Related Experiment Videos
Identification and evaluation of cognitive deficits in schizophrenia using "Machine learning"
Antonella Vacca1, Roberto Longo, Corrado Mencar
1Città Solidale Società Cooperativa Sociale, Via Anania Lamarina, 75, 72022 Latiano (BR), Italy, antonellavacca@yahoo.com.
Psychiatria Danubina
|September 7, 2019
Summary
Machine learning accurately identifies schizophrenia's key cognitive deficits, including working memory and executive functions. This approach aids early diagnosis and intervention for schizophrenia patients.
Area of Science:
- Neuroscience
- Cognitive Science
- Artificial Intelligence
Background:
- Schizophrenia is characterized by neocortical pathology and persistent cognitive dysfunction, significantly impacting psychosocial functioning.
- Cognitive impairments in schizophrenia are crucial predictors of disease severity and disability.
- Neurocognitive tests are vital for assessing schizophrenia's impact on daily life.
Purpose of the Study:
- To identify specific cognitive deficits in schizophrenia using machine learning.
- To develop a predictive system for diagnosing schizophrenia based on neurocognitive test data.
Main Methods:
- Utilized a dataset of 201 participants (86 schizophrenia, 115 healthy) from the University of Bari.
- Applied machine learning algorithms including Decision Tree, Random Forest, Logistic Regression, k-NN, Neural Network, and SVM.
- Performed feature selection and stratified 20-fold cross-validation to evaluate classification accuracy.
Main Results:
- Identified 14 key influential variables, primarily related to working memory, executive functions, attention, and verbal fluency.
- Support Vector Machine (SVM) achieved 87.8% accuracy, and Neural Network achieved 84.8% accuracy on the test set.
Conclusions:
- Machine learning offers cost-effective, non-invasive methods for early schizophrenia detection and intervention.
- Integrating comprehensive neuropsychological evaluations is essential for schizophrenia diagnosis.