A deep joint-learning proteomics model for diagnosis of six conditions associated with dementia - PubMed
5 hours ago
- #proteomics
- #AI-diagnosis
- #dementia
- Introduces ProtAIDe-Dx, a deep joint-learning proteomics model for diagnosing six dementia-related conditions.
- Uses plasma proteomics from 17,187 patients/controls to provide simultaneous probabilistic diagnosis.
- Achieves cross-validated balanced classification accuracy of 70-95% and AUC >78% across all conditions.
- Model highlights subgroups with co-pathologies and links to pathology-specific biomarkers, even in individuals without cognitive impairment.
- Interpretation reveals protein networks marking shared and specific biological processes, identifying novel and known discriminating proteins.
- Significantly improves biomarker-based differential diagnosis in memory clinic samples, pinpointing proteins at an individual level.
- Demonstrates the promise of plasma proteomics for patient-level diagnostic workup with a single blood draw.