Frontiers in psychiatry · 2026
Patient complexity profiles in depression: a machine learning approach to personalized mental health.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Dagnino Paula, Salinas Matias, Salas Rodrigo
Аннотация
BACKGROUND: Patient complexity in mental health varies substantially, yet treatment approaches often rely on standardized protocols. Identifying distinct complexity profiles may support stratified care and more personalized intervention planning in depression.
OBJECTIVE: To identify distinct patient complexity profiles in depression across sociodemographic, clinical, and psychosocial indicators and to evaluate their clinical relevance for personalized treatment planning.
METHODS: We analyzed complete-case data from 270 adults with major depression using a Knowledge Discovery in Databases framework. Twelve indicators were analyzed via Principal Component Analysis followed by K-means clustering. Robustness was evaluated using supervised validation with a Random Forest classifier and SHAP-based interpretability analysis. Between-profile comparisons were conducted, and expert clinicians evaluated clinical relevance.
RESULTS: Model selection supported a three-cluster solution (k = 3: Low-, Moderate-, and High-complexity profiles). The solution was validated using a Random Forest classifier with strong performance (accuracy = 0.91). Statistical comparisons showed that the Low-complexity profile (n = 100, 37.0%) was older and more often partnered and employed, with lower depressive symptoms and better personality functioning. The Moderate-complexity profile (n = 87, 32.2%) was younger, predominantly unpartnered, and had the lowest employment rate and medical comorbidity. The High-complexity profile (n = 83, 30.7%) showed the most severe presentation, characterized by higher depressive symptoms, greater childhood maltreatment, and impaired personality functioning. Clinical experts confirmed interpretability and suggested tailored strategies for each profile.
CONCLUSIONS: Machine learning identified clinically meaningful patient complexity profiles with significant differences across multiple domains. These profiles provide a framework for stratified care and personalized intervention planning, moving beyond one-size-fits-all approaches.
OBJECTIVE: To identify distinct patient complexity profiles in depression across sociodemographic, clinical, and psychosocial indicators and to evaluate their clinical relevance for personalized treatment planning.
METHODS: We analyzed complete-case data from 270 adults with major depression using a Knowledge Discovery in Databases framework. Twelve indicators were analyzed via Principal Component Analysis followed by K-means clustering. Robustness was evaluated using supervised validation with a Random Forest classifier and SHAP-based interpretability analysis. Between-profile comparisons were conducted, and expert clinicians evaluated clinical relevance.
RESULTS: Model selection supported a three-cluster solution (k = 3: Low-, Moderate-, and High-complexity profiles). The solution was validated using a Random Forest classifier with strong performance (accuracy = 0.91). Statistical comparisons showed that the Low-complexity profile (n = 100, 37.0%) was older and more often partnered and employed, with lower depressive symptoms and better personality functioning. The Moderate-complexity profile (n = 87, 32.2%) was younger, predominantly unpartnered, and had the lowest employment rate and medical comorbidity. The High-complexity profile (n = 83, 30.7%) showed the most severe presentation, characterized by higher depressive symptoms, greater childhood maltreatment, and impaired personality functioning. Clinical experts confirmed interpretability and suggested tailored strategies for each profile.
CONCLUSIONS: Machine learning identified clinically meaningful patient complexity profiles with significant differences across multiple domains. These profiles provide a framework for stratified care and personalized intervention planning, moving beyond one-size-fits-all approaches.