سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / اولین همایش بین المللی تحول خدمات سلامت: تلفیق مراقبت، درمان، بازتوانی و امیدبخشی

Anxiety Disorder Detection Using Digital Biomarkers and Machine Learning: A Comparative Study of Logistic Regression and Naïve Bayes

نویسندگان :

Parsa Asadnezhad¹ Mohammad Ali Ranjbar² Foad Ghaderi³

1- دانشگاه تربیت مدرس 2- دانشگاه تربیت مدرس 3- دانشگاه تربیت مدرس

کلمات کلیدی :

Anxiety Disorder Detection،Digital Biomarkers،Machine Learning،Logistic Regression،Naïve Bayes

چکیده :

Abstract: Background: Anxiety disorder is among the most prevalent mental health conditions worldwide, and its timely, objective diagnosis remains a critical clinical challenge. Recent advances in machine learning have enabled more effective analysis of complex biomedical and behavioral data for mental health assessment. Method: This study presents a data driven framework for the automated detection of anxiety disorder using classical, lightweight machine learning classifiers. Physiological and behavioral features were extracted from a publicly available social anxiety dataset (10,000+ samples), followed by systematic preprocessing and feature selection. Two supervised algorithms—Logistic Regression and Naïve Bayes—were implemented and compared. These lightweight models were chosen for their high interpretability and low computational cost, which are essential for clinical deployment and scalable screening. Performance was assessed using accuracy, precision, recall, F1 score, and ROC AUC. Results: Logistic Regression achieved the best performance with approximately 93% accuracy and an ROC AUC of 0.954. Naïve Bayes also yielded competitive results (accuracy ≈93%, AUC = 0.948) with even lower computational complexity. Conclusion: Lightweight, interpretable machine learning models can reliably identify anxiety related patterns. Their low computational requirements make them particularly suitable for integration into wearable devices and mobile health platforms for real time, continuous monitoring. The proposed framework offers a practical, scalable solution for computer aided anxiety assessment and has the potential to support clinical decision making through early detection and personalized mental healthcare.