Journal of affective disorders · 2026
Identifying neuroimaging biomarkers of adolescent major depressive disorder from cortical hemodynamic responses using machine learning approaches.
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.
Liu Xiaoli, Peng Ziyang, Cheng Fang, Hu Linling, Lu Changtai, Zhang Wenwu, Song Qinghai, Zhou Dongsheng
Аннотация
BACKGROUND: There is a pressing need to identify objective biomarkers for adolescent major depressive disorder (MDD). This study aimed to assist in identifying neuroimaging biomarkers based on cortical hemodynamic responses using machine learning.
METHOD: A total of 197 adolescents (114 adolescents with MDD and 83 healthy controls) completed both a verbal fluency task (VFT) and a working memory (WM) task during functional near-infrared spectroscopy (fNIRS) recording. Five machine learning algorithms were applied to build identification model.
RESULT: Adolescents with MDD showed reduced mean oxyhemoglobin (Oxy-Hb) activation in frontotemporal regions during both VFT and WM tasks, accompanied by altered slope feature. The combined VFT-WM model achieved the best performance, with Random Forest both reaching an area under the curve (AUC) of 0.93. The slope of the right superior frontal gyrus, derived from the working memory retrieval phase, was the most influential feature in the Random Forest model, with the highest contribution (mean |SHAP| = 0.07). Subgroup analysis showed improved performance in patients with longer disease duration (Random Forest and SVM AUC = 0.96).
CONCLUSION: WM-related prefrontal hemodynamic features, particularly retrieval-related dynamic responses, may serve as promising neuroimaging biomarkers of adolescent MDD. Integrating task-based fNIRS with interpretable machine learning may facilitate the objective identification of adolescent MDD. Given the inpatient sample, the model may be better suited for supporting diagnosis in clinically established cases rather than for early screening.
METHOD: A total of 197 adolescents (114 adolescents with MDD and 83 healthy controls) completed both a verbal fluency task (VFT) and a working memory (WM) task during functional near-infrared spectroscopy (fNIRS) recording. Five machine learning algorithms were applied to build identification model.
RESULT: Adolescents with MDD showed reduced mean oxyhemoglobin (Oxy-Hb) activation in frontotemporal regions during both VFT and WM tasks, accompanied by altered slope feature. The combined VFT-WM model achieved the best performance, with Random Forest both reaching an area under the curve (AUC) of 0.93. The slope of the right superior frontal gyrus, derived from the working memory retrieval phase, was the most influential feature in the Random Forest model, with the highest contribution (mean |SHAP| = 0.07). Subgroup analysis showed improved performance in patients with longer disease duration (Random Forest and SVM AUC = 0.96).
CONCLUSION: WM-related prefrontal hemodynamic features, particularly retrieval-related dynamic responses, may serve as promising neuroimaging biomarkers of adolescent MDD. Integrating task-based fNIRS with interpretable machine learning may facilitate the objective identification of adolescent MDD. Given the inpatient sample, the model may be better suited for supporting diagnosis in clinically established cases rather than for early screening.