Urban–rural disparities in the prevalence and trends of loneliness among Chinese older adults and their associated factors: Evidence from machine learning analysis

Abstract

In the context of rapid aging development, exploring the predictive factors of older adults' loneliness and its urban–rural differences is of great significance for promoting the psychological health of older adults. This study selected 30 variables from the three waves of China Family Panel Studies (CFPS) data in 2016, 2018, and 2020, using a random forest classifier to explore the predictive factors of loneliness. The sense of loneliness among rural older adults is significantly higher than that of urban older adults. Among the top 10 predictors of loneliness, there are seven common factors in urban and rural, including sleep quality, marital status, confidence in the future, weekly family dinners, life satisfaction, physical discomfort in the past 2 weeks, and relationship with children. The other three different predictive factors for urban older adults are weekly movie and TV duration, family size, and family net worth, while self-rated health, health change, and per capita family income set the rural older adults apart. In addition, the urban–rural differences in the predictive factors of older adults' loneliness show different development trends in the time dimension. Paying attention to the predictive factors that contribute to the high ranking of older adults' loneliness and the widening trend of urban–rural differences is highly required.