Overview of the team’s analysis pipeline. Credit: Translational Psychiatry (2025). DOI: 10.1038/s41398-025-03224-7
Major depressive disorder (MDD) is a debilitating mental health condition characterized by persistent low mood, loss of interest in everyday activities, appetite changes, sleep disturbances and, in extreme cases, suicidal thoughts. Millions of individuals worldwide have experienced a depressive episode throughout the course of their lives and reached out to psychiatrists seeking treatment.
There are currently many treatment options for depression, including different types of medications and psychotherapies, as well as targeted interventions combining both. While studies suggest that some of these treatments can be more effective than others, different people are known to often respond differently to available treatments.
Therefore, finding the medication and therapy that works best for a particular individual can take time, sometimes resulting in a long trial-and-error process. Developing effective strategies to identify the optimal treatment for specific patients early could thus be highly advantageous, as it would help them to recover faster, without having to try various medications that are not effective for them.
Researchers at the National University of Singapore and other institutes recently carried out a study exploring the possibility of predicting the response that people with MDD will have to specific treatments using a combination of functional near-infrared spectroscopy (fNIRS) and machine learning. Their paper, published in Translational Psychiatry, pin-points some biomarkers visible via fNIRS that appear to be correlated with how depressed individuals respond to treatments.
“Depression treatment responses vary widely among individuals,” wrote Cyrus Su Hui Ho, Jinyuan Wang and their colleagues in their paper. “Identifying objective biomarkers with predictive accuracy for therapeutic outcomes can enhance treatment efficiency and avoid ineffective therapies. This study investigates whether fNIRS and clinical assessment information can predict treatment response in MDD through machine-learning techniques.”
To explore the potential of fNIRS and clinical data for predicting people’s response to different treatments for depression, the researchers carried out a longitudinal study spanning across a period of six months. This study examined the responses to treatment of 70 people diagnosed with MDD, which were quantified using a questionnaire designed to help with diagnosing depression, known as the Hamilton Depression Rating Scale (HAM-D).
“fNIRS and clinical information were strictly evaluated using nested cross-validation to predict responders and non-responders based on machine-learning models, including support vector machine, random forest, XGBoost, discriminant analysis, Naïve Bayes, and transformers,” wrote Ho, Wang and their colleagues.
The researchers analyzed both fNIRS data and information collected from the patients during clinical assessments using various state-of-the-art machine learning models. Their analyses allowed them to uncover a biomarker linked with how people responded to treatments, namely the concentration of the protein hemoglobin in the dorsolateral prefrontal cortex (dlPFC), a brain region involved in executive functions (i.e., cognitive flexibility, working memory and decision-making).
“The task change of total hemoglobin (HbT), defined as the difference between pre-task and post-task average HbT concentrations, in the dlPFC is significantly correlated with treatment response (p < 0.005),” wrote the researchers.
“Leveraging a Naïve Bayes model, inner cross-validation performance (bAcc = 70% [SD = 4], AUC = 0.77 [SD = 0.04]) and outer cross-validation results (bAcc = 73% [SD = 3], AUC = 0.77 [SD = 0.02]) were yielded for predicting response using solely fNIRS data. The bimodal model combining fNIRS and clinical data showed inferior performance in outer cross-validation (bAcc = 68%, AUC = 0.70) compared to the fNIRS-only model.”
Overall, the results of this recent study highlight the potential of machine learning as a tool to explore the factors associated with how people respond to different treatments for mental health disorders.
In the future, this work could inspire other researchers to use machine learning to analyze clinical and fNIRS data, while also potentially contributing to the development of new protocols designed to rapidly identify optimal therapeutic interventions for individuals diagnosed with MDD.
More information:
Su Hui Ho et al, Application of functional near-infrared spectroscopy and machine learning to predict treatment response after six months in major depressive disorder, Translational Psychiatry (2025). DOI: 10.1038/s41398-025-03224-7
© 2025 Science X Network
Citation:
Using machine learning to predict how people diagnosed with major depressive disorder respond to treatment (2025, January 22)
retrieved 22 January 2025
from https://medicalxpress.com/news/2025-01-machine-people-major-depressive-disorder.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
