Exploring nationwide patterns of sleep problems from late adolescence to adulthood using machine learning

Abstract

Sleep problems among young adults pose a major public health challenge. Leveraging nationwide health surveys and registers from Denmark, we investigated patterns of sleep problems from late adolescence to adulthood and explored early life-course determinants. We generated life-course embeddings using unsupervised machine learning on data from 2.2 million individuals born from 1980 to 2015. We used this landscape to identify neighboring factors of sleep problems. We observed a substantial increase in self-reported sleep problems among individuals aged 15 to 45, from 34 to 49% between 2010 and 2021, and a 10-fold increase in melatonin use. We also found relevant clusters of sleep-related prescriptions, diagnoses, and procedures with age-specific incidence patterns. Specific childhood adversities, such as sibling psychiatric illness, foster care, and parental divorce, were shared factors across multiple sleep disorders such as insomnia and nightmares. These findings underscore the complex interplay between medical and psychosocial factors in sleep.

Samir Bhatt

Professor, FMedSci, MAE; Scientific Director of GPAP

Statistics, machine learning and Bayesian inference for public health and infectious diseases.