Economic Deprivation, Diminished Social Support, and Loneliness: An Empirical Test of the EDSL Model Using UK Biobank Data

Economic Deprivation, Diminished Social Support, and Loneliness: An Empirical Test of the EDSL Model Using UK Biobank Data | 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 Economic Deprivation, Diminished Social Support, and Loneliness: An Empirical Test of the EDSL Model Using UK Biobank Data Anhui Kong, Ziyue Wu, Fangqing Liu, Jinxuan Ni, JiaXing Deng, and 14 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7553119/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Background Loneliness is increasingly recognized as a major public health concern shaped by both economic and social structures. However, prior research often examined these determinants in isolation, limiting understanding of their combined effects. Methods We proposed and tested the Economic Deprivation, Support Diminution, and Loneliness (EDSL) Model using UK Biobank data (N = 232,477). Economic deprivation was assessed with the Townsend Deprivation Index (TDI) and the Index of Multiple Deprivation (IMD), while social support was measured by household size, visit frequency, and confiding frequency. Loneliness was coded as a binary outcome. Machine learning models predicted loneliness, with SHAP values identifying influential predictors. Structural equation modeling (SEM) examined both direct and indirect pathways from economic deprivation to loneliness. Results XGBoost predicted loneliness well (AUC = 81.35%). SHAP values showed confiding frequency, IMD, and household size as key predictors. SEM confirmed economic deprivation’s direct ( β = 0.06) and indirect effects on loneliness via reduced social support (indirect β = 0.07). Conclusions Findings provide robust empirical support for the EDSL Model, showing that economic deprivation fosters loneliness both directly and indirectly by eroding social support. These results highlight the need for integrated public health strategies that address economic hardship while strengthening social connectedness to reduce loneliness. Trial registration: Not applicable. Loneliness Economic Deprivation Social Support EDSL Model UK Biobank Machine Learning Structural Equation Modeling Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction Loneliness, defined by the discrepancy between desired and actual social connections, has been identified as a significant global public health concern with impacts on diverse populations worldwide ( 1 ). Current research has found that the important factors causing loneliness are far more than individual psychological factors, but are closely related to the entire social structure, resource allocation and other factors ( 2 ). Effectively addressing loneliness necessitates a comprehensive understanding of the complex interplay between social and economic structures that shape the formation, maintenance, and quality of social relationships. However, previous research has often investigated these determinants in isolation. For instance, numerous studies have established strong correlations between higher economic deprivation and increased loneliness (e.g., ( 3 , 4 ). Concurrently, a significant body of evidence underscores the protective role of social support (e.g., ( 5 , 6 ). These studies often fail to elucidate the synergistic effects and potential effects pathways—specifically, how economic deprivation might systematically erode social support, thereby exacerbating loneliness. Even foundational theoretical frameworks, such as social network theory (SNT), which emphasizes interpersonal relationship structures (7), and the social determinants of health framework (SDOH), which highlights broader socioeconomic conditions ( 8 ), have tended to maintain distinct focuses—SNT on relational dynamics and SDOH on structural inequalities. Consequently, explicit theoretical and empirical examination of how structural economic factors transact with social relational factors to jointly produce loneliness has been less common, limiting the development of holistic theoretical models and comprehensive intervention strategies. This fragmentation underscores a critical gap: while an integrative perspective arguing that economic deprivation restricts social participation and deepens isolation is gaining traction (e.g. ( 9 , 10 ), the precise mechanisms and empirical support for comprehensive models detailing these interactions, particularly using large-scale datasets, are still developing. To address this gap, this study introduces and empirically tests an Economic Deprivation, Support Diminution, and Loneliness (EDSL) Model. The EDSL Model is predicated on the understanding that economic deprivation is often accompanied by the limitation of the sources of social support—a process where individuals are marginalized due to a lack of resources or opportunities, preventing normal participation in community social activities. By definition, economic deprivation can mean the inability to afford activities that promote social integration. As classic economic deprivation researchers have noted, individuals experiencing economic deprivation are often unable to participate in ordinary social outings, hobbies, celebrations, or civic activities, because even seemingly minor expenses (such as a cup of coffee with a friend, bus fare to visit relatives, a birthday gift, or a club membership fee) can become prohibitive ( 11 ). This economic deprivation, therefore, can lead to the impairment of social support pathways, further amplifying loneliness. The EDSL Model posits that economic deprivation impacts loneliness through two primary pathways: H1) there is a direct pathway where economic deprivation; H2) Economic deprivation indirectly exacerbates loneliness by causing a diminution of social support. This diminution occurs as the aforementioned processes of exclusion stemming from economic deprivation systematically erode an individual’s access to, and the quality of, their social support networks. This diminished social support then acts as a key mediating factor that further amplifies the experience of loneliness. The concept map of this model is presented in Fig. 1 . A growing body of empirical research lends preliminary support to the EDSL Model’s tenets. For instance, longitudinal studies demonstrate that material deprivation significantly predicts higher loneliness levels even two years later, an effect partially mediated by reduced engagement in cultural activities ( 12 ). Concurrently, research correlates lower social capital with increased loneliness, found that lower levels of social participation, social connection, and reciprocity were significantly associated with higher odds of loneliness in older adults ( 13 ). Furthermore, studies employing structural equation modeling (SEM) or similar path analyses are beginning to unravel the complex interplay between socioeconomic factors, social resources, and loneliness-related outcomes. Research by ( 14 ) showed that social support mediated the relationship between social engagement (itself often linked to socioeconomic opportunities) and loneliness. While these studies are valuable, they often do not comprehensively test an integrated model that explicitly delineates both the direct impact of various economic deprivation indicators on loneliness and the concurrent mediating role of multifaceted social support measures (reflecting support diminution) within a large-scale cohort, as proposed by the EDSL Model. This study seeks to build upon this emerging evidence by providing a more robust and nuanced empirical validation of the EDSL Model. The empirical basis for this analysis is data from the UK Biobank. Economic deprivation is combined the Townsend Deprivation Index (TDI) and the Index of Multiple Deprivation (IMD). This dual approach allows a nuanced examination of economic deprivation’s facets. Social support is operationalized using established indicators: “Total number of household members” (Household Size), “Frequency of visits with friends or family” (Visit Frequency), and “Frequency of confiding in others” (Confiding Frequency). Methodologically, this research first employs machine learning to determine the predictive effect of these factors on loneliness, using Shapley Additive Explanations (SHAP) values to ascertain influence direction. Subsequently, these insights inform a SEM to rigorously test the hypothesized direct effects of economic deprivation on loneliness and the indirect effects mediated by social support indicators within the EDSL. This study aims to provide robust empirical evidence for the EDSL, thereby enhancing our understanding of the interconnected socioeconomic and social-relational drivers of loneliness 2. Methods 2.1 Participants This study utilized data from the UK Biobank, employing the data release dated September 17, 2019. The UK Biobank is a large-scale, prospective cohort study that enrolled over 500,000 participants aged 40 to 69 years at baseline (recruitment period: 2007–2010) across England, Scotland, and Wales. For the present analysis, a total of 232,477 participants were included to investigate the relationship between loneliness and a range of independent variables. Participants were bifurcated into two groups based on self-reported loneliness status: the “loneliness group” (n = 73,884) and the “non-loneliness group” (n = 158,593). Ethical approval for the UK Biobank was granted by the National Health Service Research Ethics Service (approval date: June 17, 2011; reference: 11/NW/0382). Access to and analysis of the data for this study were authorized by the UK Biobank under Application Number 37292. 2.2 Quality control and descriptive statistics Firstly, we calculated the variance inflation factor (VIF) for each independent variable. A VIF value below 5 was considered acceptable, indicating no severe multicollinearity. And we calculated the correlation coefficients between the variables to ensure that there was no excessive correlation affecting the model performance ( r < 0.7). Subsequently, we conducted descriptive statistics on the two sets of data and used chi-square tests and independent sample t-tests and chi-square test to examine whether there were significant differences in other variables between the two groups. 2.3 Measurement Data for this study were sourced from the UK Biobank database, encompassing variables across demographic, social support, and socioeconomic dimensions. The sole demographic variable was participant self-reported Gender, coded dichotomously (1 = Male, 0 = Female). Indicators of social support included the count variable Household Size, Visit Frequency, and Confiding Frequency; for the latter two categorical variables, higher scores signify greater frequency. Socioeconomic deprivation was assessed using two area-based indices linked via participant postcode: the Townsend Deprivation Index (TDI), a continuous variable calculated from census data on unemployment, car non-ownership, non-home ownership, and household overcrowding; and the Index of Multiple Deprivation (IMD), a composite continuous score or rank integrating indicators across domains such as income, employment, health, education, housing, crime, and environment. For both TDI and IMD, higher scores or ranks denote greater levels of deprivation. The dependent variable, Loneliness, was measured using the question “Do you often feel lonely?” and operationalized as a dichotomous variable, with ‘Yes’ responses coded as 1 and ‘No’ responses coded as 0; only participants providing these definitive answers were included in the final analysis. 2.4 Machine Learning To evaluate the association between loneliness and the independent variables, six machine learning algorithms were utilized: Logistic Regression (LR), Naive Bayes (NB), Decision Tree (DT), Random Forest (RF), K-nearest Neighbors (KNN), and Extreme Gradient Boosting (XGBoost). These algorithms were selected to provide a comprehensive assessment of the models’ capacity to differentiate between lonely and non-lonely individuals, to balance precision and recall, and to minimize overfitting. To ensure methodological consistency across all machine learning approaches, the dataset was partitioned into training and testing subsets, with the testing set comprising 20% of the total data. This 80/20 split is a conventional practice in machine learning, aiming to balance the requirement for adequate training data with the need for a robust evaluation set. Implementation of these algorithms was performed using the scikit-learn library ( 15 ). The performance of each algorithm was assessed using a suite of metrics, including area under the curve (AUC), accuracy, precision, recall, and the F1-score. 2.5 SHAP Value To improve the interpretability of the machine learning models, the SHAP methodology was utilized. SHAP provides a unified methodology for interpreting model outputs by quantifying the contribution of each feature to individual predictions. In this analysis, SHAP values were computed for the XGBoost model, which exhibited the optimal performance in predicting loneliness. In this study, we determined a preliminary direction for the prediction of loneliness by each variable through SHAP value. 2.6 SEM analysis We used the SEM to verify the impact of economic deprivation on loneliness and that economic deprivation further enhances loneliness by depriving social support. Therefore, we take economic deprivation as the independent variable, social support as the mediating variable, and loneliness as the dependent variable. Latent variables with only two indicator variables can be identified under certain conditions, but their estimates may not be stable enough and are more susceptible to incorrect Settings of sample characteristics and other parts of the model. Therefore, we take the total scores of TDI and IMD as the direct indicators of economic deprivation. Construct latent variables of social support through three indicators. The model was estimated using Weighted Least Squares Mean and Variance Adjusted (WLSMV), which is suitable for the classified variable. To assess model fit, we employed several widely used fit indices, including Goodness of Fit Index (GFI), Root Mean Square Residual (RMR) and Standardized Root Mean Square Residual (SRMR). A good model fit is indicated by GFI value above 0.9, RMR and SRMR values below 0.08. After the initial model fitting did not achieve satisfactory fit, we examined modification indices (MI) to identify potential areas of misspecification. A covariance between Loneliness and Visit Frequency (Loneliness ~ ~ Visit Frequency), corresponding to the largest MI, was subsequently added to the model. This modification was chosen to improve model fit while also carefully ensure model convergence. 3. Results 3.1 Quality control and descriptive statistics First of all, we ensured that there was no serious multicollinearity problem among the various variables (VIF < 5, r < 0.7, Fig. 2 B, C). The independent sample t-test and chi-square test revealed significant intergroup differences in various variables between the lonely group and the non-lonely group. The TDI of the lonely group was significantly higher than that of the non-lonely group ( t = 18.13, p < 0.001, d = 0.08). The IMD was significantly higher in the lonely group than in the non-lonely group ( t = 61.50, p < 0.001, d = 0.26). Household Size was significantly smaller in the lonely group than in the non-lonely group ( t = -51.58, p < 0.001, d = -0.21). Visit Frequency was significantly lower in the lonely group than in the non-lonely group ( t = -34.17, p < 0.001, d = -0.14). The Confiding Frequency in the Lonely group was significantly lower than that in the non-lonely group ( t = -132.11, p < 0.001, d = -0.55). There are significant differences in gender between the two groups ( χ ² = 18.13, p < 0.001). Descriptive statistics and independent sample t-test results are presented in Table 1 and Fig. 2 . Table 1 Descriptive statistics and difference test results of the two groups Items Lonely group (M ± SD) Non-Lonely Group (M ± SD) t TDI -1.31 ± 3.08 -1.55 ± 2.91 18.13 *** IMD 20.58 ± 15.61 16.73 ± 13.29 61.50 *** Household Size 2.18 2.49 -51.58 *** Visit Frequency 2.75 2.93 -34.17 *** Confiding Frequency 2.68 3.77 -132.11 *** Lonely group (Percentage) Not Lonely Group (Percentage) χ 2 Gender (Number of males) 27505 (37.22%) 75187 (47.41) 2118.74 *** Note: ***: p < 0.001 3.2 Machine learning model performances To assess the performance of various machine learning algorithms, Table 2 presents key metrics, including AUC score, Accuracy, Precision, Recall, and F1 score. The AUC of all models was greater than the chance level (AUC > 0.5), demonstrating excellent predictive performance. Specifically, XGBoost demonstrated the best predictive performance (AUC = 81.35%). This indicates that these factors can significantly predict loneliness. In Fig. 3 A, we present the ROC curves of all models. Figure 3 B presents the confusion matrix for the best-performing XGBoost model. Table 2 All machine learning model performances Metrics LR NB DT RF KNN XGBoost AUC 70.29% 69.66% 79.75% 80.82% 76.91% 81.35% Accuracy 66.69% 69.37% 77.87% 75.65% 75.66% 75.74% Precision 48.13% 53.05% 72.84% 60.46% 64.60% 60.37% Recall 61.89% 31.58% 48.45% 67.61% 51.81% 68.88% F1 score 54.15% 39.59% 58.19% 63.84% 57.50% 64.35% Note: LR: logistic regression; NB: naive bayes; DT: decision tree; RF: random forest; KNN: k-nearest neighbors; XGBoost: extreme gradient boosting. (A) Receiver operating characteristic (ROC) curves for six classifiers: random forest (RF), logistic regression (LR), naive Bayes (NB), decision tree (DT), k-nearest neighbors (KNN), and extreme gradient boosting (XGBoost). The XGBoost model yielded the highest area under the curve (AUC = 81.35). (B) Confusion matrix of the best-performing model (XGBoost) showing true and predicted classifications of loneliness. 3.3 SHAP Analysis for Feature Importance To elucidate the contribution of each feature to the model’s predictions of loneliness, a SHAP analysis was conducted on the best-performing XGBoost model. Figure 4 A presents the mean absolute SHAP values, indicating the average impact of each predictor on the model output magnitude. Confiding Frequency exhibited the largest mean SHAP value, identifying it as the most influential predictor of loneliness. This was followed by IMD and Household Size, which also demonstrated substantial impacts. TDI and Gender showed moderate importance, while Visit Frequency had a comparatively smaller, yet still relevant, impact on the model’s predictions. Figure 4 B provides a SHAP summary plot, visualizing the direction and distribution of each feature’s impact on the model output. The directionality of SHAP roughly shows the negative prediction of loneliness by social support factors and the positive prediction of loneliness by economic deprivation indicators. 3.4 SEM Results The corrected SEM model demonstrated excellent fit, as indicated by the following fit indices: GFI = 0.99, RMR = 0.05, SRMR = 0.07. The analysis revealed a significant direct effect of socioeconomic deprivation on loneliness ( β = 0.06, SE = 0.004, z = 13.70, p < 0.001, 95% CI: [0.05, 0.07]). In addition to the direct effect, the analysis identified a significant mediation pathway through which economic deprivation influences loneliness. Specifically, economic deprivation was found to reduce the support ( β = -0.08, SE = 0.004, z = -20.19, p < 0.001, 95% CI: [-0.09, -0.07]). In turn, diminished social support significantly exacerbated feelings of loneliness ( β = -0.88, SE = 0.010, z = -87.87, p < 0.001, 95% CI: [-0.90, -0.86]). The standardized estimated value of the indirect effect was 0.07 ( SE = 0.003, z = 19.69, p < 0.001). The standardized estimated value of the total effect was 0.13 ( SE = 0.003, z = 46.00, p < 0.001). In addition, it is worth noting that we have added an error covariance between visit frequency and loneliness ( ψ = -1.02, SE = 0.050, z = -16.89, p < 0.001, 95% CI: [-1.14, -0.90]), which is completely data-driven. The result of SEM is presented in Fig. 5 . 4. Discussion This study aimed to introduce and empirically test the EDSL Model, positing that economic deprivation directly affect loneliness and indirectly exacerbates it by diminishing social support. The findings from a large-scale analysis of the UK Biobank dataset offer robust support for the EDSL Model, highlighting the dual pathways through which economic hardship contributes to loneliness. 4.1 Confirmation of the EDSL Model and Hypotheses Firstly, results from machine learning analyses revealed that a model incorporating two economic deprivation indices (TDI and IMD) and three social support indicators (household size, visit frequency, and confiding frequency) predicted loneliness with high accuracy (AUC = 0.81). Among these predictors, frequency of confiding in others emerged as the most influential variable, underscoring the critical role of emotionally intimate relationships in buffering against loneliness. This finding aligns with prior research suggesting that not just the presence, but the quality of social ties—particularly those enabling emotional disclosure—is vital for mental well-being. The empirical results align closely with the proposed EDSL Model and its core hypotheses. The first hypothesis (H1), which posited a direct link between economic deprivation and increased feelings of loneliness, was supported by the significant direct effect observed in the SEM analysis ( β = 0.06, p < 0.001). This indicates that economic deprivation, in and of itself, is a direct antecedent of loneliness, potentially through mechanisms like heightened stress, stigma, or a sense of relative disadvantage. The second hypothesis (H2), proposing an indirect pathway where economic deprivation exacerbates loneliness by reducing social support (operationalized as social support in the SEM), was also clearly supported. The SEM results demonstrated that socioeconomic deprivation significantly reduced the social support ( β = -0.08, p < 0.001), and this diminished social support, in turn, significantly exacerbated feelings of loneliness ( β = -0.88, p < 0.001). The significant standardized indirect effect (0.07, p < 0.001) confirms that reduced social support is a key mediating factor in the relationship between economic deprivation and loneliness. Given that SHAP analysis has revealed the high importance of indicators reflecting the quality of emotional support, such as ‘frequency of confiding in others’ for predicting loneliness, the mediating effect of ‘reduced social support’ observed in the SEM model may also profoundly reflect the impact of economic deprivation on these high-quality social connections and deep emotional support resources, beyond just a reduction in opportunities for social participation. This pathway underscores how economic deprivation can systematically erode an individual’s access to, and the quality of, their social support networks by limiting participation in social activities, as outlined in the EDSL model’s conceptualization. 4.2 Consistency with and Extension of Prior Research These findings resonate with and build upon existing literature. The observed direct link between economic deprivation and loneliness aligns with previous studies establishing strong correlations between these two constructs ( 16 , 17 ). Similarly, the critical role of social support, or its absence, in loneliness is well-documented ( 18 , 19 ). While existing comprehensive studies have identified significant correlations between economic deprivation, social support, and loneliness ( 16 , 20 ), they have often not fully elucidated the specific mechanisms through which these factors interact. Building on this foundational research, the present study introduced and tested the EDSL Model. The EDSL model’s first proposition—a direct link from economic deprivation to loneliness—finds support in research detailing the psychological toll of financial hardship. Studies have shown that the chronic stress, feelings of shame, or perceived social stigma associated with poverty can directly foster social withdrawal and negative self-perceptions, thereby increasing vulnerability to loneliness, independent of quantifiable social interactions ( 21 ). Our findings empirically substantiate this direct adverse impact within a broader relational model. The EDSL model particularly emphasizes that economic deprivation indirectly exacerbates loneliness by diminishing social support. Consistent with previous studies, economic barriers have greatly restricted people’s participation in various social activities, hobbies and community activities, which are crucial for the formation and maintenance of social relationships ( 22 ). The EDSL model posits that this reduced participation is a key mechanism through which social support networks are eroded. This extends the work of researchers like Bai et al. (2021) and Zhao & Wu (2022), who have linked facets of social capital and support to loneliness, by explicitly modeling economic deprivation as a critical upstream driver of these support deficits. For example, where Zhao & Wu (2022) showed social support mediating the link between social engagement and loneliness, our model positions economic deprivation as a key factor limiting that initial social engagement and subsequent support. It is noteworthy that while our core theory posits that social support alleviates loneliness, the data reveal an additional negative correlation between visit frequency and loneliness. This may indicate that visit frequency not only reflects a component of perceived social support but also represents other protective machines. The impact of this behavioral dimension of social integration may not be fully captured by our social support latent variable, which places greater emphasis on emotional connections or membership identity. Just as ( 6 ) also found that support from friends has an exceptionally prominent protective effect. Friends’ visit may have additional and prominent protection mechanisms. Face-to-face visits offer genuine and perceptible social interaction and companionship. The lack of high-quality social interaction is at the core of loneliness ( 23 ). This regular visiting behavior directly meets human beings’ basic needs for a sense of belonging and connection ( 24 ) and directly reduce the sense of social isolation and the feeling of being ignored for one’s own value. Our results lend empirical weight to this indirect pathway, demonstrating that the erosion of social support (manifested as reduced social support in our SEM) acts as a significant mediator. Notably, the results indicate that this synergistic, indirect effect, where economic deprivation compromises social support and thereby amplifies loneliness, is a substantial contributor to the overall burden of loneliness. The magnitude of the standardized indirect effect (0.07) in relation to the direct effect (0.06) in our study underscores its critical importance. This study therefore extends prior work by not just confirming associations but by providing empirical evidence for specific, theoretically grounded mechanisms detailing how economic hardship translates into increased loneliness via the crucial mediating role of diminished social support. It moves beyond a general understanding of correlations (e.g.,( 25 , 26 ) by testing an integrated model with clearly defined direct and indirect effects within a large-scale cohort. 4.3 Implications and Future Directions The findings have significant theoretical and practical implications. Theoretically, this study underscores the necessity of adopting integrated models like the EDSL Model to understand multifaceted public health concerns such as loneliness. It provides strong empirical backing for the idea that social and economic structures are deeply intertwined in shaping emotional well-being. Practically, the dual pathways identified suggest that interventions to combat loneliness must be multi-pronged. Addressing economic deprivation directly through economic policies, strategies aimed at reducing economic deprivation, and support for those facing financial hardship could alleviate one direct source of loneliness. Simultaneously, interventions aimed at bolstering social support and facilitating social participation are crucial, especially for individuals in economically precarious situations. This could include funding for community programs, creating accessible social spaces, and supporting initiatives that reduce the financial barriers to social engagement. Such interventions are consistent with those described by( 27 ). Future research should aim to replicate these findings in diverse populations and cultural contexts. Longitudinal studies are essential to further elucidate the causal dynamics and temporal precedence of the relationships within the EDSL Model. Investigating other potential mediators (e.g., mental health status, perceived social status) and moderators (e.g., personality traits, community resources) could provide a more comprehensive understanding. Furthermore, exploring the specific types of social support (emotional, instrumental, informational) most affected by economic deprivation and most protective against loneliness would be a valuable avenue for research. Intervention studies based on the EDSL Model are also needed to test the real-world effectiveness of strategies targeting these dual pathways. 4.4 Limitations However, certain limitations should be acknowledged. First, while the UK Biobank is a powerful resource, its participants are aged 40–69 at baseline, so findings may not generalize to younger or much older populations without further research. Second, loneliness was measured using a single dichotomous question (“Do you often feel lonely?”); while widely used, this may not capture the full spectrum or intensity of loneliness compared to multi-item scales. Third, although the EDSL model posits causal pathways, the cross-sectional nature of the specific analyses conducted for the SEM limits the ability to definitively infer causality; longitudinal panel data analysis would be beneficial to further substantiate these causal claims. Finally, while social support was the term used in the SEM diagram for the mediator, the underlying construct of social support was measured by three distinct indicators; future research could explore how different facets of social support are uniquely impacted by economic deprivation and contribute to loneliness. 5. Conclusion This study provides evidence for the EDSL Model, demonstrating that economic deprivation directly fosters loneliness and indirectly amplifies it by diminishing social support, reflected in an overall reduction of support metrics. These findings highlight the profound impact of economic hardship on social well-being and call for integrated strategies that address both material needs and social connectedness to effectively combat the global public health challenge of loneliness. Abbreviations AUC: Area Under the Curve β : Standardized regression coefficient CI: Confidence Interval EDSL: Economic Deprivation, Support Diminution, and Loneliness GFI: Goodness of Fit Index IMD: Index of Multiple Deprivation KNN: K-nearest Neighbors LR: Logistic Regression NB: Naive Bayes RF: Random Forest RMR: Root Mean Square Residual ROC: Receiver Operating Characteristic SE: Standard Error SEM: Structural Equation Modeling SHAP: Shapley Additive Explanations SRMR: Standardized Root Mean Square Residual TDI: Townsend Deprivation Index UKB: UK Biobank VIF: Variance Inflation Factor WLSMV: Weighted Least Squares Mean and Variance Adjusted XGBoost: Extreme Gradient Boosting Declarations Ethics approval and consent to participate Ethical approval for the UK Biobank was granted by the National Health Service Research Ethics Service (approval date: June 17, 2011; reference: 11/NW/0382). All participants provided informed consent at enrollment. Access to and analysis of the data for this study were authorized by the UK Biobank under Application Number 37292. Consent for publication Not applicable. Availability of data and materials The dataset(s) analyzed during the current study are available in the UK Biobank repository (https://www.ukbiobank.ac.uk/), under Application Number 37292. Derived datasets generated for this study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests. Funding Not applicable. Authors’ contributions A.K. wrote the main manuscript text and analyzed the data. Z.Y.W., F.L., J.N., J.D., and P.Z. contributed to data analysis and preparation, while the other authors assisted with additional analyses and preparation, and reviewed and supervised the study. All authors read and approved the final manuscript. Acknowledgements We are grateful to the UK Biobank participants and coordinators for providing the data resource that made this study possible. References Wang F, Gao Y, Han Z, Yu Y, Long Z, Jiang X, et al. A systematic review and meta-analysis of 90 cohort studies of social isolation, loneliness and mortality. Nat Hum Behav. 2023 June 19;7(8):1307–19. Cacioppo JT, Cacioppo S. The growing problem of loneliness. The Lancet. 2018 Feb 3;391(10119):426. Kung CSJ, Pudney SE, Shields MA. Economic gradients in loneliness, social isolation and social support: Evidence from the UK Biobank. Social Science & Medicine. 2022 Aug;306:115122. Victor CR, Pikhartova J. Lonely places or lonely people? Investigating the relationship between loneliness and place of residence. BMC Public Health. 2020 Dec;20(1):778. Mao S, Lou VWQ, Lu N. Perceptions of neighborhood environment and loneliness among older Chinese adults: the mediator role of cognitive and structural social capital. Aging & Mental Health. 2023 Mar 4;27(3):595–603. Zhang X, Dong S. The relationships between social support and loneliness: A meta-analysis and review. Acta Psychologica. 2022 July;227:103616. Gamper M. Social Network Theories: An Overview. In: Klärner A, Gamper M, Keim-Klärner S, Moor I, von der Lippe H, Vonneilich N, editors. Social Networks and Health Inequalities: A New Perspective for Research [Internet]. Cham: Springer International Publishing; 2022. p. 35–48. Available from: https://doi.org/10.1007/978-3-030-97722-1_3 Holt-Lunstad J. Social connection as a public health issue: the evidence and a systemic framework for prioritizing the “social” in social determinants of health. Annual Review of Public Health. 2022;43(1):193–213. Peng J, Qi H, Fan Z, Zhou Q, Lin Y. Social support and health behaviors of older adults during the COVID-19 pandemic in China: a moderated mediation model of loneliness and economic income. BMC Public Health. 2024 Oct 11;24(1):2780. Stephens C, Alpass F, Towers A. Economic hardship among older people in New Zealand: the effects of low living standards on social support loneliness and mental health. 2010; Eckhard J. Does Poverty Increase the Risk of Social Isolation? Insights Based on Panel Data from Germany. The Sociological Quarterly. 2018 Apr 3;59(2):338–59. Tang VFY, Chou KL. An exploratory study on material deprivation and loneliness among older adults in Hong Kong. BMC Geriatr. 2024 May 6;24(1):400. Bai Z, Wang Z, Shao T, Qin X, Hu Z. Association between social capital and loneliness among older adults: a cross-sectional study in Anhui Province, China. BMC Geriatr. 2021 Dec;21(1):26. Zhao L, Wu L. The Association between Social Participation and Loneliness of the Chinese Older Adults over Time—The Mediating Effect of Social Support. IJERPH. 2022 Jan 12;19(2):815. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2012 Jan 2;12. Algren MH, Ekholm O, Nielsen L, Ersbøll AK, Bak CK, Andersen PT. Social isolation, loneliness, socioeconomic status, and health-risk behaviour in deprived neighbourhoods in Denmark: A cross-sectional study. SSM - Population Health. 2020 Apr;10:100546. Davis AJ, Cohen E, Nettle D. Associations amongst poverty, loneliness, and a defensive symptom cluster characterised by pain, fatigue, and low mood. Public Health. 2025 May;242:272–7. Czaja SJ, Moxley JH, Rogers WA. Social Support, Isolation, Loneliness, and Health Among Older Adults in the PRISM Randomized Controlled Trial. Front Psychol. 2021 Oct 5;12:728658. Wang J, Mann F, Lloyd-Evans B, Ma R, Johnson S. Associations between loneliness and perceived social support and outcomes of mental health problems: a systematic review. BMC Psychiatry. 2018 Dec;18(1):156. Menec VH, Newall NE, Mackenzie CS, Shooshtari S, Nowicki S. Examining social isolation and loneliness in combination in relation to social support and psychological distress using Canadian Longitudinal Study of Aging (CLSA) data. Reppermund S, editor. PLoS ONE. 2020 Mar 23;15(3):e0230673. Inglis G, Sosu E, McHardy F, Witteveen I, Jenkins P, Knifton L. Testing the associations between poverty stigma and mental health: The role of received stigma and perceived structural stigma. International Journal of Social Psychiatry. 2024;00207640241296055. Wu J, Zhang J, Fokkema T. The micro-macro interplay of economic factors in late-life loneliness: Evidence from Europe and China. Front Public Health. 2022 Sept 13;10:968411. Hawkley LC, Cacioppo JT. Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms. ann behav med. 2010 Oct;40(2):218–27. Baumeister RF, Leary MR. The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin. 1995;117(3):497–529. Creed PA, Reynolds J. Economic deprivation, experiential deprivation and social loneliness in unemployed and employed youth. Community & Applied Soc Psy. 2001 May;11(3):167–78. Golden J, Conroy RM, Bruce I, Denihan A, Greene E, Kirby M, et al. Loneliness, social support networks, mood and wellbeing in community‐dwelling elderly. Int J Geriat Psychiatry. 2009 July;24(7):694–700. O’Rourke HM, Collins L, Sidani S. Interventions to address social connectedness and loneliness for older adults: a scoping review. BMC Geriatr. 2018 Dec;18(1):214. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 07 Oct, 2025 Editor invited by journal 08 Sep, 2025 Editor assigned by journal 08 Sep, 2025 Submission checks completed at journal 08 Sep, 2025 First submitted to journal 06 Sep, 2025 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-7553119","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":531741077,"identity":"cd95f627-95b4-4878-bfbc-264a75d9ecab","order_by":0,"name":"Anhui Kong","email":"","orcid":"","institution":"Zhejiang Sci-Tech University","correspondingAuthor":false,"prefix":"","firstName":"Anhui","middleName":"","lastName":"Kong","suffix":""},{"id":531741078,"identity":"7c7b5a21-51ef-4976-8795-9db586359ad8","order_by":1,"name":"Ziyue Wu","email":"","orcid":"","institution":"Beijing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Ziyue","middleName":"","lastName":"Wu","suffix":""},{"id":531741079,"identity":"172fe1c5-836a-48ab-b565-7202646e6b5d","order_by":2,"name":"Fangqing Liu","email":"","orcid":"","institution":"University of Manchester","correspondingAuthor":false,"prefix":"","firstName":"Fangqing","middleName":"","lastName":"Liu","suffix":""},{"id":531741080,"identity":"e9f72745-ff18-40d5-a3d7-b01b717ac429","order_by":3,"name":"Jinxuan Ni","email":"","orcid":"","institution":"Xi’an Jiaotong-Liverpool University","correspondingAuthor":false,"prefix":"","firstName":"Jinxuan","middleName":"","lastName":"Ni","suffix":""},{"id":531741081,"identity":"e89811a1-1c83-4db9-a5f9-43ebb7cc5fea","order_by":4,"name":"JiaXing Deng","email":"","orcid":"","institution":"Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"JiaXing","middleName":"","lastName":"Deng","suffix":""},{"id":531741082,"identity":"7b368b76-60dd-446b-a858-fb7fa64d849a","order_by":5,"name":"Pengyi Zhang","email":"","orcid":"","institution":"Henan University","correspondingAuthor":false,"prefix":"","firstName":"Pengyi","middleName":"","lastName":"Zhang","suffix":""},{"id":531741083,"identity":"ad98d9da-7ae9-47af-a8e8-e1b61957a26c","order_by":6,"name":"Siji Huang","email":"","orcid":"","institution":"Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"Siji","middleName":"","lastName":"Huang","suffix":""},{"id":531741084,"identity":"874bfea2-d04a-40ff-8ecf-edb30065db76","order_by":7,"name":"Yujie Hu","email":"","orcid":"","institution":"Renmin University of China","correspondingAuthor":false,"prefix":"","firstName":"Yujie","middleName":"","lastName":"Hu","suffix":""},{"id":531741085,"identity":"8f2dfcdc-ff2f-431c-9122-0131137f02d6","order_by":8,"name":"Zige Wang","email":"","orcid":"","institution":"The Hong Kong Polytechnic University","correspondingAuthor":false,"prefix":"","firstName":"Zige","middleName":"","lastName":"Wang","suffix":""},{"id":531741086,"identity":"f07edd28-84d7-4f5d-ae24-bec49e070978","order_by":9,"name":"Xinyu Yu","email":"","orcid":"","institution":"Beijing Normal-Hong Kong Baptist University","correspondingAuthor":false,"prefix":"","firstName":"Xinyu","middleName":"","lastName":"Yu","suffix":""},{"id":531741087,"identity":"6b4f41eb-933c-48ab-9fb9-86d4185bccaf","order_by":10,"name":"Yujie Li","email":"","orcid":"","institution":"Jiangxi Normal University","correspondingAuthor":false,"prefix":"","firstName":"Yujie","middleName":"","lastName":"Li","suffix":""},{"id":531741088,"identity":"1dfcaf1b-00bf-4d8a-bb81-50106e3717d8","order_by":11,"name":"Xiaoyu Hu","email":"","orcid":"","institution":"Catholic University of the Sacred Heart","correspondingAuthor":false,"prefix":"","firstName":"Xiaoyu","middleName":"","lastName":"Hu","suffix":""},{"id":531741089,"identity":"2bee5fc3-bb0d-4d54-872c-76e454ab97a7","order_by":12,"name":"Hui Zhu","email":"","orcid":"","institution":"Xi’an Jiaotong-Liverpool University","correspondingAuthor":false,"prefix":"","firstName":"Hui","middleName":"","lastName":"Zhu","suffix":""},{"id":531741090,"identity":"cbcfe7f0-49a3-41b8-8915-bd921f8a962f","order_by":13,"name":"Bohan Li","email":"","orcid":"","institution":"Hainan Normal University","correspondingAuthor":false,"prefix":"","firstName":"Bohan","middleName":"","lastName":"Li","suffix":""},{"id":531741091,"identity":"7c965f44-dc92-43a3-a138-759386b3eb5a","order_by":14,"name":"Zheng Jiang","email":"","orcid":"","institution":"City University of Hong Kong","correspondingAuthor":false,"prefix":"","firstName":"Zheng","middleName":"","lastName":"Jiang","suffix":""},{"id":531741092,"identity":"6a9983d1-c00b-40d5-b78c-8be5dc304cfc","order_by":15,"name":"Zhenhong He","email":"","orcid":"","institution":"Shenzhen University","correspondingAuthor":false,"prefix":"","firstName":"Zhenhong","middleName":"","lastName":"He","suffix":""},{"id":531741093,"identity":"a109956b-e837-4f7b-8101-aa4f1c13492a","order_by":16,"name":"Nils Muhlert","email":"","orcid":"","institution":"The University of Manchester","correspondingAuthor":false,"prefix":"","firstName":"Nils","middleName":"","lastName":"Muhlert","suffix":""},{"id":531741094,"identity":"4987667b-792d-4267-9e9d-ff43565a33f9","order_by":17,"name":"Yunkai Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYFACNgaGBAMJOSCL8QCQSbwWYwYGZgYStABBYgPRWgxupCV+eFBgkd4vkX/gwIeKNAb+9m78+oBaDksAHZY7c0Yyw8EZZ3IYJM6c3UBAS3oDWMuGG8kMh3nbKhhAbEJamn8AtaQbkKAl7RjIlgSolhzCWiTPPEuzAGoxnNnz2ADolzQegn7hO55mfPPHnzp5fvbEhw8+VCTL8bf34teicCEBVYAHr3IQkO8/QFDNKBgFo2AUjHQAABjiSzLowMJ4AAAAAElFTkSuQmCC","orcid":"","institution":"Shenzhen University","correspondingAuthor":true,"prefix":"","firstName":"Yunkai","middleName":"","lastName":"Yang","suffix":""},{"id":531741095,"identity":"fa8fbb24-f557-4cf0-bdb3-20cb41b3f6e1","order_by":18,"name":"Rebecca Elliott","email":"","orcid":"","institution":"University of Manchester","correspondingAuthor":false,"prefix":"","firstName":"Rebecca","middleName":"","lastName":"Elliott","suffix":""}],"badges":[],"createdAt":"2025-09-06 22:08:01","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7553119/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7553119/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":93938003,"identity":"13bc6621-e349-4472-9247-18ce790b3e59","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":486349,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/5d986bffce1fb3a0bfc6ee5f.docx"},{"id":93938001,"identity":"1fb49c24-0146-47a3-8695-70cba33fb5ff","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17442,"visible":true,"origin":"","legend":"","description":"","filename":"a8116f93629b4de1b4eb4145e59f13e3.json","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/aa3d78d2d07d7d8c4d977e87.json"},{"id":93938376,"identity":"127adb10-6a5e-4e6b-82a6-7cd441f1d948","added_by":"auto","created_at":"2025-10-20 13:21:26","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":101929,"visible":true,"origin":"","legend":"","description":"","filename":"a8116f93629b4de1b4eb4145e59f13e31enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/c9476238172cf04e934335f9.xml"},{"id":93940509,"identity":"a0e20724-dcc0-43b8-8168-d22752ef3141","added_by":"auto","created_at":"2025-10-20 13:37:26","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":114159,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/9f858b7d21f6380f5c7955b1.png"},{"id":93942057,"identity":"084973d6-107f-49bb-bf94-f33096a722d8","added_by":"auto","created_at":"2025-10-20 13:45:26","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":77600,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/a8d1df2d845aa356a3226627.png"},{"id":93938011,"identity":"850aabe1-28ba-4c30-8114-1239c799f859","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":72599,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/f6b78a665bc6ddd691db8ab1.png"},{"id":93938014,"identity":"8bb0f383-d27f-477a-b938-0cc9d597ba24","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":117011,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/946941c959e00d23248afbb5.png"},{"id":93938380,"identity":"7d44e2f9-1914-4f6d-a713-e0c3d43e3551","added_by":"auto","created_at":"2025-10-20 13:21:26","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28145,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/d67d005373492101ad8505cb.png"},{"id":93938004,"identity":"79fb953e-b798-481b-b76c-4970231f6242","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28196,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/c630d82408b3b6b4dc81d3c7.png"},{"id":93938015,"identity":"1ccf8443-bae6-4dd5-a8a3-29335f25e28d","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23450,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/d1ab75c394c26f2398ac90bf.png"},{"id":93938012,"identity":"b710ce4d-1f17-47f3-8d2f-b4bd35df4fab","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22508,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/9a3b889d862386f3a996c5a0.png"},{"id":93938016,"identity":"2dee9798-6fb6-40c9-b906-48c69f92ab38","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":32100,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/a70c916ba5a9e35366042258.png"},{"id":93940098,"identity":"d52e3e9a-1086-45d1-abe3-4a01e594904a","added_by":"auto","created_at":"2025-10-20 13:29:26","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10694,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/9218f969a8f5915691a4eac7.png"},{"id":93938018,"identity":"b6d7eb71-8a32-40d5-9756-578c5730ce6b","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":100372,"visible":true,"origin":"","legend":"","description":"","filename":"a8116f93629b4de1b4eb4145e59f13e31structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/0a3991baa98444fa97cc5fa6.xml"},{"id":93938381,"identity":"9844cebc-c32b-4fa3-b56e-5a16dac0c2e4","added_by":"auto","created_at":"2025-10-20 13:21:26","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":113200,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/052fabaa48eb255359493de8.html"},{"id":93938374,"identity":"00005838-4eb2-4d0d-aeab-871c9e845ecd","added_by":"auto","created_at":"2025-10-20 13:21:26","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":130296,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConceptual model of the relationship between economic deprivation, social support, and loneliness.\u003c/strong\u003e The model illustrates two hypothesized pathways: (H1) Economic deprivation is associated with increased loneliness; (H2) Economic deprivation contributes to loneliness indirectly by reducing the social support (social support).\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/a42d3d6fba5444a6f92ac96e.png"},{"id":93937998,"identity":"aeea8474-a9ff-4f3e-9c27-f76b0da70932","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":87925,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDescriptive statistics and correlational patterns of predictors of loneliness.\u003c/strong\u003e (A) Boxplots compare lonely and non-lonely individuals across six variables: Townsend Deprivation Index (TDI), Index of Multiple Deprivation (IMD), visit frequency, confiding frequency, and household size. (B) Results of variance inflation factor (VIF). All VIFs are lower than 5 (yellow dotted line). (C) Correlation matrix presents pairwise associations among variables, with color intensity representing the strength and direction of correlations. Note: ***: \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/b1758eb7ff6a8149fa9aa5c9.png"},{"id":93937999,"identity":"a1ec62d3-e8f0-4e52-9316-62e5d3e560f0","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":86719,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredictive performance of machine learning models for loneliness classification.\u003c/strong\u003e\u003cbr\u003e\n(A) Receiver operating characteristic (ROC) curves for six classifiers: random forest (RF), logistic regression (LR), naive Bayes (NB), decision tree (DT), k-nearest neighbors (KNN), and extreme gradient boosting (XGBoost). The XGBoost model yielded the highest area under the curve (AUC = 81.35). (B) Confusion matrix of the best-performing model (XGBoost) showing true and predicted classifications of loneliness.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/d25c6298fe8b1051f12f7990.png"},{"id":93938375,"identity":"66a14d1a-0567-403d-948f-fe1721a0cb64","added_by":"auto","created_at":"2025-10-20 13:21:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":128279,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSHAP analysis of feature contributions to model predictions of loneliness. \u003c/strong\u003e(A) Mean SHAP values indicating the average importance of each predictor variable in the XGBoost model. Confiding frequency exerted the strongest influence. (B) SHAP summary plot visualizing the distribution of SHAP values by feature and value magnitude. Blue and pink indicate lower and higher original feature values, respectively.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/edc3490dff064c0ac5e37db1.png"},{"id":93938005,"identity":"e46b963c-2903-4292-a20e-884aa17f4b8f","added_by":"auto","created_at":"2025-10-20 13:13:26","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":36460,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMediation model testing the indirect effect of economic deprivation on loneliness via social support.\u003c/strong\u003e Path analysis results show that economic deprivation is negatively associated with social support (\u003cem\u003eβ\u003c/em\u003e= −0.08, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001), and social support is negatively associated with loneliness (\u003cem\u003eβ\u003c/em\u003e = −0.88, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001). Economic deprivation also has a significant direct effect on loneliness (\u003cem\u003eβ\u003c/em\u003e = 0.06, \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001), indicating partial mediation.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/c8bba249fc7da26621c8d247.png"},{"id":93958442,"identity":"df76d9a5-e2b2-412c-9a60-8f573c8198fd","added_by":"auto","created_at":"2025-10-20 16:39:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1387807,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7553119/v1/5b900dfa-7792-448a-80e9-2a453c0f8bcd.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Economic Deprivation, Diminished Social Support, and Loneliness: An Empirical Test of the EDSL Model Using UK Biobank Data","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eLoneliness, defined by the discrepancy between desired and actual social connections, has been identified as a significant global public health concern with impacts on diverse populations worldwide (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). Current research has found that the important factors causing loneliness are far more than individual psychological factors, but are closely related to the entire social structure, resource allocation and other factors (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Effectively addressing loneliness necessitates a comprehensive understanding of the complex interplay between social and economic structures that shape the formation, maintenance, and quality of social relationships.\u003c/p\u003e\u003cp\u003eHowever, previous research has often investigated these determinants in isolation. For instance, numerous studies have established strong correlations between higher economic deprivation and increased loneliness (e.g., (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Concurrently, a significant body of evidence underscores the protective role of social support (e.g., (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). These studies often fail to elucidate the synergistic effects and potential effects pathways\u0026mdash;specifically, how economic deprivation might systematically erode social support, thereby exacerbating loneliness. Even foundational theoretical frameworks, such as social network theory (SNT), which emphasizes interpersonal relationship structures (7), and the social determinants of health framework (SDOH), which highlights broader socioeconomic conditions (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e), have tended to maintain distinct focuses\u0026mdash;SNT on relational dynamics and SDOH on structural inequalities. Consequently, explicit theoretical and empirical examination of how structural economic factors transact with social relational factors to jointly produce loneliness has been less common, limiting the development of holistic theoretical models and comprehensive intervention strategies.\u003c/p\u003e\u003cp\u003eThis fragmentation underscores a critical gap: while an integrative perspective arguing that economic deprivation restricts social participation and deepens isolation is gaining traction (e.g. (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e), the precise mechanisms and empirical support for comprehensive models detailing these interactions, particularly using large-scale datasets, are still developing. To address this gap, this study introduces and empirically tests an Economic Deprivation, Support Diminution, and Loneliness (EDSL) Model. The EDSL Model is predicated on the understanding that economic deprivation is often accompanied by the limitation of the sources of social support\u0026mdash;a process where individuals are marginalized due to a lack of resources or opportunities, preventing normal participation in community social activities. By definition, economic deprivation can mean the inability to afford activities that promote social integration. As classic economic deprivation researchers have noted, individuals experiencing economic deprivation are often unable to participate in ordinary social outings, hobbies, celebrations, or civic activities, because even seemingly minor expenses (such as a cup of coffee with a friend, bus fare to visit relatives, a birthday gift, or a club membership fee) can become prohibitive (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). This economic deprivation, therefore, can lead to the impairment of social support pathways, further amplifying loneliness. The EDSL Model posits that economic deprivation impacts loneliness through two primary pathways: H1) there is a direct pathway where economic deprivation; H2) Economic deprivation indirectly exacerbates loneliness by causing a diminution of social support. This diminution occurs as the aforementioned processes of exclusion stemming from economic deprivation systematically erode an individual\u0026rsquo;s access to, and the quality of, their social support networks. This diminished social support then acts as a key mediating factor that further amplifies the experience of loneliness. The concept map of this model is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eA growing body of empirical research lends preliminary support to the EDSL Model\u0026rsquo;s tenets. For instance, longitudinal studies demonstrate that material deprivation significantly predicts higher loneliness levels even two years later, an effect partially mediated by reduced engagement in cultural activities (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Concurrently, research correlates lower social capital with increased loneliness, found that lower levels of social participation, social connection, and reciprocity were significantly associated with higher odds of loneliness in older adults (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Furthermore, studies employing structural equation modeling (SEM) or similar path analyses are beginning to unravel the complex interplay between socioeconomic factors, social resources, and loneliness-related outcomes. Research by (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e) showed that social support mediated the relationship between social engagement (itself often linked to socioeconomic opportunities) and loneliness. While these studies are valuable, they often do not comprehensively test an integrated model that explicitly delineates both the direct impact of various economic deprivation indicators on loneliness and the concurrent mediating role of multifaceted social support measures (reflecting support diminution) within a large-scale cohort, as proposed by the EDSL Model. This study seeks to build upon this emerging evidence by providing a more robust and nuanced empirical validation of the EDSL Model.\u003c/p\u003e\u003cp\u003eThe empirical basis for this analysis is data from the UK Biobank. Economic deprivation is combined the Townsend Deprivation Index (TDI) and the Index of Multiple Deprivation (IMD). This dual approach allows a nuanced examination of economic deprivation\u0026rsquo;s facets. Social support is operationalized using established indicators: \u0026ldquo;Total number of household members\u0026rdquo; (Household Size), \u0026ldquo;Frequency of visits with friends or family\u0026rdquo; (Visit Frequency), and \u0026ldquo;Frequency of confiding in others\u0026rdquo; (Confiding Frequency). Methodologically, this research first employs machine learning to determine the predictive effect of these factors on loneliness, using Shapley Additive Explanations (SHAP) values to ascertain influence direction. Subsequently, these insights inform a SEM to rigorously test the hypothesized direct effects of economic deprivation on loneliness and the indirect effects mediated by social support indicators within the EDSL. This study aims to provide robust empirical evidence for the EDSL, thereby enhancing our understanding of the interconnected socioeconomic and social-relational drivers of loneliness\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Participants\u003c/h2\u003e\u003cp\u003eThis study utilized data from the UK Biobank, employing the data release dated September 17, 2019. The UK Biobank is a large-scale, prospective cohort study that enrolled over 500,000 participants aged 40 to 69 years at baseline (recruitment period: 2007\u0026ndash;2010) across England, Scotland, and Wales. For the present analysis, a total of 232,477 participants were included to investigate the relationship between loneliness and a range of independent variables. Participants were bifurcated into two groups based on self-reported loneliness status: the \u0026ldquo;loneliness group\u0026rdquo; (n\u0026thinsp;=\u0026thinsp;73,884) and the \u0026ldquo;non-loneliness group\u0026rdquo; (n\u0026thinsp;=\u0026thinsp;158,593). Ethical approval for the UK Biobank was granted by the National Health Service Research Ethics Service (approval date: June 17, 2011; reference: 11/NW/0382). Access to and analysis of the data for this study were authorized by the UK Biobank under Application Number 37292.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Quality control and descriptive statistics\u003c/h2\u003e\u003cp\u003eFirstly, we calculated the variance inflation factor (VIF) for each independent variable. A VIF value below 5 was considered acceptable, indicating no severe multicollinearity. And we calculated the correlation coefficients between the variables to ensure that there was no excessive correlation affecting the model performance (\u003cem\u003er\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.7). Subsequently, we conducted descriptive statistics on the two sets of data and used chi-square tests and independent sample t-tests and chi-square test to examine whether there were significant differences in other variables between the two groups.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Measurement\u003c/h2\u003e\u003cp\u003eData for this study were sourced from the UK Biobank database, encompassing variables across demographic, social support, and socioeconomic dimensions. The sole demographic variable was participant self-reported Gender, coded dichotomously (1\u0026thinsp;=\u0026thinsp;Male, 0\u0026thinsp;=\u0026thinsp;Female). Indicators of social support included the count variable Household Size, Visit Frequency, and Confiding Frequency; for the latter two categorical variables, higher scores signify greater frequency. Socioeconomic deprivation was assessed using two area-based indices linked via participant postcode: the Townsend Deprivation Index (TDI), a continuous variable calculated from census data on unemployment, car non-ownership, non-home ownership, and household overcrowding; and the Index of Multiple Deprivation (IMD), a composite continuous score or rank integrating indicators across domains such as income, employment, health, education, housing, crime, and environment. For both TDI and IMD, higher scores or ranks denote greater levels of deprivation. The dependent variable, Loneliness, was measured using the question \u0026ldquo;Do you often feel lonely?\u0026rdquo; and operationalized as a dichotomous variable, with \u0026lsquo;Yes\u0026rsquo; responses coded as 1 and \u0026lsquo;No\u0026rsquo; responses coded as 0; only participants providing these definitive answers were included in the final analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Machine Learning\u003c/h2\u003e\u003cp\u003eTo evaluate the association between loneliness and the independent variables, six machine learning algorithms were utilized: Logistic Regression (LR), Naive Bayes (NB), Decision Tree (DT), Random Forest (RF), K-nearest Neighbors (KNN), and Extreme Gradient Boosting (XGBoost). These algorithms were selected to provide a comprehensive assessment of the models\u0026rsquo; capacity to differentiate between lonely and non-lonely individuals, to balance precision and recall, and to minimize overfitting. To ensure methodological consistency across all machine learning approaches, the dataset was partitioned into training and testing subsets, with the testing set comprising 20% of the total data. This 80/20 split is a conventional practice in machine learning, aiming to balance the requirement for adequate training data with the need for a robust evaluation set. Implementation of these algorithms was performed using the scikit-learn library (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). The performance of each algorithm was assessed using a suite of metrics, including area under the curve (AUC), accuracy, precision, recall, and the F1-score.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 SHAP Value\u003c/h2\u003e\u003cp\u003eTo improve the interpretability of the machine learning models, the SHAP methodology was utilized. SHAP provides a unified methodology for interpreting model outputs by quantifying the contribution of each feature to individual predictions. In this analysis, SHAP values were computed for the XGBoost model, which exhibited the optimal performance in predicting loneliness. In this study, we determined a preliminary direction for the prediction of loneliness by each variable through SHAP value.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.6 SEM analysis\u003c/h2\u003e\u003cp\u003eWe used the SEM to verify the impact of economic deprivation on loneliness and that economic deprivation further enhances loneliness by depriving social support. Therefore, we take economic deprivation as the independent variable, social support as the mediating variable, and loneliness as the dependent variable. Latent variables with only two indicator variables can be identified under certain conditions, but their estimates may not be stable enough and are more susceptible to incorrect Settings of sample characteristics and other parts of the model. Therefore, we take the total scores of TDI and IMD as the direct indicators of economic deprivation. Construct latent variables of social support through three indicators.\u003c/p\u003e\u003cp\u003eThe model was estimated using Weighted Least Squares Mean and Variance Adjusted (WLSMV), which is suitable for the classified variable. To assess model fit, we employed several widely used fit indices, including Goodness of Fit Index (GFI), Root Mean Square Residual (RMR) and Standardized Root Mean Square Residual (SRMR). A good model fit is indicated by GFI value above 0.9, RMR and SRMR values below 0.08. After the initial model fitting did not achieve satisfactory fit, we examined modification indices (MI) to identify potential areas of misspecification. A covariance between Loneliness and Visit Frequency (Loneliness\u0026thinsp;~\u0026thinsp;~\u0026thinsp;Visit Frequency), corresponding to the largest MI, was subsequently added to the model. This modification was chosen to improve model fit while also carefully ensure model convergence.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Quality control and descriptive statistics\u003c/h2\u003e\u003cp\u003eFirst of all, we ensured that there was no serious multicollinearity problem among the various variables (VIF\u0026thinsp;\u0026lt;\u0026thinsp;5, \u003cem\u003er\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.7, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB, C). The independent sample t-test and chi-square test revealed significant intergroup differences in various variables between the lonely group and the non-lonely group. The TDI of the lonely group was significantly higher than that of the non-lonely group (\u003cem\u003et\u003c/em\u003e\u0026thinsp;=\u0026thinsp;18.13, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.08). The IMD was significantly higher in the lonely group than in the non-lonely group (\u003cem\u003et\u003c/em\u003e\u0026thinsp;=\u0026thinsp;61.50, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.26). Household Size was significantly smaller in the lonely group than in the non-lonely group (\u003cem\u003et\u003c/em\u003e = -51.58, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, \u003cem\u003ed\u003c/em\u003e = -0.21). Visit Frequency was significantly lower in the lonely group than in the non-lonely group (\u003cem\u003et\u003c/em\u003e = -34.17, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, \u003cem\u003ed\u003c/em\u003e = -0.14). The Confiding Frequency in the Lonely group was significantly lower than that in the non-lonely group (\u003cem\u003et\u003c/em\u003e = -132.11, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, \u003cem\u003ed\u003c/em\u003e = -0.55). There are significant differences in gender between the two groups (\u003cem\u003eχ\u003c/em\u003e\u0026sup2; = 18.13, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Descriptive statistics and independent sample t-test results are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\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\u003eDescriptive statistics and difference test results of the two groups\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eItems\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLonely group (M\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNon-Lonely Group (M\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003et\u003c/em\u003e\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTDI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-1.31\u0026thinsp;\u0026plusmn;\u0026thinsp;3.08\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-1.55\u0026thinsp;\u0026plusmn;\u0026thinsp;2.91\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18.13\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIMD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20.58\u0026thinsp;\u0026plusmn;\u0026thinsp;15.61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e16.73\u0026thinsp;\u0026plusmn;\u0026thinsp;13.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e61.50\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHousehold Size\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.18\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-51.58\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVisit Frequency\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.75\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-34.17\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eConfiding Frequency\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.68\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3.77\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-132.11\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLonely group (Percentage)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNot Lonely Group (Percentage)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eχ\u003c/em\u003e\u003csup\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGender (Number of males)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e27505 (37.22%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e75187 (47.41)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2118.74\u003csup\u003e***\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eNote: ***: \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\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\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Machine learning model performances\u003c/h2\u003e\u003cp\u003eTo assess the performance of various machine learning algorithms, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents key metrics, including AUC score, Accuracy, Precision, Recall, and F1 score. The AUC of all models was greater than the chance level (AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.5), demonstrating excellent predictive performance. Specifically, XGBoost demonstrated the best predictive performance (AUC\u0026thinsp;=\u0026thinsp;81.35%). This indicates that these factors can significantly predict loneliness. In Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, we present the ROC curves of all models. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB presents the confusion matrix for the best-performing XGBoost model.\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\u003e\u003cb\u003eAll machine learning model performances\u003c/b\u003e\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMetrics\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNB\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDT\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eRF\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eXGBoost\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAUC\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e70.29%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e69.66%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e79.75%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e80.82%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e76.91%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e81.35%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e66.69%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e69.37%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e77.87%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e75.65%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e75.66%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e75.74%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e48.13%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e53.05%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e72.84%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e60.46%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e64.60%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e60.37%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e61.89%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e31.58%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e48.45%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e67.61%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e51.81%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e68.88%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e54.15%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e39.59%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e58.19%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e63.84%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e57.50%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e64.35%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"7\" nameend=\"c7\" namest=\"c1\"\u003e\u003cp\u003eNote: LR: logistic regression; NB: naive bayes; DT: decision tree; RF: random forest; KNN: k-nearest neighbors; XGBoost: extreme gradient boosting.\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\u003c/p\u003e\u003cp\u003e(A) Receiver operating characteristic (ROC) curves for six classifiers: random forest (RF), logistic regression (LR), naive Bayes (NB), decision tree (DT), k-nearest neighbors (KNN), and extreme gradient boosting (XGBoost). The XGBoost model yielded the highest area under the curve (AUC\u0026thinsp;=\u0026thinsp;81.35). (B) Confusion matrix of the best-performing model (XGBoost) showing true and predicted classifications of loneliness.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.3 SHAP Analysis for Feature Importance\u003c/h2\u003e\u003cp\u003eTo elucidate the contribution of each feature to the model\u0026rsquo;s predictions of loneliness, a SHAP analysis was conducted on the best-performing XGBoost model. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA presents the mean absolute SHAP values, indicating the average impact of each predictor on the model output magnitude. Confiding Frequency exhibited the largest mean SHAP value, identifying it as the most influential predictor of loneliness. This was followed by IMD and Household Size, which also demonstrated substantial impacts. TDI and Gender showed moderate importance, while Visit Frequency had a comparatively smaller, yet still relevant, impact on the model\u0026rsquo;s predictions. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB provides a SHAP summary plot, visualizing the direction and distribution of each feature\u0026rsquo;s impact on the model output. The directionality of SHAP roughly shows the negative prediction of loneliness by social support factors and the positive prediction of loneliness by economic deprivation indicators.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.4 SEM Results\u003c/h2\u003e\u003cp\u003eThe corrected SEM model demonstrated excellent fit, as indicated by the following fit indices: GFI\u0026thinsp;=\u0026thinsp;0.99, RMR\u0026thinsp;=\u0026thinsp;0.05, SRMR\u0026thinsp;=\u0026thinsp;0.07. The analysis revealed a significant direct effect of socioeconomic deprivation on loneliness (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.06, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.004, \u003cem\u003ez\u003c/em\u003e\u0026thinsp;=\u0026thinsp;13.70, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, 95% CI: [0.05, 0.07]). In addition to the direct effect, the analysis identified a significant mediation pathway through which economic deprivation influences loneliness. Specifically, economic deprivation was found to reduce the support (\u003cem\u003eβ\u003c/em\u003e = -0.08, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.004, \u003cem\u003ez\u003c/em\u003e = -20.19, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, 95% CI: [-0.09, -0.07]). In turn, diminished social support significantly exacerbated feelings of loneliness (\u003cem\u003eβ\u003c/em\u003e = -0.88, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.010, \u003cem\u003ez\u003c/em\u003e = -87.87, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, 95% CI: [-0.90, -0.86]). The standardized estimated value of the indirect effect was 0.07 (\u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.003, \u003cem\u003ez\u003c/em\u003e\u0026thinsp;=\u0026thinsp;19.69, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The standardized estimated value of the total effect was 0.13 (\u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.003, \u003cem\u003ez\u003c/em\u003e\u0026thinsp;=\u0026thinsp;46.00, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In addition, it is worth noting that we have added an error covariance between visit frequency and loneliness (\u003cem\u003eψ\u003c/em\u003e = -1.02, \u003cem\u003eSE\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.050, \u003cem\u003ez\u003c/em\u003e = -16.89, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, 95% CI: [-1.14, -0.90]), which is completely data-driven. The result of SEM is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis study aimed to introduce and empirically test the EDSL Model, positing that economic deprivation directly affect loneliness and indirectly exacerbates it by diminishing social support. The findings from a large-scale analysis of the UK Biobank dataset offer robust support for the EDSL Model, highlighting the dual pathways through which economic hardship contributes to loneliness.\u003c/p\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e4.1 Confirmation of the EDSL Model and Hypotheses\u003c/h2\u003e\u003cp\u003eFirstly, results from machine learning analyses revealed that a model incorporating two economic deprivation indices (TDI and IMD) and three social support indicators (household size, visit frequency, and confiding frequency) predicted loneliness with high accuracy (AUC\u0026thinsp;=\u0026thinsp;0.81). Among these predictors, frequency of confiding in others emerged as the most influential variable, underscoring the critical role of emotionally intimate relationships in buffering against loneliness. This finding aligns with prior research suggesting that not just the presence, but the quality of social ties\u0026mdash;particularly those enabling emotional disclosure\u0026mdash;is vital for mental well-being.\u003c/p\u003e\u003cp\u003eThe empirical results align closely with the proposed EDSL Model and its core hypotheses. The first hypothesis (H1), which posited a direct link between economic deprivation and increased feelings of loneliness, was supported by the significant direct effect observed in the SEM analysis (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.06, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This indicates that economic deprivation, in and of itself, is a direct antecedent of loneliness, potentially through mechanisms like heightened stress, stigma, or a sense of relative disadvantage.\u003c/p\u003e\u003cp\u003eThe second hypothesis (H2), proposing an indirect pathway where economic deprivation exacerbates loneliness by reducing social support (operationalized as social support in the SEM), was also clearly supported. The SEM results demonstrated that socioeconomic deprivation significantly reduced the social support (\u003cem\u003eβ\u003c/em\u003e = -0.08, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and this diminished social support, in turn, significantly exacerbated feelings of loneliness (\u003cem\u003eβ\u003c/em\u003e = -0.88, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The significant standardized indirect effect (0.07, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) confirms that reduced social support is a key mediating factor in the relationship between economic deprivation and loneliness. Given that SHAP analysis has revealed the high importance of indicators reflecting the quality of emotional support, such as \u0026lsquo;frequency of confiding in others\u0026rsquo; for predicting loneliness, the mediating effect of \u0026lsquo;reduced social support\u0026rsquo; observed in the SEM model may also profoundly reflect the impact of economic deprivation on these high-quality social connections and deep emotional support resources, beyond just a reduction in opportunities for social participation. This pathway underscores how economic deprivation can systematically erode an individual\u0026rsquo;s access to, and the quality of, their social support networks by limiting participation in social activities, as outlined in the EDSL model\u0026rsquo;s conceptualization.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e4.2 Consistency with and Extension of Prior Research\u003c/h2\u003e\u003cp\u003eThese findings resonate with and build upon existing literature. The observed direct link between economic deprivation and loneliness aligns with previous studies establishing strong correlations between these two constructs (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Similarly, the critical role of social support, or its absence, in loneliness is well-documented (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). While existing comprehensive studies have identified significant correlations between economic deprivation, social support, and loneliness (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e), they have often not fully elucidated the specific mechanisms through which these factors interact. Building on this foundational research, the present study introduced and tested the EDSL Model.\u003c/p\u003e\u003cp\u003eThe EDSL model\u0026rsquo;s first proposition\u0026mdash;a direct link from economic deprivation to loneliness\u0026mdash;finds support in research detailing the psychological toll of financial hardship. Studies have shown that the chronic stress, feelings of shame, or perceived social stigma associated with poverty can directly foster social withdrawal and negative self-perceptions, thereby increasing vulnerability to loneliness, independent of quantifiable social interactions (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). Our findings empirically substantiate this direct adverse impact within a broader relational model.\u003c/p\u003e\u003cp\u003eThe EDSL model particularly emphasizes that economic deprivation indirectly exacerbates loneliness by diminishing social support. Consistent with previous studies, economic barriers have greatly restricted people\u0026rsquo;s participation in various social activities, hobbies and community activities, which are crucial for the formation and maintenance of social relationships (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). The EDSL model posits that this reduced participation is a key mechanism through which social support networks are eroded. This extends the work of researchers like Bai et al. (2021) and Zhao \u0026amp; Wu (2022), who have linked facets of social capital and support to loneliness, by explicitly modeling economic deprivation as a critical upstream driver of these support deficits. For example, where Zhao \u0026amp; Wu (2022) showed social support mediating the link between social engagement and loneliness, our model positions economic deprivation as a key factor limiting that initial social engagement and subsequent support.\u003c/p\u003e\u003cp\u003eIt is noteworthy that while our core theory posits that social support alleviates loneliness, the data reveal an additional negative correlation between visit frequency and loneliness. This may indicate that visit frequency not only reflects a component of perceived social support but also represents other protective machines. The impact of this behavioral dimension of social integration may not be fully captured by our social support latent variable, which places greater emphasis on emotional connections or membership identity. Just as (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e) also found that support from friends has an exceptionally prominent protective effect. Friends\u0026rsquo; visit may have additional and prominent protection mechanisms. Face-to-face visits offer genuine and perceptible social interaction and companionship. The lack of high-quality social interaction is at the core of loneliness (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). This regular visiting behavior directly meets human beings\u0026rsquo; basic needs for a sense of belonging and connection (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) and directly reduce the sense of social isolation and the feeling of being ignored for one\u0026rsquo;s own value.\u003c/p\u003e\u003cp\u003eOur results lend empirical weight to this indirect pathway, demonstrating that the erosion of social support (manifested as reduced social support in our SEM) acts as a significant mediator. Notably, the results indicate that this synergistic, indirect effect, where economic deprivation compromises social support and thereby amplifies loneliness, is a substantial contributor to the overall burden of loneliness. The magnitude of the standardized indirect effect (0.07) in relation to the direct effect (0.06) in our study underscores its critical importance. This study therefore extends prior work by not just confirming associations but by providing empirical evidence for specific, theoretically grounded mechanisms detailing how economic hardship translates into increased loneliness via the crucial mediating role of diminished social support. It moves beyond a general understanding of correlations (e.g.,(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e) by testing an integrated model with clearly defined direct and indirect effects within a large-scale cohort.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e4.3 Implications and Future Directions\u003c/h2\u003e\u003cp\u003eThe findings have significant theoretical and practical implications. Theoretically, this study underscores the necessity of adopting integrated models like the EDSL Model to understand multifaceted public health concerns such as loneliness. It provides strong empirical backing for the idea that social and economic structures are deeply intertwined in shaping emotional well-being.\u003c/p\u003e\u003cp\u003ePractically, the dual pathways identified suggest that interventions to combat loneliness must be multi-pronged. Addressing economic deprivation directly through economic policies, strategies aimed at reducing economic deprivation, and support for those facing financial hardship could alleviate one direct source of loneliness. Simultaneously, interventions aimed at bolstering social support and facilitating social participation are crucial, especially for individuals in economically precarious situations. This could include funding for community programs, creating accessible social spaces, and supporting initiatives that reduce the financial barriers to social engagement. Such interventions are consistent with those described by(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eFuture research should aim to replicate these findings in diverse populations and cultural contexts. Longitudinal studies are essential to further elucidate the causal dynamics and temporal precedence of the relationships within the EDSL Model. Investigating other potential mediators (e.g., mental health status, perceived social status) and moderators (e.g., personality traits, community resources) could provide a more comprehensive understanding. Furthermore, exploring the specific types of social support (emotional, instrumental, informational) most affected by economic deprivation and most protective against loneliness would be a valuable avenue for research. Intervention studies based on the EDSL Model are also needed to test the real-world effectiveness of strategies targeting these dual pathways.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003e4.4 Limitations\u003c/h2\u003e\u003cp\u003eHowever, certain limitations should be acknowledged. First, while the UK Biobank is a powerful resource, its participants are aged 40\u0026ndash;69 at baseline, so findings may not generalize to younger or much older populations without further research. Second, loneliness was measured using a single dichotomous question (\u0026ldquo;Do you often feel lonely?\u0026rdquo;); while widely used, this may not capture the full spectrum or intensity of loneliness compared to multi-item scales. Third, although the EDSL model posits causal pathways, the cross-sectional nature of the specific analyses conducted for the SEM limits the ability to definitively infer causality; longitudinal panel data analysis would be beneficial to further substantiate these causal claims. Finally, while social support was the term used in the SEM diagram for the mediator, the underlying construct of social support was measured by three distinct indicators; future research could explore how different facets of social support are uniquely impacted by economic deprivation and contribute to loneliness.\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eThis study provides evidence for the EDSL Model, demonstrating that economic deprivation directly fosters loneliness and indirectly amplifies it by diminishing social support, reflected in an overall reduction of support metrics. These findings highlight the profound impact of economic hardship on social well-being and call for integrated strategies that address both material needs and social connectedness to effectively combat the global public health challenge of loneliness.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAUC: Area Under the Curve\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026beta;\u003c/em\u003e: Standardized regression coefficient\u003c/p\u003e\n\u003cp\u003eCI: Confidence Interval\u003c/p\u003e\n\u003cp\u003eEDSL: Economic Deprivation, Support Diminution, and Loneliness\u003c/p\u003e\n\u003cp\u003eGFI: Goodness of Fit Index\u003c/p\u003e\n\u003cp\u003eIMD: Index of Multiple Deprivation\u003c/p\u003e\n\u003cp\u003eKNN: K-nearest Neighbors\u003c/p\u003e\n\u003cp\u003eLR: Logistic Regression\u003c/p\u003e\n\u003cp\u003eNB: Naive Bayes\u003c/p\u003e\n\u003cp\u003eRF: Random Forest\u003c/p\u003e\n\u003cp\u003eRMR: Root Mean Square Residual\u003c/p\u003e\n\u003cp\u003eROC: Receiver Operating Characteristic\u003c/p\u003e\n\u003cp\u003eSE: Standard Error\u003c/p\u003e\n\u003cp\u003eSEM: Structural Equation Modeling\u003c/p\u003e\n\u003cp\u003eSHAP: Shapley Additive Explanations\u003c/p\u003e\n\u003cp\u003eSRMR: Standardized Root Mean Square Residual\u003c/p\u003e\n\u003cp\u003eTDI: Townsend Deprivation Index\u003c/p\u003e\n\u003cp\u003eUKB: UK Biobank\u003c/p\u003e\n\u003cp\u003eVIF: Variance Inflation Factor\u003c/p\u003e\n\u003cp\u003eWLSMV: Weighted Least Squares Mean and Variance Adjusted\u003c/p\u003e\n\u003cp\u003eXGBoost: Extreme Gradient Boosting\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthical approval for the UK Biobank was granted by the National Health Service Research Ethics Service (approval date: June 17, 2011; reference: 11/NW/0382). All participants provided informed consent at enrollment. Access to and analysis of the data for this study were authorized by the UK Biobank under Application Number 37292.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset(s) analyzed during the current study are available in the UK Biobank repository (https://www.ukbiobank.ac.uk/), under Application Number 37292. Derived datasets generated for this study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA.K. wrote the main manuscript text and analyzed the data. Z.Y.W., F.L., J.N., J.D., and P.Z. contributed to data analysis and preparation, while the other authors assisted with additional analyses and preparation, and reviewed and supervised the study. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe are grateful to the UK Biobank participants and coordinators for providing the data resource that made this study possible.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWang F, Gao Y, Han Z, Yu Y, Long Z, Jiang X, et al. A systematic review and meta-analysis of 90 cohort studies of social isolation, loneliness and mortality. Nat Hum Behav. 2023 June 19;7(8):1307\u0026ndash;19. \u003c/li\u003e\n\u003cli\u003eCacioppo JT, Cacioppo S. The growing problem of loneliness. The Lancet. 2018 Feb 3;391(10119):426. \u003c/li\u003e\n\u003cli\u003eKung CSJ, Pudney SE, Shields MA. Economic gradients in loneliness, social isolation and social support: Evidence from the UK Biobank. Social Science \u0026amp; Medicine. 2022 Aug;306:115122. \u003c/li\u003e\n\u003cli\u003eVictor CR, Pikhartova J. Lonely places or lonely people? Investigating the relationship between loneliness and place of residence. BMC Public Health. 2020 Dec;20(1):778. \u003c/li\u003e\n\u003cli\u003eMao S, Lou VWQ, Lu N. Perceptions of neighborhood environment and loneliness among older Chinese adults: the mediator role of cognitive and structural social capital. Aging \u0026amp; Mental Health. 2023 Mar 4;27(3):595\u0026ndash;603. \u003c/li\u003e\n\u003cli\u003eZhang X, Dong S. The relationships between social support and loneliness: A meta-analysis and review. Acta Psychologica. 2022 July;227:103616. \u003c/li\u003e\n\u003cli\u003eGamper M. Social Network Theories: An Overview. In: Kl\u0026auml;rner A, Gamper M, Keim-Kl\u0026auml;rner S, Moor I, von der Lippe H, Vonneilich N, editors. Social Networks and Health Inequalities: A New Perspective for Research [Internet]. Cham: Springer International Publishing; 2022. p. 35\u0026ndash;48. Available from: https://doi.org/10.1007/978-3-030-97722-1_3\u003c/li\u003e\n\u003cli\u003eHolt-Lunstad J. Social connection as a public health issue: the evidence and a systemic framework for prioritizing the \u0026ldquo;social\u0026rdquo; in social determinants of health. Annual Review of Public Health. 2022;43(1):193\u0026ndash;213. \u003c/li\u003e\n\u003cli\u003ePeng J, Qi H, Fan Z, Zhou Q, Lin Y. Social support and health behaviors of older adults during the COVID-19 pandemic in China: a moderated mediation model of loneliness and economic income. BMC Public Health. 2024 Oct 11;24(1):2780. \u003c/li\u003e\n\u003cli\u003eStephens C, Alpass F, Towers A. Economic hardship among older people in New Zealand: the effects of low living standards on social support loneliness and mental health. 2010; \u003c/li\u003e\n\u003cli\u003eEckhard J. Does Poverty Increase the Risk of Social Isolation? Insights Based on Panel Data from Germany. The Sociological Quarterly. 2018 Apr 3;59(2):338\u0026ndash;59. \u003c/li\u003e\n\u003cli\u003eTang VFY, Chou KL. An exploratory study on material deprivation and loneliness among older adults in Hong Kong. BMC Geriatr. 2024 May 6;24(1):400. \u003c/li\u003e\n\u003cli\u003eBai Z, Wang Z, Shao T, Qin X, Hu Z. Association between social capital and loneliness among older adults: a cross-sectional study in Anhui Province, China. BMC Geriatr. 2021 Dec;21(1):26. \u003c/li\u003e\n\u003cli\u003eZhao L, Wu L. The Association between Social Participation and Loneliness of the Chinese Older Adults over Time\u0026mdash;The Mediating Effect of Social Support. IJERPH. 2022 Jan 12;19(2):815. \u003c/li\u003e\n\u003cli\u003ePedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2012 Jan 2;12. \u003c/li\u003e\n\u003cli\u003eAlgren MH, Ekholm O, Nielsen L, Ersb\u0026oslash;ll AK, Bak CK, Andersen PT. Social isolation, loneliness, socioeconomic status, and health-risk behaviour in deprived neighbourhoods in Denmark: A cross-sectional study. SSM - Population Health. 2020 Apr;10:100546. \u003c/li\u003e\n\u003cli\u003eDavis AJ, Cohen E, Nettle D. Associations amongst poverty, loneliness, and a defensive symptom cluster characterised by pain, fatigue, and low mood. Public Health. 2025 May;242:272\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eCzaja SJ, Moxley JH, Rogers WA. Social Support, Isolation, Loneliness, and Health Among Older Adults in the PRISM Randomized Controlled Trial. Front Psychol. 2021 Oct 5;12:728658. \u003c/li\u003e\n\u003cli\u003eWang J, Mann F, Lloyd-Evans B, Ma R, Johnson S. Associations between loneliness and perceived social support and outcomes of mental health problems: a systematic review. BMC Psychiatry. 2018 Dec;18(1):156. \u003c/li\u003e\n\u003cli\u003eMenec VH, Newall NE, Mackenzie CS, Shooshtari S, Nowicki S. Examining social isolation and loneliness in combination in relation to social support and psychological distress using Canadian Longitudinal Study of Aging (CLSA) data. Reppermund S, editor. PLoS ONE. 2020 Mar 23;15(3):e0230673. \u003c/li\u003e\n\u003cli\u003eInglis G, Sosu E, McHardy F, Witteveen I, Jenkins P, Knifton L. Testing the associations between poverty stigma and mental health: The role of received stigma and perceived structural stigma. International Journal of Social Psychiatry. 2024;00207640241296055. \u003c/li\u003e\n\u003cli\u003eWu J, Zhang J, Fokkema T. The micro-macro interplay of economic factors in late-life loneliness: Evidence from Europe and China. Front Public Health. 2022 Sept 13;10:968411. \u003c/li\u003e\n\u003cli\u003eHawkley LC, Cacioppo JT. Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms. ann behav med. 2010 Oct;40(2):218\u0026ndash;27. \u003c/li\u003e\n\u003cli\u003eBaumeister RF, Leary MR. The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin. 1995;117(3):497\u0026ndash;529. \u003c/li\u003e\n\u003cli\u003eCreed PA, Reynolds J. Economic deprivation, experiential deprivation and social loneliness in unemployed and employed youth. Community \u0026amp;amp; Applied Soc Psy. 2001 May;11(3):167\u0026ndash;78. \u003c/li\u003e\n\u003cli\u003eGolden J, Conroy RM, Bruce I, Denihan A, Greene E, Kirby M, et al. Loneliness, social support networks, mood and wellbeing in community‐dwelling elderly. Int J Geriat Psychiatry. 2009 July;24(7):694\u0026ndash;700. \u003c/li\u003e\n\u003cli\u003eO\u0026rsquo;Rourke HM, Collins L, Sidani S. Interventions to address social connectedness and loneliness for older adults: a scoping review. BMC Geriatr. 2018 Dec;18(1):214. \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Loneliness, Economic Deprivation, Social Support, EDSL Model, UK Biobank, Machine Learning, Structural Equation Modeling","lastPublishedDoi":"10.21203/rs.3.rs-7553119/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7553119/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eLoneliness is increasingly recognized as a major public health concern shaped by both economic and social structures. However, prior research often examined these determinants in isolation, limiting understanding of their combined effects.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eWe proposed and tested the Economic Deprivation, Support Diminution, and Loneliness (EDSL) Model using UK Biobank data (N\u0026thinsp;=\u0026thinsp;232,477). Economic deprivation was assessed with the Townsend Deprivation Index (TDI) and the Index of Multiple Deprivation (IMD), while social support was measured by household size, visit frequency, and confiding frequency. Loneliness was coded as a binary outcome. Machine learning models predicted loneliness, with SHAP values identifying influential predictors. Structural equation modeling (SEM) examined both direct and indirect pathways from economic deprivation to loneliness.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eXGBoost predicted loneliness well (AUC\u0026thinsp;=\u0026thinsp;81.35%). SHAP values showed confiding frequency, IMD, and household size as key predictors. SEM confirmed economic deprivation\u0026rsquo;s direct (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.06) and indirect effects on loneliness via reduced social support (indirect \u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.07).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eFindings provide robust empirical support for the EDSL Model, showing that economic deprivation fosters loneliness both directly and indirectly by eroding social support. These results highlight the need for integrated public health strategies that address economic hardship while strengthening social connectedness to reduce loneliness.\u003c/p\u003e\u003ch2\u003eTrial registration:\u003c/h2\u003e\u003cp\u003eNot applicable.\u003c/p\u003e","manuscriptTitle":"Economic Deprivation, Diminished Social Support, and Loneliness: An Empirical Test of the EDSL Model Using UK Biobank Data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-20 13:13:21","doi":"10.21203/rs.3.rs-7553119/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-10-07T13:20:08+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-08T13:36:28+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-08T12:44:09+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-08T12:40:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Public Health","date":"2025-09-06T21:52:03+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0133057b-ae98-469b-bc78-6961e3705adb","owner":[],"postedDate":"October 20th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-10-20T13:13:21+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-20 13:13:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7553119","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7553119","identity":"rs-7553119","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}