JMIR Publications

ABSTRACT

Background:

Dysphagia affects more than half of older adults with dementia and is associated with a 10-fold increase in mortality. The development of accessible, objective, and reliable screening tools is crucial for early detection and management.

Objective:

This systematic scoping review aims to examine the current state-of-the-art in artificial intelligence (AI) and sensor-based technologies for dysphagia screening.

Methods:

We conducted a systematic literature search across CINAHL, Embase, PubMed, and Web of Science, performed by two independent researchers. The findings were reported in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) framework. The methodological quality of the eligible articles was assessed using the QUADAS-2 tool, to which we added a “model” domain for a more comprehensive evaluation. Data extracted from the included studies were synthesized narratively.

Results:

The review included 24 studies involving 2,979 participants. Most studies focused on per-individual classification rather than per-swallow event classification. Acoustic (n = 13) and vibratory signals (n = 9) are the primary modality sources utilized in dysphagia screening. Additional sources included nasal airflow, electromyography (EMG), strain and motion analysis, and optical methods. Notably, six studies employed multimodal approaches, while the remaining 18 studies focused on a single modality. Machine learning models, particularly support vector machines (n = 13), were frequently utilized, while deep learning approaches have been gaining traction. Performance metrics varied widely across studies, with some reporting high accuracy and AUC values, despite small testing samples. Multimodal systems appeared to perform better than unimodal systems. The methodological quality assessment revealed a high risk of bias in many studies, particularly in patient sampling, blinding procedures, and model development. The lack of external validation and domain adaptation testing raises concerns about the transferability and real-world applicability of these AI-based systems.

Conclusions:

This review provides an overview of technological advancements in the use of artificial intelligence and sensors for dysphagia screening. These promising developments in screening tools pave the way for continuous long-term tele-swallowing assessments.