INTRODUCTION

In recent years, Korea has experienced a rapid increase in its aging population, transitioning to an aging society in 2010 and projected to become a super-aging society by 2025 [1]. Concurrently, there has been a significant rise in the number of older adults living alone, a trend that may be linked to the fracturing of traditional family structures in Korea [2]. According to a 2021 statistical report, the number of older adults living alone in Korea rose from 1.22 million in 2015 to 1.53 million in 2019 and 1.66 million in 2020. If this trend persists, the number of older adults living alone is projected to reach 2.27 million by 2027 and 3.35 million by 2037 [1].

Older adults who live alone are defined as individuals residing in single-person households without the presence of a spouse, children, or any family members [3]. Recent research indicates that living alone in community dwellings poses a significant risk to well-being, primarily due to an increased likelihood of poor health outcomes, such as deteriorated self-rated health, mental health, and emotional well-being [4–6]. Choi [7] found that older adults living alone had a higher incidence of chronic diseases, lower scores on cognitive tests, and more symptoms of depression compared to those living with others. Additionally, they participate less in social activities than their counterparts who reside with family [8,9]. Previous studies have also suggested that older adults living alone are more prone to social isolation than those who live with others [10,11].

These risk factors for well-being can significantly impact the lives of older adults living alone. In their 2018 study, Moon and Kim [12] observed a link between self-rated health (p<0.01), lifestyle (p<0.01), and social participation (p<0.01) and life satisfaction in older adults living alone, while controlling for leisure and social activities and residential environment. This study highlighted the physical environment of older adults living alone, a focus that diverges from previous research centered on individual-level factors. Meanwhile, Kang and Yoon [13] suggested that loneliness mediates the relationship between participation in social activities and life satisfaction in older adults living alone, underscoring the significance of engaging in social activities for this demographic’s life satisfaction. However, although previous studies have explored factors that either hinder or enhance life satisfaction in this group, they have often overlooked a broad array of factors influencing life satisfaction, such as physical functions, physical health, economic activity, and mental health status [14,15]. To better understand the life satisfaction of older adults living alone, it is necessary to consider variables associated with these dimensions.

Korea collects national data on older adults through the Korea Senior Survey and the Korean Longitudinal Study of Aging to better understand the dynamics of its aging population. In this context, employing big data and data-centric analysis to examine the life satisfaction of older adults living alone could yield significant insights. Machine learning (ML), with its capability to develop predictive and classification models by identifying hidden patterns in large datasets, offers an objective approach that may surpass traditional statistical methods in complex data analysis [16]. ML includes both supervised and unsupervised learning. Supervised learning involves training models for categorization or prediction using data with known outcomes, such as regression models and decision trees. It produces results by creating a classification or prediction model for an outcome variable using the same data as the training data. Meanwhile, unsupervised learning focuses on reducing dimensions or grouping data into clusters [17]. Although both traditional statistics and ML aim to predict, classify, and infer data, traditional statistics primarily concentrate on inferences derived from probabilistic models [18]. However, ML may be more suitable for big data analyses as it excels at predicting and classifying patterns in large, complex datasets [18]. Acknowledging the limitations of previous studies, our study employs ML techniques to investigate the various factors affecting the life satisfaction of older adults living alone, aiming to identify key variables that influence this demographic’s life satisfaction.

This study explored the factors influencing life satisfaction among older adults living alone by developing ML models that classify their levels of satisfaction. The objective was to identify the key variables that impact the life satisfaction of older adults living alone.

METHODS

Data Source and Participants

We extracted demographic and clinical characteristics from 10 097 adults using data from the 2020 Korea Senior Survey. This national survey, initiated in 2007, aims to explore the characteristics of older adults in various domains including health, economics, and social participation, and to track changes over time. After excluding participants with missing data, those aged 64 and under, and individuals not living alone, our final sample consisted of 3112 older adults.

Study Variables

The outcome variable in this study was the overall life satisfaction of older adults living alone. We measured life satisfaction with the question, “How satisfied are you with your overall life?” Responses were scored on a 5-point Likert scale (1: very satisfied, 2: satisfied, 3: neutral, 4: dissatisfied, and 5: very dissatisfied). For the purposes of analysis, the life satisfaction variables were converted into dummy variables and used as target variables, with 1 representing very stisfied or satisfied, and 0 representing neutral, dissatisfied, or very dissatisfied.

The predictor variables included demographics, health, functionality, environment, and activity participation. Within the demographic domain, variables such as gender, age, economic status, residential area (1: metropolitan city, 0: regional city), marital status, and educational attainment were considered for older adults living alone.

The health domains included 32 chronic diseases, such as hypertension, stroke, and diabetes, coded as 1 for yes and 0 for no. They also included the total number of chronic diseases diagnosed by a doctor, the number of medications prescribed by a doctor for more than three months, the presence of a disability diagnosis (1 for yes, 0 for no), current smoking status, alcohol consumption in the past year, experience of falls in the past year, utilization of a medical institution in the previous month, and admission to a nursing home or other healthcare facility in the past year.

The functional domains assessed included the use of orthoses (visual, hearing, and chewing) for performing activities of daily living (ADL), physical capacity, cognitive function, and depression. Physical capacity was evaluated through various activities: running and walking across a field, climbing ten steps, bending, squatting, or kneeling, reaching and touching objects above head level, and lifting an 8 kg object. A higher score indicated impaired physical function in older adults. Cognitive function was measured with the Korean version of the Mini-Mental State Examination for Dementia Screening (MMSE-DS), which is utilized to detect dementia in this population. Depression was assessed using the short form of the Korean version of the Geriatric Depression Scale (SGDS-K).

The environmental domains included the type of residence, categorized as 1: owned, 2: annually rented, 3: monthly rented, 4: free; housing type, classified as 1: house, 2: apartment, 3: multicomplex house, 4: other; residence satisfaction, scored as 1: very satisfied or satisfied, 0: neutral, dissatisfied, or very dissatisfied; satisfaction with the community environment; unmet medical care needs in the previous year; travel time to the hospital; the number of adult children; and the frequency of visits from adult children. Satisfaction with the community environment was evaluated using a 5-point Likert scale (1: very satisfied, 2: satisfied, 3: neutral, 4: dissatisfied, 5: very dissatisfied) across seven areas: convenience facilities, public transportation, green spaces, security, proximity to adult children and other family members, opportunities for neighborly interaction, and the overall environment.

The activity participation domains were basic activities of daily living (BADL), instrumental activities of daily living (IADL), social activities (3 domains: travel, leisure, educational activity), exercise, and nutritional activity (2 domains: less than 2 meals per day, 3 or more drinks per day).

Statistical Analysis

The participants’ demographic and clinical characteristics were analyzed using descriptive statistics. Categorical variables were presented as frequency distributions and percentages, while numerical variables were described using means and standard deviations (SDs). We utilized the chi-square and Wilcoxon signed rank sum tests to compare life satisfaction levels between the satisfied and dissatisfied groups. In this paper, we explored various ML models, including logistic Lasso regression, classification and regression tree (CART), C5.0, random forest, and extreme gradient boosting (XGBoost), to determine the optimal classification model. The decision tree algorithm used the CART, C5.0, random forest, and XGBoost models. Decision trees classify data by dividing them into independent and dependent variables and develop predictive models. The split points are determined using metrics such as entropy (for C5.0) or the Gini index (for CART [16]). Random forest is an ensemble model that utilizes bagging to resample data and construct multiple decision trees. It then aggregates the results to enhance classification accuracy [19]. Similarly, XGBoost is another ensemble model that uses boosting techniques. It builds decision trees sequentially, focusing on correcting errors from previous trees while optimizing certain processes through parallelization [20]. The dataset was split into training and test data, comprising 80% and 20% of the total, respectively. We fine-tuned the hyperparameters using both random and grid search methods, along with tenfold cross-validation. To evaluate the ML models’ classification performance, we used accuracy, precision, recall, the F1-score, and the area under the curve (AUC). Accuracy refers to the ratio of accurate predictions (either true positive or true negative) out of all predictions made. Precision measures the percentage of true positives among the predicted positive results. Recall, also known as sensitivity, calculates the proportion of true positives out of all actual positives. The F1-score combines precision and recall into a single metric, offering a balanced assessment of model performance by considering both metrics [16]. Important variables were identified from each model using information gain and Shapley additive explanations (SHAP) values. SHAP enhances the interpretability of the ML model by applying Shapley values from game theory to determine the impact of each input variable on the model’s output by evaluating how its presence or absence affects the model’s output. SHAP values provide a consistent measure of both positive and negative contributions, capturing the overall influence of variables, unlike information gain. These values are calculated by determining the average change in model output with and without the inclusion of a specific variable [21]. In SHAP plots, the x-axis represents the SHAP values, indicating the direction of correlation with the classification model. The y-axis lists the explanatory variables in order of their impact on classification performance. The color of each dot reflects the intensity of the feature value, ranging from red (positive values) to blue (negative values), with darker shades indicating stronger magnitudes of positive or negative values. SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) was utilized to extract and preprocess data. The Python package scikit-learn (version 3.7.6) was employed to develop the ML model and identify important variables.

Ethics Statement

The Institutional Review Board of Yonsei University Mirae Campus approved our study and granted an exemption from review (IRB No. 1041849-202208-SB-139-01).

RESULTS

Table 1 presents the demographic and clinical characteristics of the participants. Among older adults living alone, 1701 (54.7%) reported dissatisfaction with their lives. This group included 364 (21.4%) men and 1337 (78.6%) women, with an average age of 76.1 years (SD, 6.9). Additionally, 929 (54.6%) of these individuals resided in metropolitan areas. Conversely, 1411 (45.3%) older adults living alone expressed satisfaction with their lives. This satisfied cohort consisted of 275 (19.5%) men and 1136 (80.5%) women, with a mean age of 74.5 years (SD, 6.7).

Table 2 shows the performance results of the ML models. XGBoost surpassed the other ML models in accuracy, precision, and AUC, with scores of accuracy=0.75, precision=0.73, recall=0.71, F1-score=0.72, and AUC=0.75. Logistic Lasso regression and random forest exhibited similar performance metrics, though both were less effective than XGBoost (logistic Lasso regression: accuracy=0.74, precision=0.70, recall=0.74, F1-score=0.72, AUC=0.74; random forest: accuracy=0.74, precision=0.72, recall=0.69, F1-score=0.71, AUC=0.74). CART and C5.0 showed comparable results in terms of accuracy, precision, recall, F1-score, and AUC (CART: accuracy=0.69, precision=0.63, recall=0.80, F1-score=0.70, AUC=0.70; C5.0: accuracy=0.69, precision=0.63, recall=0.78, F1-score=0.70, AUC=0.70). Therefore, XGBoost was chosen as the final ML model based on its F1-scores and AUC. Table 3 summarizes the five most significant variables in the XGBoost model according to information gain. The model identified the following key variables affecting life satisfaction among older adults: overall community satisfaction, self-rated health, community satisfaction regarding opportunities to interact with neighbors, community satisfaction related to proximity to a child, and satisfaction with residence.

Figure 1 illustrates the factors and their significance in categorizing the life satisfaction of older adults living alone, as determined by SHAP values. Factors positively associated with high life satisfaction include good self-rated health, high overall satisfaction with the community environment, satisfaction related to the proximity of adult children and other family members, high satisfaction with their residence, satisfaction related to opportunities for neighborly interactions, ownership of the residence, and frequent visits from adult children. Conversely, severe depressive symptoms, a high number of chronic diseases, and a low frequency of participation in social activities were found to negatively impact life satisfaction.

DISCUSSION

In this study, we developed ML models to identify factors associated with life satisfaction among older adults living alone, focusing on demographic, health, activity, and environmental domains. We used XGBoost as the final ML model and identified key variables such as overall community satisfaction, self-rated health, community satisfaction regarding opportunities to interact with neighbors, community satisfaction related to proximity to a child, and residence satisfaction. These variables were deemed important based on the information gain from the XGBoost model. Additionally, we determined the magnitude of these important variables and their relationship to life satisfaction by extracting the SHAP value.

To identify measures that could improve the life satisfaction of older adults living alone, we determined that satisfaction with the general community environment is a modifiable factor. This study utilized satisfaction in seven distinct areas of the community environment as predictors. Satisfaction with the overall community environment emerged as a significant variable in XGBoost, playing a crucial role in reducing impurity during group classification. Park and Kang [22] identified a correlation between older adults’ life satisfaction and their satisfaction with the community environment. Additionally, several studies have demonstrated that the community environment significantly impacts the quality of life of older adults [12,23,24] . Moreover, satisfaction with the general community environment also indirectly influences older adults’ overall life satisfaction. Liu and Heo [25] highlighted a positive association between subjective neighborhood relationships and social participation among older adults who were content with the physical aspects of their community, suggesting a potential increase in life satisfaction. This study corroborates previous findings and analyzes data specifically concerning older adults living alone, providing insights that may elucidate the unique characteristics of this demographic. Recent research on age-friendly environments, guided by the World Health Organization’s framework for age-friendly cities, supports our findings and underscores the importance of creating age-friendly environments for older adults living alone [26,27].

One variable that effectively explains life satisfaction is the self-rated health of older adults living alone; this study reveals that self-rated health is a useful indicator for classifying their life satisfaction. Previous studies have previously indicated a positive correlation between self-rated health and life satisfaction among older adults, a finding supported by this research [28,29]. Self-rated health shows significant variability in conditions such as psychiatric and cerebrovascular diseases, and cancer, and regression models demonstrate that it captures a wide range of variances in actual health status [30]. Furthermore, self-rated health is associated with mortality among older adults, underscoring the importance of its regular assessment [31]. According to the 2020 Senior Survey Results Report, self-rated health is a crucial factor in determining life satisfaction for older adults living alone; however, their self-rated health scores were lower compared to those living with children or other family members [1]. These results highlight the need for a national understanding of the unique characteristics of older adults living alone and the development of tailored support systems and welfare programs to address their specific needs [29]. Each country have to establish and maintain a suitable social support system to promote healthy aging for older adults living alone and enhance it to help subsidize their health and welfare costs.

Improving the quality of life for older adults living alone can be facilitated by geographic proximity between older adults and their adult children. Previous studies have shown that older adults who live near their adult children are more satisfied with their lives than those who do not [32,33]. Additionally, interactions with family members and acquaintances have been linked to increased life satisfaction among older adults living alone [34]. Our study focused on satisfaction with the distance from an adult child rather than specific geographic proximity. Nevertheless, our findings align with the context of earlier results.

Previous studies have shown a negative correlation between certain variables and the quality of life of older adults living alone. The relationship between depression and social participation is complex, potentially affecting social isolation or increasing the risk of suicide among this demographic. Research has indicated that external factors such as family bonds, support systems, and social capital can help alleviate these adverse effects [35–37]. This study’s results suggest that the community environment, the frequency of interactions with their children, and a neighborhood conducive to engaging with neighbors significantly impact the life satisfaction of older adults living alone. The Korean government has initiated programs aimed at increasing social participation among older adults, including those who live alone [38]. Our findings highlight the importance of external factors, including social participation, the community environment, and communication with children, in enhancing the quality of life for older adults living alone.

This study employed various analytical methods to assess the life satisfaction of older adults living alone, including logistic Lasso regression, decision tree models (CART and C5.0), random forest, and XGBoost. Both ensemble and non-ensemble models were utilized, with random forest and XGBoost representing the ensemble approaches. Ensemble models, which are sophisticated ML techniques, create multiple decision tree models by resampling the entire dataset and selecting the model that demonstrates the best performance, rather than relying on a single decision tree model [19]. These methods help reduce the variability that can occur with data in a single model, which explains why the ensemble models in this study outperformed the single models, CART and C5.0. Previous studies across various fields have compared single and ensemble models, consistently finding that ensemble models offer superior performance; our study’s findings align with these observations [39, 40]. Furthermore, this study addressed the reduced interpretability of ensemble models compared to single models by providing explanations based on information gains and SHAP values.

Notes

Conflict of Interest

The authors have no conflicts of interest associated with the material presented in this paper.

Funding

This research was supported by the BK21 FOUR (Fostering Outstanding Universities for Research) funded by the Ministry of Education (MOE) of the Republic of Korea and National Research Foundation of Korea (NRF) (Big data specialized education and research team for cognitive health and social integration of community-dwelling older adults).

Acknowledgements

None.

Author Contributions

Conceptualization: Bae S. Data curation: Bae S. Funding acquisition: Hong I. Methodology: Bae S, Lee MJ, Hong I. Writing – original draft: Bae S. Writing – review & editing: Lee MJ, Hong I.

Figure 1

Feature importance of the life satisfaction classification model based on Shapley additive explanation (SHAP) values. SGDS-K, Korean version of the short form of the Geriatric Depression Scale. The variables are listed in descending order, indicating their significance as measured by the magnitude of the SHAP values. SHAP values show how each parameter interacts with the model output, with higher parameter values in red and lower values in blue.

REFERENCES

• 1. Statistics Korea. 2020 Korea Senior Survey result report; 2020 [cited 2024 Sep 7]. Available from: https://www.mohw.go.kr/board.es?mid=a10107010100&bid=0040&act=view&list_no=374195&tag=&cg_code=&list_depth=1 (Korean)
• 2. Park KS, Kim MS. The impact of changes in household living arrangements on the changes in poverty rate among elderly people in South Korea. Korean J Sociol 2016;50(1):221-253. (Korean).Article
• 3. Choi Y. Economic and health status, social support and depression of the elderly living alone. Soc Sci Res 2008;24(4):103-123. (Korean).
• 4. You KS, Lee H. The physical, mental, and emotional health of older people who are living alone or with relatives. Arch Psychiatr Nurs 2006;20(4):193-201. https://doi.org/10.1016/j.apnu.2005.12.008ArticlePubMed
• 5. Shin SH, Sok SR. A comparison of the factors influencing life satisfaction between Korean older people living with family and living alone. Int Nurs Rev 2012;59(2):252-258. https://doi.org/10.1111/j.1466-7657.2011.00946.xArticlePubMed
• 6. Nam SI, See JH, Jung YE, Chae J. Meaning in life, depression, and suicidal ideation in old age: a comparative study of living alone and not living alone older adults. Korean J Gerontol Soc Welf 2019;74(2):163-192. (Korean).Article
• 7. Choi M. Association between social participation and cognitive function among community-dwelling older adults living alone: analysis of a nationally representative survey. Int J Nurs Pract 2020;26(4):e12841. https://doi.org/10.1111/ijn.12841ArticlePubMed
• 8. Kang E, Lee M. Identifying the effect of living alone on life in later adulthood: comparison between living alone and those living with others with a propensity score matching analysis. Health Soc Welf Rev 2018;38(4):196-226. (Korean). https://doi.org/10.15709/hswr.2018.38.4.196Article
• 9. Sakurai R, Kawai H, Suzuki H, Kim H, Watanabe Y, Hirano H, et al. Poor social network, not living alone, is associated with incidence of adverse health outcomes in older adults. J Am Med Dir Assoc 2019;20(11):1438-1443. https://doi.org/10.1016/j.jamda.2019.02.021ArticlePubMed
• 10. Broughton K, Beggs BA. Leisure satisfaction of older adults. Act Adapt Aging 2007;31(1):1-18. https://doi.org/10.1300/J016v31n01_01Article
• 11. Yu CY, Hou SI, Miller J. Health for older adults: the role of social capital and leisure-time physical activity by living arrangements. J Phys Act Health 2018;15(2):150-158. https://doi.org/10.1123/jpah.2017-0006ArticlePubMed
• 12. Moon JH, Kim D. Factors influencing life satisfaction in elderly living alone. J Korea Contents Assoc 2017;18(1):44-54. (Korean). https://doi.org/10.5392/JKCA.2018.18.01.044Article
• 13. Kang JP, Yoon J. A study on multi-level models in life satisfaction of elderly living alone: moderating effect of elderly’s leisure activity and social support. J Digit Converg 2017;15(5):89-98. (Korean). https://doi.org/10.14400/JDC.2017.15.5.89Article
• 14. Yi Y, Park YH. Structural equation model of the relationship between functional ability, mental health, and quality of life in older adults living alone. PLoS One 2022;17(8):e0269003. https://doi.org/10.1371/journal.pone.0269003ArticlePubMedPMC
• 15. Sohn SY. Factors affecting health related quality of life between living alone and living together in the elderly. J Digit Converg 2020;18(12):293-302. (Korean). https://doi.org/10.14400/JDC.2020.18.12.293Article
• 16. Kubat M. An introduction to machine learning; 2nd ed. Cham: Springer; 2017 https://doi.org/10.1007/978-3-319-63913-0
• 17. Dineva K, Atanasova T. Systematic look at machine learning algorithms–advantages, disadvantages and practical applications. International Multidisciplinary Scientific GeoConference: SGEM. 2020; 20(2.1):317-324 https://doi.org/10.5593/sgem2020/2T/s07.041Article
• 18. Bzdok D, Altman N, Krzywinski M. Statistics versus machine learning. Nat Methods 2018;15(4):233-234ArticlePubMedPMCPDF
• 19. Zhou ZH. Ensemble methods: foundations and algorithms. Boca Raton: CRC Press; 2012. p. 15-20 p. 57-59
• 20. González S, García S, Del Ser J, Rokach L, Herrera F. A practical tutorial on bagging and boosting based ensembles for machine learning: algorithms, software tools, performance study, practical perspectives and opportunities. Inf Fusion 2020;64: 205-237. https://doi.org/10.1016/j.inffus.2020.07.007Article
• 21. Lundberg SM, Erion GG, Lee SI. Consistent individualized feature attribution for tree ensembles. arXiv [Preprint]. 2019 [cited 2024 Sep 7]. Available from: https://doi.org/10.48550/arXiv.1802.03888
• 22. Park JH, Kang SW. Factors related to life satisfaction of older adults at home: a focus on residential conditions. Healthcare (Basel) 2022;10(7):1279. https://doi.org/10.3390/healthcare10071279ArticlePMC
• 23. Bae NR, Park CS. A study on the ecological factors affecting the quality of life among the elderly people. J Korean Gerontol Soc 2009;29(2):761-779. (Korean).
• 24. Hwang E. A study on comparison of predictive factors on happiness among male and female aged living alone. J Korea Acad Ind Coop Soc 2019;20(8):392-402. (Korean). https://doi.org/10.5762/KAIS.2019.20.8.392Article
• 25. Liu D, Heo Y. The effect of satisfaction with community physical environment on life satisfaction among the elderly: a focus on the double mediating effects of subjective neighborhood relationships and social participation. J Korean Gerontol Soc 2022;42(1):17-35. (Korean). https://doi.org/10.31888/JKGS.2022.42.1.17Article
• 26. Park S, Lee S. Age-friendly environments and life satisfaction among South Korean elders: person-environment fit perspective. Aging Ment Health 2017;21(7):693-702. https://doi.org/10.1080/13607863.2016.1154011ArticlePubMed
• 27. Sánchez-González D, Rojo-Pérez F, Rodríguez-Rodríguez V, Fernández-Mayoralas G. Environmental and psychosocial interventions in age-friendly communities and active ageing: a systematic review. Int J Environ Res Public Health 2020;17(22):8305. https://doi.org/10.3390/ijerph17228305ArticlePubMedPMC
• 28. Kim JH. Influential factors on the change in life satisfaction of elderly households -longitudinal analysis using a latent growth model. J Korea Contents Assoc 2019;19(1):339-349. (Korean). https://doi.org/10.5392/JKCA.2019.19.01.339Article
• 29. Park H, Jang E, Hong S. The influence of satisfaction with interpersonal relationships and social support on subjective health status among the elderly people living alone. Soc Sci Res Rev 2020;36(2):1-19. (Korean). https://doi.org/10.18859/ssrr.2020.5.36.2.1Article
• 30. Choi Y. Is self-rated health a sufficient proxy for true health. Korean J Gerontol Soc Welf 2018;73(4):7-28. (Korean). https://doi.org/10.21194/kjgsw.73.4.201812.7Article
• 31. Wuorela M, Lavonius S, Salminen M, Vahlberg T, Viitanen M, Viikari L. Self-rated health and objective health status as predictors of all-cause mortality among older people: a prospective study with a 5-, 10-, and 27-year follow-up. BMC Geriatr 2020;20(1):120. https://doi.org/10.1186/s12877-020-01516-9ArticlePubMedPMC
• 32. van der Pers M, Mulder CH, Steverink N. Geographic proximity of adult children and the well-being of older persons. Res Aging 2015;37(5):524-551. https://doi.org/10.1177/0164027514545482ArticlePubMed
• 33. Wei Y, Tsay WJ. Does distance make happiness? Geographic proximity of adult children and the well-being of older persons. J Aging Soc Policy 2024;36(2):222-240. https://doi.org/10.1080/08959420.2022.2080464ArticlePubMed
• 34. Lu N, Spencer M, Sun Q, Lou VW. Family social capital and life satisfaction among older adults living alone in urban China: the moderating role of functional health. Aging Ment Health 2021;25(4):695-702. https://doi.org/10.1080/13607863.2019.1709155ArticlePubMed
• 35. Holtfreter K, Reisig MD, Turanovic JJ. Depression and infrequent participation in social activities among older adults: the moderating role of high-quality familial ties. Aging Ment Health 2017;21(4):379-388. https://doi.org/10.1080/13607863.2015.1099036ArticlePubMed
• 36. Vanderhorst RK, McLaren S. Social relationships as predictors of depression and suicidal ideation in older adults. Aging Ment Health 2005;9(6):517-525. https://doi.org/10.1080/13607860500193062ArticlePubMed
• 37. Oh HJ, Bae SM. The moderating effects of individual and community social capital on the relationship between depressive symptoms and suicide in the elderly. Curr Psychol 2021;40(8):4164-4171. https://doi.org/10.1007/s12144-021-01589-8Article
• 38. Bae S, Hong I, Park JH. Current status and implications of health and welfare policy and support system for older adults: focusing on lifestyle. Korea J Geriatr Occup Ther 2023;5(1):49-63. (Korean).
• 39. Usmani S, Saboor A, Haris M, Khan MA, Park H. Latest research trends in fall detection and revention using machine learning: a systematic review. Sensors (Basel) 2021;21(15):5134. https://doi.org/10.3390/s21155134ArticlePubMedPMC
• 40. Xin Y, Ren X. Predicting depression among rural and urban disabled elderly in China using a random forest classifier. BMC Psychiatry 2022;22(1):118. https://doi.org/10.1186/s12888-022-03742-4ArticlePubMedPMC

Figure & Data

References

Citations

Citations to this article as recorded by