Schizophrenia Diagnosis from Resting-State FMRI Using a Hybrid Machine Learning Approach Based on Functional Connectivity Patterns
8th International Congress on Human-Computer Interaction, Optimization and Robotic Applications, ICHORA 2026, Ankara, Türkiye, 21 - 23 Mayıs 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/ichora69329.2026.11537243
- Basıldığı Şehir: Ankara
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: COBRE, dynamic functional connectivity, machine learning, rs-fMRI, Schizophrenia, stacking
- Yozgat Bozok Üniversitesi Adresli: Evet
Özet
Schizophrenia is a chronic neuropsychiatric disorder with complex clinical symptoms and a heterogeneous neurobiological basis, which renders diagnostic processes particularly challenging. In this study, the discriminative power of dynamic functional connectivity (dFC) patterns derived from resting-state functional magnetic resonance imaging (rs-fMRI) data was investigated for distinguishing schizophrenia patients from healthy controls. Following rigorous quality control and motion scrubbing on the COBRE dataset, regional time series were extracted using the AAL2 atlas, and dynamic connectivity features were obtained via a sliding window approach. The resulting high-dimensional dFC representations were classified using a stacking-based hybrid machine learning framework that integrates Support Vector Machines (SVM) and XGBoost. Model evaluation was performed using a subject-level, GroupKFoldbased five-fold cross-validation strategy to prevent data leakage. The proposed approach demonstrated higher and more stable performance in terms of accuracy and ROC-AUC compared to single classifiers. These findings indicate that subject-level aggregated dFC representations combined with stacking-based hybrid learning strategies can provide a reliable and generalizable framework for schizophrenia classification.