Facial Emotion Recognition Predicts Alexithymia Using Machine Learning.

Area of Science:

• Psychiatry and Psychology
• Artificial Intelligence in Healthcare
• Computational Neuroscience

Background:

• Alexithymia, characterized by emotional processing deficits, impacts psychiatric disorder diagnosis.
• Traditional assessment tools for alexithymia have limitations, including insufficient insight and inconsistent results.
• Developing novel, objective screening tools for alexithymia is crucial for accurate diagnosis.

Purpose of the Study:

• To introduce a novel screening tool for alexithymia.
• To leverage machine learning (ML) models for alexithymia prediction.
• To utilize facial emotion recognition (FER) task scores as a basis for ML models.

Main Methods:

• A cross-sectional study involving 55 university students.
• Administration of the Toronto Alexithymia Scale (TAS-20), SCL-90-R, BAI, and BDI-II.
• Implementation of Support Vector Machine (SVM) and Feedforward Neural Network (FNN) classifiers with K-fold cross-validation on FER task data.

Main Results:

• ML models achieved an accuracy of 72.7-81.8% after feature selection and optimization.
• The models demonstrated the ability to distinguish individuals with and without alexithymia.
• Key predictive features for alexithymia were successfully identified.

Conclusions:

• ML models integrating FER task, SCL-90-R, BDI-II, and BAI scores can effectively diagnose alexithymia.
• These models identify influential factors for predicting alexithymia.
• The proposed method shows potential as a clinical instrument for earlier and more accurate diagnosis.