Antidepressant Treatment Response Prediction With Early Assessment of Functional Near-Infrared Spectroscopy and...

Area of Science:

• Neuroscience
• Genetics
• Computational Biology

Background:

• Functional near-infrared spectroscopy (fNIRS) shows potential for diagnosing major depressive disorder (MDD).
• Predicting antidepressant treatment response (ATR) in MDD patients remains a clinical challenge.
• Integrating neuroimaging and genetic data may improve treatment prediction.

Purpose of the Study:

• To investigate the prediction of MDD ATR across three response levels using fNIRS and micro-ribonucleic acids (miRNAs).
• To develop and evaluate a novel algorithm for reducing inter-subject variability in predictive modeling.

Main Methods:

• A custom algorithm incorporating Principal Component Analysis (PCA) was used to reduce inter-subject variability.
• fNIRS and miRNA data were analyzed using a Radial Basis Function (RBF) Support Vector Machine (SVM).
• The algorithm was tested on groups of non-responders, partial-responders, and responders.

Main Results:

• The proposed algorithm achieved 82.70% accuracy, 78.44% sensitivity, 86.15% precision, and 91.02% specificity.
• This performance surpassed conventional machine learning approaches using combined clinical, sociodemographic, and genetic data.
• The custom algorithm effectively minimized inter-subject variability.

Conclusions:

• The fusion of fNIRS and miRNA data significantly enhances MDD ATR prediction accuracy.
• Addressing inter-subject variability is crucial for improving predictive models.
• This approach offers a practical tool for clinical decision support systems in personalized MDD treatment.