Scholarlink

Schizophrenia is a complex neuropsychiatric disorder that significantly impairs cognitive function, perception, and social behavior. Early and accurate diagnosis is crucial for effective treatment, yet traditional diagnostic methods remain subjective and time-consuming. This study presents a hybrid deep learning approach that leverages BERT (Bidirectional Encoder Representations from Transformers) for feature extraction and XGBoost (Extreme Gradient Boosting) for classification to analyze EEG signals for schizophrenia detection. The proposed model extracts meaningful representations from EEG data using BERT, while XGBoost performs the final classification, optimizing predictive performance. Additionally, data augmentation techniques are applied to enhance generalization. Experimental results demonstrate an accuracy exceeding 80%, indicating the potential of transformer-based models in EEG-based mental health diagnostics. To ensure interpretability, multiple visualizations, including confusion matrices and dataset distributions, are provided. This study underscores the effectiveness of hybrid AI models in improving the objectivity and efficiency of schizophrenia diagnosis. We got an accuracy of 72% with f1 score of 0.75 for detecting schizophrenia.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimization-Based Deep Learning Approaches for Efficient Schizophrenia Detection Technique

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Namdeo Badhe,
  • Navin Bohara,
  • Yog Dalal,
  • Riddhi Dayma,
  • Khojiakbar Egamberdiev

摘要

Schizophrenia is a complex neuropsychiatric disorder that significantly impairs cognitive function, perception, and social behavior. Early and accurate diagnosis is crucial for effective treatment, yet traditional diagnostic methods remain subjective and time-consuming. This study presents a hybrid deep learning approach that leverages BERT (Bidirectional Encoder Representations from Transformers) for feature extraction and XGBoost (Extreme Gradient Boosting) for classification to analyze EEG signals for schizophrenia detection. The proposed model extracts meaningful representations from EEG data using BERT, while XGBoost performs the final classification, optimizing predictive performance. Additionally, data augmentation techniques are applied to enhance generalization. Experimental results demonstrate an accuracy exceeding 80%, indicating the potential of transformer-based models in EEG-based mental health diagnostics. To ensure interpretability, multiple visualizations, including confusion matrices and dataset distributions, are provided. This study underscores the effectiveness of hybrid AI models in improving the objectivity and efficiency of schizophrenia diagnosis. We got an accuracy of 72% with f1 score of 0.75 for detecting schizophrenia.