BACKGROUND: This study addresses the challenge of accurately classifying the severity of schizophrenia in patients through a clever approach. By leveraging electroencephalography (EEG) signals, we aim to establish a method for evaluating patient conditions, thereby contributing to the psychiatric diagnosis and treatment field. METHODS: Our research methodology encompasses a comprehensive system framework designed to analyze EEG signals with the Positive and Negative Syndrome Scale (PANSS) for correlation analysis. The process involves: (1) administering the PANSS test to create a database of schizophrenia patients; (2) developing a visual concentration test system that measures EEG signals in real-time; (3) processing these signals to construct an EEG feature database; (4) employing support vector machine and decision tree methods for illness severity classification; (5) conducting statistical analysis to correlate PANSS scores with EEG features, assessing the effectiveness of these correlations in clinical applications. RESULTS: The study successfully demonstrated the potential of a concentration detection system, integrating EEG signal analysis with PANSS scores, to classify schizophrenia severity accurately. Applying SVM and decision tree methods established significant correlations between EEG features and clinical scales, indicating the system's efficacy in supporting psychiatric diagnosis. CONCLUSIONS: Our findings suggest that the proposed analytical methods, focusing on EEG signals and employing a novel system framework, can effectively assist in classifying the severity of schizophrenia. This approach offers promising implications for enhancing diagnostic accuracy and tailoring treatment strategies for patients with schizophrenia.