Predicting depression severity among people living with HIV in Oman using random forest

Area of Science:

• Medical Informatics
• Psychiatry
• Epidemiology

Background:

• Depression is a common comorbidity in people with HIV, linked to worse health outcomes.
• Current screening tools often miss crucial sociodemographic and psychosocial influences.
• Machine learning (ML) offers advanced methods to analyze complex variable interactions.

Purpose of the Study:

• To investigate the utility of ML, specifically Random Forest, in predicting depression severity among people living with HIV (PLWH).
• To identify key sociodemographic and psychosocial predictors of depression in PLWH.
• To explore the integration of ML for enhanced mental health screening in HIV care.

Main Methods:

• A cross-sectional, multicentre study involving 256 PLWH in Oman.
• Depression severity assessed using the PHQ-9 scale.
• Random Forest regression model implemented to analyze depression predictors.

Main Results:

• A depression prevalence of 26.3% was observed in the study population.
• The Random Forest model achieved moderate predictive accuracy, explaining 35% of the variance in PHQ-9 scores.
• Self-efficacy and perceived social support emerged as the most significant predictors of depression severity.

Conclusions:

• Machine learning, particularly Random Forest, shows promise for predicting depression in PLWH.
• Identifying key psychosocial factors like self-efficacy and social support can inform targeted interventions.
• This approach can potentially improve mental health care integration within routine HIV services.