Gerçek zamanlı depresyon tahmini için makine öğrenmesi modelleri

Öz

Depresyon, dünya genelinde milyonlarca insanı etkileyen yaygın bir küresel sağlık sorunudur ve zamanında teşhisi, etkili müdahaleler için kritik öneme sahiptir. Geleneksel tanı yöntemleri çoğunlukla öznel klinik değerlendirmelere dayanmakta olup, bu durum sıklıkla geç teşhis ve yetersiz tedavi sonuçlarına yol açmaktadır. Bu çalışma, yapay zekâ (YZ) ile giyilebilir teknolojinin entegrasyonunu araştırarak depresyon şiddetinin gerçek zamanlı, sürekli ve nesnel olarak tahmin edilmesini sağlamayı amaçlamaktadır. Giyilebilir cihazlardan toplanan fizyolojik ve davranışsal veriler kullanılarak, Uzun Kısa Süreli Bellek (LSTM), Rastgele Orman (RF) ve Gradyan Artırma (GB) gibi gelişmiş makine öğrenmesi algoritmaları depresyon şiddetini sınıflandırmak için uygulanmıştır. Başlangıçta sınıflandırma beş kategori üzerinden yapılmış, ancak model performansını artırmak amacıyla daha sonra üç sınıfa indirgenmiştir. Doğruluk (accuracy), ve Ortalama Kare Hatası (MSE) gibi değerlendirme metrikleri, Gradient Boosting modelinin diğer yaklaşımlara kıyasla tutarlı şekilde üstün performans gösterdiğini ortaya koymuştur. Bu bulgular, yapay zekâ destekli giyilebilir platformların erken ve doğru depresyon tespiti sağlayarak kişiselleştirilmiş ruh sağlığı bakımı ve tedavi sonuçlarını iyileştirme konusunda dönüştürücü bir potansiyele sahip olduğunu vurgulamaktadır.

Anahtar Kelimeler

• Depresyon Tahmini
• Makine Öğrenmesi
• LSTM
• Gradient Boosting
• Rastgele Orman (Random Forest)

Machine learning models for real-time depression prediction

Öz

Depression is a pervasive global health challenge affecting millions worldwide, and its timely diagnosis remains critical for effective intervention. Traditional diagnostic approaches predominantly rely on subjective clinical assessments, which often lead to delayed detection and suboptimal treatment outcomes. This study investigates the integration of artificial intelligence (AI) with wearable technology to enable continuous, objective prediction of depression severity in real time. Leveraging physiological and behavioral data collected from wearable devices, advanced machine learning algorithms—including Long Short-Term Memory (LSTM), Random Forest (RF), and Gradient Boosting (GB)—were applied to classify depression severity. Initially, the classification framework comprised five categories but was subsequently streamlined to three classes to improve model performance. Evaluation using accuracy and Mean Squared Error (MSE) metrics revealed that the Gradient Boosting model consistently outperformed other approaches. These findings underscore the transformative potential of AI-enabled wearable platforms to facilitate early, accurate depression detection, thereby enhancing personalized mental health care and treatment outcomes.

Anahtar Kelimeler

• Depression Prediction
• Wearable De-vice
• Machine Learning
• LSTM
• Gradient Boosting
• Random Forest

Kaynakça

1. World Health Organization (WHO), Depression is one of the leading causes of disability worldwide, affecting over 280 million people. Available: https://www.who.int/news-room/fact-sheets/detail/mental-disorders/, [Accessed June 13, 2026].
2.     F. Zafar, L. F. Alam, R. R. Vivas, J. Wang, S. J. Whei, S. Mehmood, A. Sadeghzadegan, M. Lakkimsetti and Z. Nazir, The Role of Artificial Intelligence in Identifying Depression and Anxiety: A Comprehensive Literature Review, Cureus, 16(3), e56472 2024. https://doi.org/10.7759/cureus.56472
3.     A. Olyanasab and M. Annabestani, Leveraging Machine Learning for Personalized Wearable Biomedical Devices: A Review, Journal of Personalized Medicine, 14(2), 203, 2024. https://doi.org/10.3390/jpm14020203
4.     N. Prasad, I. Chien, T. Regan, A. Enrique, D. Keegan, J. Palacios, U. Munir, R. Tanno, H. Richardson, A. Nori, D. Richards, G. Doherty, D. Belgrave and A. Thieme, Deep learning for the prediction of clinical outcomes in internet-delivered CBT for depression and anxiety, PLOS One, 18(11), e0272685, 2023. https://doi.org/10.1371/journal.pone.0272685
5.     N. Narziev, H. Goh, K. Toshnazarov, S. A. Lee, K. M. Chung and Y. Noh, STDD: Short-Term Depression Detection with Passive Sensing, Sensors, 20(5), 1396, 2020. https://doi.org/10.3390/s20051396
6.     K. Kroenke, R. L. Spitzer and J. B. W. Willams, The PHQ-9: validity of a brief depression severity measure, Journal of General Internal Medicine, 16(9), 606-613, 2001. https://doi.org/10.1046/j.1525-1497.2001.016009606.x
7.     A. A. Abd-Alrazaq, R. AlSaad, F. Shuweihdi, A. Ahmed, S. Aziz and J. Sheikh, Systematic review and meta-analysis of performance of wearable artificial intelligence in detecting and predicting depression, NPJ Digital Medicine, 6(1), 84, 2023. https://doi.org/10.1038/s41746-023-00828-5
8.     A. Abd-alrazaq, R. AlSaad, S. Aziz, A. Ahmed, K. Denecke, M. Househ, F. Farooq and J. Sheikh, Wearable Artificial Intelligence for Anxiety and Depression: Scoping Review, Journal of Medical Internet Research, 25, e42672, 2023. https://doi.org/10.2196/42672

9.     T. Mullick, A. Radovic, S. Shaaban and A. Doryab, Predicting Depression in Adolescents Using Mobile and Wearable Sensors: Multimodal Machine Learning–Based Exploratory Study, JMIR Formative Research, 6(6), e35807, 2022. https://doi.org/10.2196/35807
10.   A. Pavlopoulos, T. Rachiotis and I. Maglogianni, An Overview of Tools and Technologies for Anxiety and Depression Management Using AI, Applied Sciences, 14(19), 9068, 2024. https://doi.org/10.3390/app14199068
11.   D. Patel, S. A. Sumner, D. Bowen, M. Zwald, E. Yard, J. Wang, R. Law, K. Holland, T. Nguyen, G. Mower, Y. Chen, J. I. Johnson, M. Jespersen, E. Mytty, J. M. Lee and M. Bauer, Predicting state level suicide fatalities in the united states with realtime data and machine learning, npj Mental Health Research, 3(1), 3, 2024. https://www.nature.com/articles/s44184-023-00045-8
12.   X. Shui, H. Xu, S. Tan and D. Zhang, Depression Recognition Using Daily Wearable-Derived Physiological Data, Sensors, 25(2), 567, 2025. https://doi.org/10.3390/s25020567
13.   S. Burchert, A. Kerber, J. Zimmermann and C. Knaevelsrud, 14-day smartphone ambulatory assessment of depression symptoms and mood dynamics in a general population sample: comparison with the PHQ-9 depression screening, Frontiers in Psychology, 10, 704 2019.
14.   S. Burchert, A. Kerber, J. Zimmermann and C. Knaevelsrud, Screening accuracy of a 14-day smartphone ambulatory assessment of depression symptoms and mood dynamics in a general population sample: Comparison with the PHQ-9 depression screening, PLOS One, 16(1), e0244955, 2021. https://doi.org/10.1371/journal.pone.0244955
15.   A. S. Yasin and H. Osmanoğlu, Predicting Stock Market Trends Using Regression and Deep Learning, International Conference on Computing, Intelligence and Data Analytics, pp. 631-644, 2025. https://doi.org/10.1007/978-3-031-87154-2_50
16.   W. Z. T. Tareq, Sleep Disorders Detection and Classification Using Random Forests Algorithm, In Allahviranloo, T., Hosseinzadeh Lotfi, F., Moghaddas, Z., Vaez-Ghasemi, M. (eds.) Decision Making in Healthcare Systems. Studies in Systems, Decision and Control, Springer, Cham, pp. 257-266, 2024. https://doi.org/10.1109/SUSTAINED63638.2024.11074079
17.   W. Z. T. Tareq and M. Davud, Classification and clustering, In Decision Making Models, Academic Press, pp. 351-359, 2024. https://doi.org/10.1016/B978-0-443-16147-6.00024-4
18.   D. Stephen, A. A. Attallah and M. Anugerah Ayu, Improving Early Detection and Prevention of Depression Using an Interactive Mobile App, 2022 IEEE 8th International Conference on Computing, Engineering and Design (ICCED), pp. 1-6, Sukabumi, Indonesia, 2022. https://doi.org/10.1109/ICCED56140.2022.10010529