Related Experiment Video
Updated: Oct 25, 2025
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
1.4K
Sch-net: a deep learning architecture for automatic detection of schizophrenia
Biomedical Engineering Online
|August 4, 2021
Summary
This study introduces Sch-net, a novel deep learning model for detecting schizophrenia through speech analysis. Sch-net achieves high accuracy, offering an objective method for diagnosing schizophrenia and specific language impairment.
Area of Science:
- Artificial Intelligence
- Machine Learning
- Computational Linguistics
Background:
- Schizophrenia diagnosis is challenging, often relying on subjective clinical experience, especially for negative symptoms.
- Current methods for detecting schizophrenia via speech analysis are hindered by the difficulty of manual feature extraction from variable speech signals.
Purpose of the Study:
- To develop an objective and effective deep learning method for diagnosing schizophrenia using speech patterns.
- To overcome limitations of traditional feature engineering in schizophrenic speech detection.
Main Methods:
- A novel convolutional neural network, Sch-net, was designed for end-to-end schizophrenic speech detection.
- Sch-net incorporates skip connections for feature fusion and a convolutional block attention module (CBAM) for feature weighting, avoiding manual feature engineering.
Main Results:
- Sch-net achieved 97.68% accuracy in detecting schizophrenia from speech data.
- The model demonstrated generalization by achieving 99.52% accuracy in classifying specific language impairment (SLI) in children.
Conclusions:
- The proposed Sch-net model offers a promising, data-driven approach for aiding the diagnosis of schizophrenia and SLI.
- This deep learning technique can provide objective, aided diagnostic information, improving upon current subjective methods.
