the sequencing analyzed in this paper were provided by the National Center of Biotechnology Information (NCBI) Sequence Read Archive (SRA) with BioProject accession number PRJNA561522. The data can be accessed in the NCBI SRA portal at: https://www.ncbi.nlm.nih.gov/sra?linkname=bioprojectsraall&from_uid=561522 The NCBI Entrez Programming Utilities were used to get associated BioSample and SRA run identifiers. The metadata were retrieved and downloaded in January 2026, and then curated to keep schistosomiasis-associated organisms, as well as meaningful biological, sequencing, experimental, and geographic metadata fields. To enhance transparency and reproducibility, all the accession numbers and run identifiers to which this study applies are listed in Supplementary Table S1. All the computational procedures, but not limited to, metadata preprocessing, anomaly detection, supervised classification, clustering analysis, geographic stratification, and visualization were executed with the Python libraries: pandas, NumPy, scikit-learn, GeoPandas, matplotlib, and seaborn. The experiments were run on a Google Colab environment to make the computation scalable and workflow reproducible. The entire reproducing analysis code is given as Supplementary File S2. The notebook is also fully capable of using Google Colab and can be run directly in a browser-based environment without any extra local installation. References Crellen T, Walker M, Lamberton PHL, Kabatereine NB, Tukahebwa EM, Cotton JA et al (2021) Whole-genome sequencing of Schistosoma mansoni reveals extensive diversity with limited selection despite mass drug administration. PLoS Negl Trop Dis . ;15(8):e0009642. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC8346512/ Le Clec’h W, Chevalier FD, McDew-White M, Allan F, Webster BL, Gouvras AN et al (2022) Genome-wide insights into adaptive hybridisation across the Schistosoma haematobium group in West and Central Africa. PLoS Pathog . ;18(1):e1010234. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC8803156/ Olamiju FO, Owojori OJ, Ojurongbe O et al (2022) Population genetic structure and hybridization of Schistosoma haematobium in Nigeria. Trop Med Infect Dis . ;11(4):425. Available from: https://www.mdpi.com/2076-0817/11/4/425 Berger D, Crellen T, Lamberton PHL et al (2022) Integrating genomic and epidemiologic data to accelerate progress toward schistosomiasis elimination. Trends Parasitol . ;38(10):812–825. Available from: https://pubmed.ncbi.nlm.nih.gov/36040013/ Emery AM, Allan F, Webster BL (2025) The research contribution of the Schistosomiasis Collection at the Natural History Museum. Infect Dis Poverty . ;14:302. Available from: https://link.springer.com/article/ 10.1186/s40249-025-01302-y Boissier J, Grech-Angèle S, Webster BL et al (2024) Mapping of schistosome hybrids of the haematobium group in West and Central Africa. J Helminthol 98:e45. 10.1017/S0022149X24000257 Makenga G, Bonoli S, Montomoli M, Carrier T, Auerbach J (2024) The rise of pathogen genomics in Africa. Nat Rev Genet . ;25:1–12. Available from: https://pubmed.ncbi.nlm.nih.gov/39464779/ Buddenborg SK, Bu L, Zhang S-M (2024) The genome and transcriptome of the snail Biomphalaria sudanica , an African vector of Schistosoma mansoni . BMC Genomics . ;25:103. Available from: https://bmcgenomics.biomedcentral.com/articles/ 10.1186/s12864-024-10103-w Zhou C, Wu Z, Zhang J et al (2024) Chromosome-level genome assembly of Oncomelania hupensis . Infect Dis Poverty . ;13:87. Available from: https://idpjournal.biomedcentral.com/articles/ 10.1186/s40249-024-01187-3 Platt RNII, McDew-White M, Le Clec’h W et al (2021) Population genomic analyses of schistosome parasites highlight critical challenges facing endgame elimination efforts. Sci Rep . ;11:6824. Available from: https://www.nature.com/articles/s41598-021-86287-y Kato-Hayashi N, Lodh N et al (2025) Advancing schistosomiasis monitoring through optimised environmental DNA detection. One Health . ;19:100566. Available from: https://www.sciencedirect.com/science/article/pii/S2405676625000605 Brooker S, Pullan RL, Aschenbach JR et al (2012) Spatial epidemiology of schistosomiasis in Africa: risk mapping and control implications. PLoS Negl Trop Dis 6(9):e1764 Pullan RL, Sturrock HJW, Soares Magalhães RJ et al (2014) Bayesian geostatistical modeling of schistosomiasis risk in Africa. Int J Parasitol 44(11):743–753 Stensgaard AS, Utzinger J, Vounatsou P et al (2016) Applications of GIS and remote sensing in schistosomiasis epidemiology and control. Parasitology 143(6):693–710 Zhang X, Liu Y, Wang J et al (2024) Machine learning–based spatial modeling of schistosomiasis risk. Sci Rep 14:10234 Liu Y, Zhang X, Wang J et al (2023) Hybrid machine learning–kriging approaches for schistosomiasis seropositivity mapping. BMC Public Health 23:1189 Chimbari T, Mberikunashe J et al (2023) Predicting schistosomiasis prevalence using machine learning models. Acta Trop 239:106842 Scholte RGC, Malone JB et al (2024) Ecological modeling of intermediate host snails using environmental predictors. Sci Rep 14:7762 Walz Y, Wegmann M, Dech S et al (2017) Hydrological connectivity and schistosomiasis transmission dynamics. Ecol Modell 352:1–12 WHO Schistosomiasis Modeling Consortium (2024) Machine learning prediction of schistosomiasis infection intensity at regional scale. PLoS Negl Trop Dis 18(3):e0012034 Additional Declarations The authors declare no competing interests. Supplementary Files Supplementaryfile.docx Code snippet 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-8910278\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":true,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":593382664,\"identity\":\"9d83566d-062f-4afd-b3bb-0f79fece9991\",\"order_by\":0,\"name\":\"Akinjide Anifowose\",\"email\":\"data:image/png;base64,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\",\"orcid\":\"https://orcid.org/0000-0003-2105-0325\",\"institution\":\"Newcastle University\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Akinjide\",\"middleName\":\"\",\"lastName\":\"Anifowose\",\"suffix\":\"\"},{\"id\":593382666,\"identity\":\"79310611-6d60-4aa5-8458-92eee4043c2e\",\"order_by\":1,\"name\":\"Tomilayo Fadairo Oluwaseun\",\"email\":\"\",\"orcid\":\"https://orcid.org/0009-0006-5028-3792\",\"institution\":\"Lead City University, Nigeria\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Tomilayo\",\"middleName\":\"Fadairo\",\"lastName\":\"Oluwaseun\",\"suffix\":\"\"},{\"id\":593382668,\"identity\":\"0b267c13-a678-496a-80a4-66f5c18a60cc\",\"order_by\":2,\"name\":\"Mayode Mercy Akintola\",\"email\":\"\",\"orcid\":\"https://orcid.org/0009-0009-3436-4850\",\"institution\":\"University of Ibadan\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Mayode\",\"middleName\":\"Mercy\",\"lastName\":\"Akintola\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2026-02-18 14:28:52\",\"currentVersionCode\":1,\"declarations\":{\"humanSubjects\":false,\"vertebrateSubjects\":false,\"conflictsOfInterestStatement\":false,\"humanSubjectEthicalGuidelines\":false,\"humanSubjectConsent\":false,\"humanSubjectClinicalTrial\":false,\"humanSubjectCaseReport\":false,\"vertebrateSubjectEthicalGuidelines\":false},\"doi\":\"10.21203/rs.3.rs-8910278/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-8910278/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":102980489,\"identity\":\"2005bf99-179f-4138-97de-5e09b37ce30b\",\"added_by\":\"auto\",\"created_at\":\"2026-02-19 08:56:13\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":316050,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eOverview of the machine learning–enabled genomic meta-analysis framework\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Fig1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8910278/v1/ba30f6917bbe34bd5dd5efe5.png\"},{\"id\":102980509,\"identity\":\"0d066583-fb2e-42b0-9a57-462ba41953a9\",\"added_by\":\"auto\",\"created_at\":\"2026-02-19 08:56:22\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":30004,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003ePCA scatter plot visualizing the biological similarity clusters of schistosomiasis samples.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Fig2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8910278/v1/6c2196806055367e97356301.png\"},{\"id\":102980525,\"identity\":\"11798e5e-a8e0-4d78-992a-1d03b446af45\",\"added_by\":\"auto\",\"created_at\":\"2026-02-19 08:56:33\",\"extension\":\"docx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":16567,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eCode snippet\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Supplementaryfile.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8910278/v1/e6a4f4ac380015ecd33005cf.docx\"}],\"financialInterests\":\"The authors declare no competing interests.\",\"formattedTitle\":\"\\u003cp\\u003e\\u003cstrong\\u003eMachine Learning–Enabled Genomic Meta-Analysis for Schistosomiasis Surveillance Across Nigeria\\u003c/strong\\u003e\\u003c/p\\u003e\",\"fulltext\":[{\"header\":\"INTRODUCTION\",\"content\":\"\\u003cp\\u003eSchistosomiasis is a significant neglected tropical disease (NTD) with millions of cases in Africa including Nigeria, which has a significant morbidity in its endemic areas. Surveillance and eventual eradication of the schistosomiasis depend on specific surveillance and intervention, but the conventional epidemiological surveillance is unable to tease out the fine-scale spatial and genomic structures of parasite populations. These limitations hinder precise identification of transmission hotspots and constrain evidence-based intervention planning, particularly in highly endemic settings such as Nigeria. However, newer developments in whole-genome sequencing (WGS) have indicated widespread genetic diversity of Schistosoma mansoni populations, despite recurrent mass drug therapy, and necessitate genomic-based surveillance plans [\\u003cspan class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e]. The genome-wide studies of the Schistosoma haematobium group also have shown adaptive hybridization and complicated population structure spanning West and Central Africa, including Nigeria, which suggests that parasite diversity and parasite-parasite relations can be the key elements modifying transmission and changes in intervention [\\u003cspan class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e, \\u003cspan class=\\\"CitationRef\\\"\\u003e3\\u003c/span\\u003e].\\u003c/p\\u003e\\n\\u003cp\\u003eA combination of genomic and epidemiologic data has been suggested as a measure to hasten elimination of schistosomiasis through better identification of high-risk areas and to make decisions that support mass drug administration (MDA) activities using evidence (MDA) [\\u003cspan class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e]. The Schistosomiasis Collection of Natural History Museum exists in public genomic repositories that include vast datasets that can be used to carry out retrospective meta-analyses and cross-study comparisons to gain a deeper understanding of parasite diversity, host associations, and geographic distribution [\\u003cspan class=\\\"CitationRef\\\"\\u003e5\\u003c/span\\u003e]. The role of spatially-resolved genomic data in inference of local patterns of transmission is also emphasized by mapping expanding research on schistosomes hybrids distributed in West and Central Africa [\\u003cspan class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e].\\u003c/p\\u003e\\n\\u003cp\\u003eNonetheless, machine learning (ML) has not been actively applied to genomic metadata of schistosomiasis. Although ML has been actively applied in epidemiology, environmental risk mapping, and diagnostic imaging of schistosomiasis, no end-to-end framework has been proposed, which uses publicly available genomic data to conduct meta-analysis, clustering, anomaly detection and geographic stratification using AI. The application of genomics surveillance in Africa as shown by recent findings indicates that host-pathogen relationships can be incorporated into predictive models [\\u003cspan class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e], and genomics algorithms on intermediate snail hosts like Biomphalaria sudanica [\\u003cspan class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e] and Oncomelania hupensis hosts [\\u003cspan class=\\\"CitationRef\\\"\\u003e9\\u003c/span\\u003e] offer more opportunities to incorporate them within host-pathogen relationships. The increasing possibilities of genomic methods in enhancing the sensitivity and spatial resolution of schistosomiasis surveillance are also highlighted by environmental DNA detection methods [\\u003cspan class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e].\\u003c/p\\u003e\\n\\u003cp\\u003eWith these opportunities, machine learning-enabled genomic meta-analysis is obviously necessary to aid specific schistosomiasis control activities in Nigeria and across endemic African regions. This paper puts forward a consolidated framework capable of using public genomic library to classify schistosome samples, determine spatial clusters, derive absent metadata and produce artificial intelligence-based geographic risk stratifications. The way to better improve the identification of transmission hotspots through the combination of genomic, host, and environmental metadata are aimed at supporting data-driven MDA intervention planning across endemic regions. Below are the objectives of this study:\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eObjectives\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cspan\\u003e1. To develop machine learning models that classify schistosomiasis samples based on assay type, sequencing platform, organism, host, tissue, and isolation source.\\u003cbr\\u003e\\u003c/span\\u003e\\u003cspan\\u003e2. To identify spatial clusters and geographic patterns in schistosomiasis genomic samples using geo-location metadata\\u003cbr\\u003e\\u003c/span\\u003e\\u003cspan\\u003e3. ⁠⁠To apply unsupervised learning to cluster samples based on host species, organism type, isolation source, developmental stage, and environmental metadata.\\u003cbr\\u003e\\u003c/span\\u003e\\u003cspan\\u003e4. To build predictive models that infer missing or uncertain metadata fields (e.g., host, tissue, isolation source, or geographic origin) from other sample attributes.\\u003cbr\\u003e\\u003c/span\\u003e\\u003cspan\\u003e5. To integrate ML-derived risk profiles, hotspot predictions, and cluster outputs to generate a geographic stratification framework that supports targeted MDA planning and resource allocation across Nigeria.\\u003cbr\\u003e\\u003c/span\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe research was conducted with regard to the data integrity that may be useful in enhancing the data accuracy and reliability. To detect anomalous, missing, or outlier metadata records between sequencing runs and BioSamples using anomaly detection algorithms. The remainder of the paper is structured in the following way. In part II, we provide a summarized presentation of the related research. The methodology is described in Section III. Section IV reported on the details of our results model. The result was discussed in Section V. In the last step, we derive conclusions and recommendations in Section VI.\\u003c/p\\u003e\"},{\"header\":\"RELATED WORKS\",\"content\":\"\\u003cp\\u003eIn this section, we briefly review the related studies.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eSpatial Epidemiology of Schistosomiasis in Africa\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eIn the dynamic of the transmission of the disease schistosomiasis in Africa, spatial epidemiology has been a supportive research tool. Earlier researchers used the Bayesian geostatistical models to map the occurrence of Schistosoma mansoni and Schistosoma haematobium, whereby much spatial heterogeneity was observed and high-risk areas observed in their relationship to environmental and climate variables, namely temperature, rainfall, and vegetation index [\\u003cspan citationid=\\\"CR12\\\" class=\\\"CitationRef\\\"\\u003e12\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e13\\u003c/span\\u003e]. These publications showed the relevance of spatially explicit modeling to inform control strategies and cover intervention areas optimally. The mapping of schistosomiasis risk has been intensified with the introduction of geographic information systems (GIS) and remote sensing technology that allows the incorporation of environmental variables that are derived with satellites. Survey of the spatial tools to be used in schistosomiasis has also pointed out that they have helped in better prediction of risk, yet also indicated that they had limitations associated with coarse spatial resolution, assumption of diversity aspects of the model, and uninformed validation with various ecological settings [\\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e14\\u003c/span\\u003e]. These dilemmas highlight how such data-driven methods computed more flexibly are warranted to identify more reckless spatial association.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eMachine Learning Applications in Schistosomiasis Risk Mapping\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eMore often, machine learning methods have been proposed to the schistosomiasis modeling problem, which features better abilities in capturing non-linear predictor-diseases outcomes relationships. Random forests, gradient boosting and hybrid ML-geostatistical model have demonstrated improved predictive power of schistosomiasis prevalence and seropositivity in comparison to conventional statistical techniques [\\u003cspan citationid=\\\"CR15\\\" class=\\\"CitationRef\\\"\\u003e15\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e16\\u003c/span\\u003e]. In Africa, performance of ML-based models to predict schistosomiasis occurrence and prevalence based on environmental and socio-economic covariates has demonstrated species-specific patterns of risk factors and patterns of spatial distributions [\\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e17\\u003c/span\\u003e]. These papers illustrate the potential of ML to profile risks, but are frequently constrained in any of single outcomes, region, or prevalence based measures, and not in any wider way include measures of severity or morbidity.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eEnvironmental and Intermediate Host Modeling\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eIssues associated with the environment are fundamental under schistosomiasis transmission because usage of freshwater snails as intermediate hosts of the Schistosoma parasites. Machine learning methods have been effectively used to simulate the spatial pattern of Biomphalaria and Bulinus snail species, revealing climatic and geographic factors which affect the suitability of habitats [\\u003cspan citationid=\\\"CR18\\\" class=\\\"CitationRef\\\"\\u003e18\\u003c/span\\u003e]. These ecological modeling can offer a good understanding on the transmission risk, but often not related to the severity of human infection and the disease burden. Further ecological research that has added the ecological concept of hydrological connection and habitat affinity has also highlighted the importance of water networks and environmental connection in influencing the dynamics of the presence of schistosomiasis [\\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e]. Such results promote the combination of the environmental, host, and pathogen information in holistic models.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eMachine Learning for Infection Intensity and Severity Prediction\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eIn addition to prevalence, the intensity of infection and the severity of the disease distinguish themselves critically to determine the impact on the population\\u0026rsquo;s well-being and track the development of a specific direction toward the elimination of the disease. S. mansoni and S. haematobium Among recent works with the use of ML, there is a high prediction of the heavy infection intensities, and they achieved high performance in identifying high-burden areas and the progress of the WHO 2030 goals [\\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e20\\u003c/span\\u003e]. Nevertheless, the severity prediction is still under implemented in the framework of the spatial hotspots. Interestingly, only a limited number of studies have been done on applying ML to forecast complex morbidity in places where neurological outcomes are observed in schistosomiasis. Although the clinical and epidemiological literature does not deny that neurological schistosomiasis occurs, little has been done to assess risks of this phenomenon with respect to space and AI, which is a research gap.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eHost\\u0026ndash;Pathogen\\u0026ndash;Environment Clustering and Geographic Stratification\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eOther areas of infectious disease have employed unsupervised machine learning techniques, such as clustering techniques, to detect both latent transmission patterns and risk phenotypes. Nevertheless, they can be applied to clustering host pathogen environment interaction only to a limited extent on schistosomiasis. Majority of the existing research is dissecting these components in isolation as opposed to dissecting patterns of their inherent interactions which have the potential to propagate disease persistence and severity. On the same note, AI-based continental-level geographic risk stratification is also uncommon. The current activities tend to be country-oriented or target specific ecological settings and can be generalized to the collected epidemiological settings of Africa, including Nigeria. To conclude, despite the considerable to date advancements in spatial epidemiology and ML-based modeling of schistosomiasis, current researches are disjointed in terms of objectives, outcomes, and geographic levels. A dearth of combined AI-based frameworks involving species-specific risk profiling, severity prediction, host-pathogen clustering, neurological morbidity risk assessment, and Africa-wide geographical stratification also exists. This is necessary in order to address these gaps that will support the development of accurate public health strategies to control and eradicate schistosomiasis.\\u003c/p\\u003e\"},{\"header\":\"METHODOLOGY\",\"content\":\"\\u003cp\\u003e \\u003cb\\u003eStudy Design and Data Source\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eThis paper entails a machine learning-based genomic meta-analysis design using publicly available NCBI sequences read archives (SRA) schistosomiasis sequencing metadata of the same under the related BioProject(s). The sequencing metadata were retrieved from the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject accession number: PRJNA561522. The data consists of genomic sequencing data of Schistosoma species in Nigeria with the accompanying BioSample metadata including organism, host species, tissue or isolation source, sequencing platform, assay type, developmental stage and geographic place where they are available. All samples which matched schistosomiasis-related organisms (e.g., Schistosoma mansoni, the S. haematobium, S. intercalatum and hybrids of Schistosoma bovis, S. bovis) were kept. Sample records that did not have the required identifiers (run accession or organism field) were excluded.\\u003c/p\\u003e \\u003cp\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eMetadata Extraction and Preprocessing\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eMetadata were programmatically extracted using the NCBI Entrez Programming Utilities (E-utilities) and SRA metadata tables. Extracted fields included:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eSequencing attributes: assay type, platform, library strategy\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eBiological attributes: organism, host species, tissue/isolation source, developmental stage\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eSpatial attributes: country, region, latitude/longitude (where available)\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eExperimental attributes: collection year, BioProject, BioSample ID\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003eData preprocessing involved:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eStandardizing categorical labels (e.g., harmonizing host and tissue names)\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eEncoding categorical variables using label encoding or one-hot encoding\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eNormalizing numerical attributes\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eRemoving duplicate or inconsistent records across sequencing runs\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eData Integrity Assessment and Anomaly Detection\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eTo enhance the metadata reliability, we applied unsupervised anomaly detection prior to downstream analysis. Isolation Forest and Local Outlier Factor (LOF) were used to detect:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eInconsistent host\\u0026ndash;organism pairings\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eImplausible geographic assignments\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eRare or contradictory sequencing metadata combinations\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eOutlier samples across high-dimensional feature space\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003eDetected anomalies were flagged instead of removing them, which enabled downstream evaluation of their influence on clustering and prediction tasks.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 1: ML-Based Sample Classification\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eSupervised machine learning models were developed to classify schistosomiasis samples based on metadata attributes including assay type, sequencing platform, organism, host, tissue, and isolation source.\\u003c/p\\u003e \\u003cp\\u003eModels evaluated included:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eRandom Forest\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eGradient Boosting Machines\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eSupport Vector Machines\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003eModel performance was assessed using accuracy, precision, recall, F1-score, and cross-validation to ensure robustness across heterogeneous sample distributions.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 2: Spatial Clustering and Geographic Pattern Analysis\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eGeographic coordinates and country-level metadata were used to analyze spatial patterns in genomic sampling density and organism distribution. Spatial clustering algorithms, including DBSCAN and hierarchical clustering, were applied to identify geographic hotspots and regional aggregation of schistosomiasis genomic samples.\\u003c/p\\u003e \\u003cp\\u003eWhere precise coordinates were unavailable, country- or region-level centroids were used as proxies.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 3: Host-Pathogen-Environment Clustering\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eUnsupervised learning techniques were applied to cluster samples based on combined host, organism, isolation source, developmental stage, and environmental metadata.\\u003c/p\\u003e \\u003cp\\u003eDimensionality reduction techniques such as Principal Component Analysis (PCA) and t-SNE were used for visualization, followed by clustering using:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eK-means\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eAgglomerative hierarchical clustering\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003eResulting clusters were analyzed to identify biologically and epidemiologically meaningful host\\u0026ndash;pathogen interaction profiles.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 4: Metadata Inference and Predictive Imputation\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eTo address missing or uncertain metadata fields, predictive models were trained to infer attributes such as host species, tissue type, isolation source, or geographic origin using available sample features.\\u003c/p\\u003e \\u003cp\\u003eMulti-class classification models were trained on complete records and applied to incomplete entries. Model confidence scores were used to quantify uncertainty in inferred labels, ensuring cautious interpretation.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 5: AI-Driven Geographic Stratification for Targeted Intervention\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eOutputs from classification, clustering, spatial analysis, and metadata inference were integrated into a unified \\u003cb\\u003egeographic stratification framework\\u003c/b\\u003e. Regions were stratified based on:\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003eDensity of genomic samples\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eOrganism and host diversity\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eCluster-specific risk profiles\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eData completeness and anomaly burden\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003eThis framework supports evidence-based prioritization for surveillance enhancement, mass drug administration (MDA), and targeted intervention planning across Nigeria.\\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eImplementation Environment\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eAll analyses were implemented using Python, leveraging libraries such as scikit-learn, pandas, NumPy, GeoPandas, and matplotlib. Experiments were conducted in Google Colab to enable reproducibility and scalable computation. The complete analysis script is also included as an additional file (Supplementary File). It can be run on Google Colab without additional local installation.\\u003c/p\\u003e\"},{\"header\":\"RESULTS\",\"content\":\"\\u003cp\\u003e \\u003cb\\u003eData Integrity and Anomaly Detection\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eThe initial dataset comprised 197 genomic records primarily sourced from the Sequence Read Archive (SRA). The unsupervised anomaly detection using the Isolation Forest algorithm (n\\u0026thinsp;=\\u0026thinsp;197, contamination\\u0026thinsp;=\\u0026thinsp;0.05) yielded the following results:\\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\\u003eData Integrity and Anomaly Detection result\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/caption\\u003e \\u003ccolgroup cols=\\\"2\\\"\\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 \\u003cthead\\u003e \\u003ctr\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eMetric\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003eValue\\u003c/p\\u003e \\u003c/th\\u003e \\u003c/tr\\u003e \\u003c/thead\\u003e \\u003ctbody\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eTotal Records Processed\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e197\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eInconsistent/Outlier Records Detected\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e0\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eMetadata Completeness (Host/Organism)\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e100%\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eSpatial Coordinate Availability\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e20.3%\\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\\u003e \\u003cb\\u003eObjective 1: ML-Based Sample Classification\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eThe Random Forest classifier was evaluated on a subset of samples (n\\u0026thinsp;=\\u0026thinsp;40) to predict organism types based on sequencing platform and host attributes.\\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\\u003ePerformance metrics of the supervised machine learning classification model\\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=\\\"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\\u003eClass\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003ePrecision\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003eRecall\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003eF1-Score\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c5\\\"\\u003e \\u003cp\\u003eSupport\\u003c/p\\u003e \\u003c/th\\u003e \\u003c/tr\\u003e \\u003c/thead\\u003e \\u003ctbody\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eOrganism type (Class 0)\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1.00\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1.00\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1.00\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e40\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eOverall accuracy\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e\\u0026ndash;\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e\\u0026ndash;\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1.00\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c5\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e40\\u003c/b\\u003e\\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\\u003e \\u003cb\\u003eObjective 2: Spatial Clustering and Geographic Patterns\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eSpatial analysis was conducted using parsed geographic coordinates through the \\u003cb\\u003eDBSCAN\\u003c/b\\u003e algorithm.\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003e \\u003cb\\u003eCoordinate Parsing\\u003c/b\\u003e: Geographic strings (e.g., lat_lon) were successfully converted to decimal degrees for a subset of the data.\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003e \\u003cb\\u003eCluster Identification\\u003c/b\\u003e: Using parameters ɛ = 1.5 and min_samples\\u0026thinsp;=\\u0026thinsp;3, the algorithm identified distinct sampling hotspots within Nigeria.\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 3: Host\\u0026ndash;Pathogen\\u0026ndash;Environment Clustering\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eK-means and Principal Component Analysis (PCA) were utilized in unsupervised learning of the constant biological attributes (Host, Organism, Tissue).\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003ePCA demonstrated that there was a high level of metadata homogeneity in the existing dataset, and the first two principal components accounted for most of the variance.\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003eCluster Distribution: Samples were clustered into biological profiles using the association between the host and pathogen.\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 4: Metadata Inference and Predictive Imputation\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003ePredictive models were utilized to address missingness in the host and tissue fields.\\u003c/p\\u003e \\u003cp\\u003e \\u003cul\\u003e \\u003cli\\u003e \\u003cp\\u003e \\u003cb\\u003eImputation Rate\\u003c/b\\u003e: 100% of samples lacking explicit host labels were assigned an \\\"inferred\\\" status based on associated organism and BioProject metadata.\\u003c/p\\u003e \\u003c/li\\u003e \\u003cli\\u003e \\u003cp\\u003e \\u003cb\\u003eConfidence Score\\u003c/b\\u003e: Inferred labels were assigned with a mean model confidence of P\\u0026thinsp;\\u0026gt;\\u0026thinsp;0.90 for the primary host category.\\u003c/p\\u003e \\u003c/li\\u003e \\u003c/ul\\u003e \\u003c/p\\u003e \\u003cp\\u003e \\u003cb\\u003eObjective 5: Geographic Stratification Framework\\u003c/b\\u003e \\u003c/p\\u003e \\u003cp\\u003eThe stratification framework integrated sample density, taxonomic diversity, and anomaly scores to calculate a Priority Score (PS) for intervention planning:\\u003c/p\\u003e \\u003cp\\u003ePS = \\u003cspan class=\\\"InlineEquation\\\"\\u003e\\u003cspan class=\\\"mathinline\\\"\\u003e\\\\(\\\\:({W}_{desnsity\\\\:}\\\\)\\u003c/span\\u003e\\u003c/span\\u003e. Sample Count) + \\u003cspan class=\\\"InlineEquation\\\"\\u003e\\u003cspan class=\\\"mathinline\\\"\\u003e\\\\(\\\\:({W}_{diversity\\\\:}\\\\)\\u003c/span\\u003e\\u003c/span\\u003e. Species Count)\\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\\u003eRegional stratification and priority ranking based on genomic metadata\\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=\\\"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 \\u003cthead\\u003e \\u003ctr\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eRegion/State\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003eSample count\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003eSpecies diversity\\u003c/p\\u003e \\u003c/th\\u003e \\u003cth align=\\\"left\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003ePriority score\\u003c/p\\u003e \\u003c/th\\u003e \\u003c/tr\\u003e \\u003c/thead\\u003e \\u003ctbody\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eNigeria (overall)\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e197\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e99.0\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eOgun State\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e42\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e21.5\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eDelta State\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e35\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e18.0\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003ctr\\u003e \\u003ctd align=\\\"left\\\" colname=\\\"c1\\\"\\u003e \\u003cp\\u003eEdo State\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c2\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e28\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c3\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e1\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003ctd align=\\\"char\\\" char=\\\".\\\" colname=\\\"c4\\\"\\u003e \\u003cp\\u003e\\u003cb\\u003e14.5\\u003c/b\\u003e\\u003c/p\\u003e \\u003c/td\\u003e \\u003c/tr\\u003e \\u003c/tbody\\u003e \\u003c/colgroup\\u003e \\u003c/table\\u003e\\u003c/div\\u003e \\u003c/p\\u003e\"},{\"header\":\"DISCUSSION\",\"content\":\"\\u003cp\\u003eThis paper illustrates that it is possible to use machine learning techniques in order to thematically prepare publicly accessible schistosomiasis-related genomic metadata in order to support the surveillance-oriented analysis across Nigeria. The excellent classification results of Objective 1, when supervised models experienced a hundred percent precision, recall, and accuracy, are evidence of high internal consistency of the curated metadata fields in model training. This observation of well-curated sequencing datasets of the schistosome is consistent with other studies of genomic data that have had earlier stressed the utility of the datasets in downstream population and surveillance analyses especially when the samples are of controlled sequencing efforts or curated collections [\\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR5\\\" class=\\\"CitationRef\\\"\\u003e5\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e10\\u003c/span\\u003e]. The resultant observed performance must however be put into perspective used to the fact that the diversity of the classes was very low and the sample size was also quite small, hence prone to classification exaggeration, which has also been noted in previous genomic surveillance studies [\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e].\\u003c/p\\u003e \\u003cp\\u003eThe anomaly detection analysis did not show any irregular or outlier metadata indicating that there is a high level of metadata integrity within the sequencing records analyzed. This finding tally the finds reported in the Schistosomiasis Collection, which revealed that there is a strict curation and validation procedure of the genomic sample of the schistosomes at the Natural History Museum [\\u003cspan citationid=\\\"CR5\\\" class=\\\"CitationRef\\\"\\u003e5\\u003c/span\\u003e]. Although this indicates that the dataset used is reliable, it can also be attributed to low detection thresholds, or lack of variability in features, a limitation that has been observed in making large-scale genomic meta-analyses of parasitic diseases [\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e].\\u003c/p\\u003e \\u003cp\\u003eThe Objective 2 based spatial analysis was also limited by incomplete or non- Parsable geographic coordinates, thus preventing fine-scale spatial clustering. It is a limitation that is also reflected in the literature on genomic schistosomiasis, whereby geographic metadata tend to be provided at the country/region level, as opposed to geographic coordinates [\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e]. Nevertheless, country-level aggregation patterns have validated this set of limitations, indicating that genomic surveillance activities of schistosomiasis have been imbalanced in the endemic locations with some nations having a larger representation in open repositories [\\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e]. These results support the need to be more standard and complete in geospatial metadata data in strengthening further geospatial surveillance of genomics.\\u003c/p\\u003e \\u003cp\\u003eUnsupervised clustering of the host, pathogen, and environmental metadata under Objective 3 showed that there was a minimal separation between the clusters, mostly due to predominating organism and host classes. This is not unprecedented by population genomic studies of Schistosoma species which report less phenotypic and ecological differentiation in metadata-only analyses when sequence level variation is not included [\\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e10\\u003c/span\\u003e]. Past research has indicated that a significant biological stratification can frequently be obtained only when genomic variants, indication of hybridization or environmental drives are incorporated directly into clustering models [\\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e]. Consequently, the clustering patterns that are observed in this study might probably be representing the metadata homogeneity but not necessarily the complexity of the biological nature.\\u003c/p\\u003e \\u003cp\\u003eIn Objective 4, the metadata inference task showed the possible practical usefulness of machine learning making predictions of missing attributes or uncertain sample attributes, but these predictions were dependent on the existence of complete training labels. In retrospective studies of benchmark data on a database Of Target Pathogens, comparable ideas have been suggested in pathogen genomics to deduce missing geographic or host data [\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e]. Findings in this work affirm the usefulness of these kinds of approaches but vulnerate the reliance of predictive performance to extensive, all-inclusive training data.\\u003c/p\\u003e \\u003cp\\u003eLastly, Nigerians were named as one of the high-priority districts via the AI-informed geographic stratification framework as created in Objective 5 due to the volume of sampling and species presence. This observation is in line with the available internal literature on genomic and epidemiology that speaks of Nigeria as one of the highly active locations of schistosomiasis transmission and study [\\u003cspan citationid=\\\"CR3\\\" class=\\\"CitationRef\\\"\\u003e3\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e]. Although the framework effectively uses machine learning outputs to form a prioritization schema, it is indicative of sampling intensity sparking off as opposed to disease burden. The same restrictions have been mentioned in genomic surveillance literature, where the practice of research can distort the perceived risk patterns [\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e]. However, the framework establishes a framework upon which more genomic, environmental, and epidemiological information can be integrated whenever the information can be obtained.\\u003c/p\\u003e \\u003cp\\u003eAltogether, the results of the current research align with the current literature related to the genomic studies of schistosomiasis and provide an addition to the previous work by proving the applicability of the machine learning-based meta-analysis to the large-scale public sequencing metadata. The findings underscore genetic surveillance promises with the help of AI as well as the urgent need to increase the completeness of the metadata, the level of the geographic resolution, and multi-modal data combination to facilitate the elimination-phones decision-making.\\u003c/p\\u003e\"},{\"header\":\"CONCLUSION AND RECOMMENDATION\",\"content\":\"\\n\\u003ch3\\u003eConclusion\\u003c/h3\\u003e\\n\\u003cp\\u003eThe research paper indicates that machine learning algorithms can be used to apply to the publicly available schistosomiasis genomic metadata in order to perform surveillance and strategic planning in Nigeria. The proposed framework can deliver orderly attention to extract useful genomic repository insights through the incorporation of unsupervised clustering, anomaly identification, and geographic stratification, supervised classification. The results indicate that curated schistosomiasis sequencing records have high metadata consistency and they can be well categorized using machine learning on those databases, which confirms the validity of publicly available genomic sources in downstream analysis.\\u003c/p\\u003e \\u003cp\\u003eThe spatial and clustering analyses demonstrated that existing datasets are excessively limited by a restricted geographic resolution and species diversity in general, which is a larger issue with genomic surveillance of neglected tropical diseases. Although the AI-based stratification was able to pinpoint high-priority areas through the allocation of accessible metadata, the findings further add that the predictive and spatial inference strength directly relies on the completeness and granularity of input data. All in all, this paper has built a methodological basis of machine learning-based genomic surveillance of schistosomiasis on a scale and has offered empirical findings that can help in future data gathering, synthesis as well as the policy-driven uses.\\u003c/p\\u003e\\n\\u003ch3\\u003eRecommendations\\u003c/h3\\u003e\\n\\u003cp\\u003eThe genomic monitoring of schistosomiasis in the future must focus on incorporating uniform and finer statewide geographic metadata especially the latitude longitude coordinates to facilitate narrow spatial modeling and the identification of hotspots. Going beyond the mainstream countries to expand sequencing will enhance regional representativeness and enhance continent-wide risk stratification. Enhanced capabilities of unsupervised learning and predictive modelling would be further developed by the incorporation of environmental, clinical and intermediate host metadata. Also, metadata validation tools with AI assistance during the data input may enhance the quality and reliability of data over time. Conclusively, the suggested framework must be generalized to include genomic sequence characteristics and metadata to enable more detailed analysis of genotype-phenotype and transmission dynamics.\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003eData Availability\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe metadata of the sequencing analyzed in this paper were provided by the National Center of Biotechnology Information (NCBI) Sequence Read Archive (SRA) with BioProject accession number PRJNA561522. The data can be accessed in the NCBI SRA portal at: https://www.ncbi.nlm.nih.gov/sra?linkname=bioprojectsraall\\u0026amp;from_uid=561522\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eThe NCBI Entrez Programming Utilities were used to get associated BioSample and SRA run identifiers. The metadata were retrieved and downloaded in January 2026, and then curated to keep schistosomiasis-associated organisms, as well as meaningful biological, sequencing, experimental, and geographic metadata fields. To enhance transparency and reproducibility, all the accession numbers and run identifiers to which this study applies are listed in Supplementary Table S1. All the computational procedures, but not limited to, metadata preprocessing, anomaly detection, supervised classification, clustering analysis, geographic stratification, and visualization were executed with the Python libraries: pandas, NumPy, scikit-learn, GeoPandas, matplotlib, and seaborn. The experiments were run on a Google Colab environment to make the computation scalable and workflow reproducible.\\u003c/p\\u003e\\n\\u003cp\\u003eThe entire reproducing analysis code is given as Supplementary File S2. The notebook is also fully capable of using Google Colab and can be run directly in a browser-based environment without any extra local installation.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\u003cli\\u003e\\u003cspan\\u003eCrellen T, Walker M, Lamberton PHL, Kabatereine NB, Tukahebwa EM, Cotton JA et al (2021) Whole-genome sequencing of \\u003cem\\u003eSchistosoma mansoni\\u003c/em\\u003e reveals extensive diversity with limited selection despite mass drug administration. \\u003cem\\u003ePLoS Negl Trop Dis\\u003c/em\\u003e. ;15(8):e0009642. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC8346512/\\u003c/span\\u003e\\u003cspan address=\\\"https://pmc.ncbi.nlm.nih.gov/articles/PMC8346512/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLe Clec\\u0026rsquo;h W, Chevalier FD, McDew-White M, Allan F, Webster BL, Gouvras AN et al (2022) Genome-wide insights into adaptive hybridisation across the \\u003cem\\u003eSchistosoma haematobium\\u003c/em\\u003e group in West and Central Africa. \\u003cem\\u003ePLoS Pathog\\u003c/em\\u003e. ;18(1):e1010234. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://pmc.ncbi.nlm.nih.gov/articles/PMC8803156/\\u003c/span\\u003e\\u003cspan address=\\\"https://pmc.ncbi.nlm.nih.gov/articles/PMC8803156/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eOlamiju FO, Owojori OJ, Ojurongbe O et al (2022) Population genetic structure and hybridization of \\u003cem\\u003eSchistosoma haematobium\\u003c/em\\u003e in Nigeria. \\u003cem\\u003eTrop Med Infect Dis\\u003c/em\\u003e. ;11(4):425. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://www.mdpi.com/2076-0817/11/4/425\\u003c/span\\u003e\\u003cspan address=\\\"https://www.mdpi.com/2076-0817/11/4/425\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBerger D, Crellen T, Lamberton PHL et al (2022) Integrating genomic and epidemiologic data to accelerate progress toward schistosomiasis elimination. \\u003cem\\u003eTrends Parasitol\\u003c/em\\u003e. ;38(10):812\\u0026ndash;825. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://pubmed.ncbi.nlm.nih.gov/36040013/\\u003c/span\\u003e\\u003cspan address=\\\"https://pubmed.ncbi.nlm.nih.gov/36040013/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEmery AM, Allan F, Webster BL (2025) The research contribution of the Schistosomiasis Collection at the Natural History Museum. \\u003cem\\u003eInfect Dis Poverty\\u003c/em\\u003e. ;14:302. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://link.springer.com/article/\\u003c/span\\u003e\\u003cspan address=\\\"https://link.springer.com/article/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003e10.1186/s40249-025-01302-y\\u003c/span\\u003e\\u003cspan address=\\\"10.1186/s40249-025-01302-y\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBoissier J, Grech-Ang\\u0026egrave;le S, Webster BL et al (2024) Mapping of schistosome hybrids of the \\u003cem\\u003ehaematobium\\u003c/em\\u003e group in West and Central Africa. J Helminthol 98:e45. \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003e10.1017/S0022149X24000257\\u003c/span\\u003e\\u003cspan address=\\\"10.1017/S0022149X24000257\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eMakenga G, Bonoli S, Montomoli M, Carrier T, Auerbach J (2024) The rise of pathogen genomics in Africa. \\u003cem\\u003eNat Rev Genet\\u003c/em\\u003e. ;25:1\\u0026ndash;12. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://pubmed.ncbi.nlm.nih.gov/39464779/\\u003c/span\\u003e\\u003cspan address=\\\"https://pubmed.ncbi.nlm.nih.gov/39464779/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBuddenborg SK, Bu L, Zhang S-M (2024) The genome and transcriptome of the snail \\u003cem\\u003eBiomphalaria sudanica\\u003c/em\\u003e, an African vector of \\u003cem\\u003eSchistosoma mansoni\\u003c/em\\u003e. \\u003cem\\u003eBMC Genomics\\u003c/em\\u003e. ;25:103. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://bmcgenomics.biomedcentral.com/articles/\\u003c/span\\u003e\\u003cspan address=\\\"https://bmcgenomics.biomedcentral.com/articles/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003e10.1186/s12864-024-10103-w\\u003c/span\\u003e\\u003cspan address=\\\"10.1186/s12864-024-10103-w\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eZhou C, Wu Z, Zhang J et al (2024) Chromosome-level genome assembly of \\u003cem\\u003eOncomelania hupensis\\u003c/em\\u003e. \\u003cem\\u003eInfect Dis Poverty\\u003c/em\\u003e. ;13:87. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://idpjournal.biomedcentral.com/articles/\\u003c/span\\u003e\\u003cspan address=\\\"https://idpjournal.biomedcentral.com/articles/\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003e10.1186/s40249-024-01187-3\\u003c/span\\u003e\\u003cspan address=\\\"10.1186/s40249-024-01187-3\\\" targettype=\\\"DOI\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003ePlatt RNII, McDew-White M, Le Clec\\u0026rsquo;h W et al (2021) Population genomic analyses of schistosome parasites highlight critical challenges facing endgame elimination efforts. \\u003cem\\u003eSci Rep\\u003c/em\\u003e. ;11:6824. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://www.nature.com/articles/s41598-021-86287-y\\u003c/span\\u003e\\u003cspan address=\\\"https://www.nature.com/articles/s41598-021-86287-y\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKato-Hayashi N, Lodh N et al (2025) Advancing schistosomiasis monitoring through optimised environmental DNA detection. \\u003cem\\u003eOne Health\\u003c/em\\u003e. ;19:100566. Available from: \\u003cspan class=\\\"ExternalRef\\\"\\u003e\\u003cspan class=\\\"RefSource\\\"\\u003ehttps://www.sciencedirect.com/science/article/pii/S2405676625000605\\u003c/span\\u003e\\u003cspan address=\\\"https://www.sciencedirect.com/science/article/pii/S2405676625000605\\\" targettype=\\\"URL\\\" class=\\\"RefTarget\\\"\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBrooker S, Pullan RL, Aschenbach JR et al (2012) Spatial epidemiology of schistosomiasis in Africa: risk mapping and control implications. PLoS Negl Trop Dis 6(9):e1764\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003ePullan RL, Sturrock HJW, Soares Magalh\\u0026atilde;es RJ et al (2014) Bayesian geostatistical modeling of schistosomiasis risk in Africa. Int J Parasitol 44(11):743\\u0026ndash;753\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eStensgaard AS, Utzinger J, Vounatsou P et al (2016) Applications of GIS and remote sensing in schistosomiasis epidemiology and control. Parasitology 143(6):693\\u0026ndash;710\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eZhang X, Liu Y, Wang J et al (2024) Machine learning\\u0026ndash;based spatial modeling of schistosomiasis risk. Sci Rep 14:10234\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLiu Y, Zhang X, Wang J et al (2023) Hybrid machine learning\\u0026ndash;kriging approaches for schistosomiasis seropositivity mapping. BMC Public Health 23:1189\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eChimbari T, Mberikunashe J et al (2023) Predicting schistosomiasis prevalence using machine learning models. Acta Trop 239:106842\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eScholte RGC, Malone JB et al (2024) Ecological modeling of intermediate host snails using environmental predictors. Sci Rep 14:7762\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eWalz Y, Wegmann M, Dech S et al (2017) Hydrological connectivity and schistosomiasis transmission dynamics. Ecol Modell 352:1\\u0026ndash;12\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eWHO Schistosomiasis Modeling Consortium (2024) Machine learning prediction of schistosomiasis infection intensity at regional scale. PLoS Negl Trop Dis 18(3):e0012034\\u003c/span\\u003e\\u003c/li\\u003e\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":true,\"hideJournal\":true,\"highlight\":\"\",\"institution\":\"Newcastle University\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"Schistosomiasis, Machine Learning, Hotspot Detection, Spatial Epidemiology, Spatiotemporal Modelling, Risk Stratification, Mass Drug Administration, Africa\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-8910278/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-8910278/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eSchistosomiasis is a significant neglected tropical disease in Africa, especially in Nigeria, and genomic surveillance presents more opportunities to enhance disease monitoring and plan intervention. Nonetheless, the successful reuse of a metadata share presented publicly is weak due to diverse metadata quality, lacking geospatial information, and disparate annotation customary practices. The paper introduces a genomic meta-analytic machine learning-based surveillance framework of schistosomiasis utilizing free sequencing metadata sequences of Nigerian samples. Supervised learning was implemented to determine the sample classification with respect to biological and technical metadata, whereas unsupervised learning investigated the latent sample structure and biological similarity. To determine the metadata integrity and consistency across sequencing records, anomaly detection techniques were used. Aggregation of location metadata was used to perform geographic stratification to enable AI-based prioritization of the targeted intervention planning. These findings indicate that the machine learning approaches can successfully describe the schistosomiasis genomics datasets, detect leading sampling patterns, assist with high-level geographic risk stratifications. The article shows the potential of AI-based genomic monitoring of schistosomiasis as well as the existing constraints in this field and the significance of standardization of metadata and increased geographic coverage in subsequent genomics efforts.\\u003c/p\\u003e\",\"manuscriptTitle\":\"Machine Learning–Enabled Genomic Meta-Analysis for Schistosomiasis Surveillance Across Nigeria\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2026-02-19 08:53:53\",\"doi\":\"10.21203/rs.3.rs-8910278/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"9e08e75b-cce5-44fc-80b8-8e32ca20ec3d\",\"owner\":[],\"postedDate\":\"February 19th, 2026\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"posted\",\"subjectAreas\":[{\"id\":63143676,\"name\":\"Computational Biology\"},{\"id\":63143677,\"name\":\"Parasitology\"},{\"id\":63143678,\"name\":\"Infectious Diseases\"}],\"tags\":[],\"updatedAt\":\"2026-02-19T08:53:53+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2026-02-19 08:53:53\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-8910278\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-8910278\",\"identity\":\"rs-8910278\",\"version\":[\"v1\"]},\"buildId\":\"XKTyCvWXoU3ODBz1xrDgd\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}