Evaluating sleep patterns and intrinsic capacity with machine learning: Results from the Gan-Dau healthy longevity plan

Abstract

Background

This study aims to examine the association between sleep and intrinsic capacity (IC), employing a machine-learning approach, to promote healthy aging and disability prevention in the community.

Methods

A cohort of 810 community-dwelling individuals aged 50 years were enrolled. Sleep patterns were assessed using the Pittsburgh Sleep Quality Index (PSQI) and its subdomains. Unsupervised machine learning through K-means clustering was applied to classify sleep patterns into four distinct categories, enabling further analysis. IC was evaluated by assessing cognitive, locomotion, vitality, psychological, and sensory functions and was subsequently rescaled using the percent of the maximum possible method.

Results

Low IC was linked to higher PSQI (OR 1.10, 95% CI 1.05-1.15, p<0.001), as well as subdomains indicating poor sleep quality, lower habitual sleep efficiency, and increased sleep disturbances. Poor sleep quality (PSQI >5) was associated with low IC and lower scores in the psychological wellbeing, and vitality subdomains. Results of K-means clustering analysis showed: Category 1 (worst sleepers) (OR 2.54, 95% CI 1.55-4.16, p<0.001), Category 2 (short and inefficient sleepers, OR 1.69, 95% CI 1.18-2.43, p=0.004), and Category 3 (inefficient sleepers, OR 1.50, 95% CI 1.02-2.20, p=0.037) exhibited a higher risk for low IC compared to robust sleepers.

Conclusions

The study highlights the crucial role of sleep quality in maintaining intrinsic capacity and promoting healthy aging. Impairments in psychological wellbeing and vitality were identified as the primary contributors. This emphasizes the importance of promoting healthy sleep habits for overall well-being.