Technophobia and Psychosocial Correlates of Depressive Symptoms in Older Adults: Machine Learning and Network Analysis.
Abstract
Objectives
To examine technophobia and psychosocial correlates of current depressive symptom status among older adults using machine learning and network analysis.
Methods
This cross-sectional study included 358 adults aged ≥ 60 years in Hubei, China. Four classifiers were evaluated using 100 repeated stratified train-test splits, with model development restricted to training data. A covariate-adjusted regularized partial correlation network examined dimension-level conditional associations.
Results
Overall, 119 participants (33.2%) had GDS-15 scores ≥ 5. Mean AUCs ranged from 0.927 to 0.949, with no clear between-model differences. Subjective age showed the greatest permutation importance; technophobia, self-perceptions of aging, and intergenerational support were consistently informative. In the adjusted network, fear of personal failure, negative outcomes/control, and interest age showed the largest positive associations with depressive symptoms, whereas emotional support showed the largest negative association. Cross-community connections linked perceived convenience with positive control and daily care support with chronic timeline.
Conclusions
Technophobia, subjective age, self-perceptions of aging, and intergenerational support emerged as consistent correlates of current depressive symptom status. Network analysis further identified dimension-level and cross-community conditional associations.Clinical implications Support may emphasize error-tolerant digital design with clear feedback, opportunities to maintain personally meaningful activities, and emotionally supportive family care that preserves autonomy.