Evaluation of a Machine Learning-Based Dysphagia Prediction Tool in Clinical Routine: A Prospective Observational Cohort Study.
Stefanie Jauk, Diether Kramer, Sai Pavan Kumar Veeranki and 5 others
PMID 36625964WHAT IT FOUND
The dysphagia prediction tool worked better in internal medicine than geriatric wards.
It identified 74.2% of internal medicine dysphagia cases and 44.4% of geriatric cases.
Key findings
01In internal medicine admissions, the tool identified 74.2% of dysphagia cases and classified 84.1% of non-dysphagia admissions as low risk.
02In geriatric admissions, the tool identified 44.4% of dysphagia cases and classified 93.0% of non-dysphagia admissions as low risk.
03All patients predicted very high risk were examined by speech-language pathologists.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The geriatric ward had only 222 admissions, so its performance estimate was imprecise. Dysphagia status was taken from routine records, so some patients with dysphagia may have been missed. The evaluation happened during COVID-19 lockdowns, when hospital case mix and ward use were unusual. The study did not evaluate whether the tool prevented aspiration pneumonia or improved patient outcomes. The tool was evaluated in a secondary care hospital, with an internal medicine ward and a geriatric ward. The tool's predicted risks were slightly higher than the observed dysphagia frequency.
Declared interests
Funding was provided by the Medical University of Graz. The electronic health record data were hosted by KAGes, the regional healthcare provider.
The easy way to misread this
Do not conclude that this tool can safely rule out dysphagia in geriatric inpatients, because it identified only 44.4% of dysphagia cases in that ward.