Hybrid Approach for Brain Disease Diagnosis: Integrating Hidden Markov Models with Deep Learning (CNN–HMM)

Abstract

Accurate and early diagnosis of brain diseases particularly functional and psychological disorders such as anxiety, stress, and depression remains one of the major challenges in modern medicine. Traditional machine learning methods rely on manual and time‑consuming feature extraction, while deep learning models, despite their high accuracy, often lack the interpretability required for clinical applications due to their “black‑box” nature. This study proposes a novel hybrid approach for the diagnosis and classification of brain disorders based on the analysis of Electroencephalography (EEG) signals. The proposed model integrates the automatic feature extraction capability of Convolutional Neural Networks (CNNs) with the temporal modeling and interpretability strengths of Hidden Markov Models (HMMs). To develop this system, a native clinical dataset comprising EEG recordings from 200 subjects across four groups (healthy, anxiety, chronic stress, and depression) was utilized. The results demonstrate that the proposed hybrid model achieves a high accuracy of 92.5% in classifying the four disease categories, outperforming conventional methods, while simultaneously providing clinically meaningful interpretability through HMM transition matrices.