Machine learning in obsessive-compulsive disorder medications.

Area of Science:

• Psychiatry
• Computational Biology
• Pharmacology

Background:

• Obsessive-compulsive disorder (OCD) is a prevalent psychiatric condition with substantial morbidity.
• Existing treatments for OCD often yield suboptimal responses in a significant patient subset.
• Novel approaches are needed to refine medication selection and treatment strategies for OCD.

Purpose of the Study:

• To classify medications used for obsessive-compulsive disorder (OCD) utilizing machine learning (ML) techniques.
• To compare the classification performance of Decision Tree (DT), Chi-Square Automatic Interaction Detection (CHAID), and linear models.
• To identify key medications and potential therapeutic adjuncts for OCD based on ML analysis.

Main Methods:

• A descriptive analytical study employing co-word analysis and artificial intelligence methods.
• Extraction of 116 drugs from 6574 articles using co-word analysis.
• Development and evaluation of Decision Tree (DT) models for medication classification, using Gini index for hyperparameter optimization.

Main Results:

• 56 drugs were classified by the DT model, categorized into six groups.
• ML analysis highlighted the efficacy of specific medications including clomipramine, duloxetine, and pindolol.
• Folate was identified as a potentially beneficial supplement, and treating comorbidities like hypothyroidism and streptococcal infections could enhance treatment outcomes.

Conclusions:

• Machine learning offers a valuable framework for classifying OCD medications and understanding treatment efficacy.
• The study identified specific pharmacological agents and highlighted the importance of managing comorbid conditions for improved OCD treatment.
• DT and other ML models provide insights into optimizing therapeutic interventions for obsessive-compulsive disorder.