Explainable and externally validated machine learning for neurocognitive diagnosis via ECGs

Journal article

Background: Electrocardiogram (ECG) analysis has emerged as a promising tool for detecting physiological changes linked to non-cardiac disorders. Given the close connection between cardiovascular and neurocognitive health, ECG abnormalities may be present in individuals with co-occurring neurocognitive conditions. This highlights the potential of ECG as a biomarker to improve detection, therapy monitoring and risk stratification in patients with neurocognitive disorders, an area that remains underexplored. Aims: We aimed to demonstrate the feasibility of predicting neurocognitive disorders from ECG features across diverse patient populations. Methods: ECG features and demographic data were used to predict neurocognitive disorders, as defined by the International Classification of Diseases 10th revision, focusing on dementia, delirium and Parkinson’s disease. Internal and external validations were performed using the Medical Information Mart for Intensive Care IV and ECG-View datasets. Predictive performance was assessed by the area under the receiver operating characteristic curve (AUROC) scores, and Shapley values were used to interpret feature contributions. Results: Significant predictive performance was observed for several neurocognitive disorders. The highest predictive performance was observed for F03: dementia, with an internal AUROC of 0.848 (95% confidence interval (CI) 0.848 to 0.848) and an external AUROC of 0.865 (95% CI 0.864 to 0.965), followed by G30: Alzheimer’s disease, with an internal AUROC of 0.809 (95% CI 0.808 to 0.810) and an external AUROC of 0.863 (95% CI 0.863 to 0.864). Feature importance analysis revealed both established and novel ECG correlates. Conclusions: These findings suggest that ECG holds promise as a non-invasive, explainable biomarker for selected neurocognitive disorders. This study demonstrates robust performance across cohorts and lays the groundwork for future clinical applications, including early detection and personalised monitoring.

Authors: Lopez Alcaraz, JM; Oloyede, E; Taylor, D; Haverkamp, W; Strodthoff, N

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: BMJ Publishing Group
• Journal: General Psychiatry
• Volume: 38
• Issue: 5
• Article number: gpsych-2025-102107
• Publication date: 2025-10-23
• Acceptance date: 2025-09-14
• DOI: 10.1136/gpsych-2025-102107
• EISSN: 2517-729X
• ISSN: 2096-5923
• Language: English
• Keywords: Models, Statistical; Neurocognitive Disorders; Diagnosis, Dual (Psychiatry); Neuropsychiatry