Ann Biomed Eng. 2026 Jul 1. doi: 10.1007/s10439-026-04267-7. Online ahead of print.
ABSTRACT
Post-stroke dysphagia (PSD) affects approximately 42% of acute stroke patients, increasing hospitalization costs and length of stay. Early identification improves outcomes, yet many patients-especially in low-resource settings-lack access to gold-standard evaluations. This scoping review explores the integration of artificial intelligence (AI) and data science-defined as the interdisciplinary use of computational methods, statistical modeling, and machine learning to extract clinically meaningful patterns from biomedical data-in PSD screening and assessment. We synthesize evidence from bedside screening instruments, acoustic voice analyses, and emerging AI-driven models for dysphagia and aspiration risk stratification, critically appraising limitations related to small datasets, overfitting risk, and the need for external validation. Traditional tools like the water swallow test show high sensitivity but varying specificity; recent studies support augmenting these with voice-based biomarkers such as post-swallow wet voice, jitter, and shimmer. While wet voice as a standalone marker has limited sensitivity (8-29%), its high specificity (75-94%) within multimodal approaches justifies continued investigation. AI models trained on acoustic parameters have demonstrated strong performance in detecting penetration-aspiration events, while mobile and voice-based platforms may expand diagnostic reach, pending further validation. We also review optimal screening timing, emphasizing assessment within 24 h of stroke onset with repeated evaluations for high-risk patients. Future directions advocate multimodal, patient-centered approaches combining wearable biosensors, cloud-based analytics, and culturally adapted algorithms, while addressing implementation challenges including infrastructure requirements, digital literacy, workflow integration, and ethical considerations. The convergence of clinical expertise and computational technologies presents a promising path to equitable, scalable, and precise dysphagia care.
PMID:42384315 | DOI:10.1007/s10439-026-04267-7
