Development of a machine learning-based depression risk identification tool for older adults with asthma

Comprehensive Summary

This study developed a Depression Risk Identification Tool (DRIT) using machine learning to help detect depression in older adults with asthma. Asthma affects quality of life, and when depression occurs alongside it, health outcomes become much worse. To build the tool, the researchers used data from 1,154 asthma patients in the China Health and Retirement Longitudinal Study (CHARLS). They first applied LASSO regression to narrow down 21 key predictors of depression from a larger set of variables. Eight different machine learning models were tested, and the glmBoost algorithm showed the best performance. In testing, it reached an AUC of 0.740, while in the validation group, it achieved 0.664. Some of the main risk factors identified were poor cognitive function, heavy exercise, being unmarried, and female gender. The researchers also used SHAP values to make the model’s predictions more interpretable by showing how each factor contributed to the results. The study showed that a machine learning based tool can reliably predict depression risk in a vulnerable population.

Outcomes and Implications

Depression often goes undiagnosed in older asthma patients, yet it can make managing asthma much harder and reduce quality of life. A tool like DRIT can give doctors a way to flag patients at risk early, even if depression symptoms are subtle. Because the model is interpretable, providers can see which factors raise the risk and use that information to guide treatment discussions. This makes it easier to connect patients with counseling, medication, or social support before problems escalate. It also supports a more integrated approach to care, treating both physical and mental health together. While the study was based on survey data and still needs validation in other populations, the results show clear potential. If it is scaled responsibly, this type of tool can improve patient outcomes and reduce the burden on healthcare systems.