Identifying patients with major depressive disorder based on tryptophan hydroxylase-2 methylation using machine learning algorithms.

Objectives: This study aimed to identify patients with major depressive disorder (MDD) by developing different machine learning (ML) models based on tryptophan hydroxylase-2 (TPH2) methylation and environmental stress.

Methods: The data were collected from 291 patients with MDD and 100 healthy control participants: individual basic information, the Negative Life Events Scale (NLES) scores, the Childhood Trauma Questionnaire (CTQ) scores and the methylation level at 38 CpG sites in TPH2. Information gain was used to select critical input variables. Support vector machine (SVM), back propagation neural network (BPNN) and random forest (RF) algorithms were used to build recognition models, which were evaluated by the 10-fold cross-validation. SHapley Additive exPlanations (SHAP) method was used to evaluate features importance.

Results: Gender, NLES scores, CTQ scores and 13 CpG sites in TPH2 gene were considered as predictors in the models. Three ML algorithms showed satisfactory performance in predicting MDD and the BPNN model indicated best prediction effects.

Conclusion: ML models with TPH2 methylation and environmental stress were identified to possess great performance in identifying patients with MDD, which provided precious experience for artificial intelligence to assist traditional diagnostic methods in the future.