Related Experiment Video
Updated: Aug 19, 2025
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
1.2K
Fingerprints as Predictors of Schizophrenia: A Deep Learning Study
Raymond Salvador1,2, María Ángeles García-León1,2, Isabel Feria-Raposo1,3,4
1FIDMAG Germanes Hospitalàries Research Foundation, Barcelona, Spain.
Schizophrenia Bulletin
|November 29, 2022
Summary
Fingerprints may serve as early predictors for psychosis. Deep learning models analyzing fingerprint patterns achieved up to 70% accuracy in distinguishing individuals with psychosis from healthy controls.
Area of Science:
- Neuroscience
- Biometrics
- Machine Learning
Background:
- Fingerprint development is linked to central nervous system growth, suggesting potential as schizophrenia risk markers.
- The complexity of fingerprint patterns necessitates advanced algorithms for accurate characterization.
Purpose of the Study:
- To investigate the utility of fingerprints as early predictors of psychosis.
- To develop and evaluate deep learning models for psychosis classification based on fingerprint data.
Main Methods:
- Convolutional neural network (CNN) based deep learning classification algorithms were developed.
- Models were trained on a dataset of 612 patients with non-affective psychosis and 844 healthy controls.
- A 5-fold cross-validation scheme was used to ensure unbiased accuracy estimates.
Main Results:
- The highest accuracy for single-finger models was 68% (right thumb).
- Multi-input models achieved a maximum accuracy of 70% using left thumb, index, and middle fingers.
- These results demonstrate the potential of fingerprint analysis in psychosis detection.
Conclusions:
- Lifelong stability of fingerprints suggests their application as early psychosis predictors.
- Fingerprint analysis could be particularly valuable in high-risk populations.
- The study highlights a novel approach for psychosis risk assessment.
