Machine learning algorithm diagnoses stroke with 83% accuracy
HospiMedica International Staff Writer
Posted on: April 11, 2023
Stroke is one of the most frequently misdiagnosed medical conditions and rapid detection is essential for effective treatment. Patients who receive treatment within her hour of onset have a better chance of survival and avoid long-term brain damage. Data show that blacks, Hispanics, women, Medicare seniors, and rural residents are less likely to be diagnosed during this critical period. is overlooked. Machine learning (ML) algorithms leverage social determinants of hospital and health data to predict stroke with 83% accuracy before clinical test results or diagnostic images are available, according to a new study. Rapid diagnosis has been shown. The findings suggest the potential to reduce stroke misdiagnosis and enhance patient monitoring. This allows medical staff to identify stroke patients or at-risk patients earlier and improve patient outcomes.
Researchers at Florida International University (Miami, Florida, USA) have developed a ML algorithm to improve stroke diagnosis using data from suspected stroke patients, including age, race, and number of underlying medical conditions. Social determinants of health (SDoH) are non-medical factors such as race, income, and housing stability that influence a wide range of health outcomes. The researchers used Florida hospital emergency department and admission records between 2012 and 2014 and combined them with her SDoH data from the American Community Survey to create an ML stroke prediction algorithm. Their analysis included 143,203 unique patient hospital visits. Patients diagnosed with stroke were typically older, had more chronic conditions, and were primarily dependent on Medicare.
Image: ML algorithms use social determinants of hospital and health data to diagnose stroke (photo credit: Freepik)
Using the researchers’ ML algorithm, when a patient arrives at the hospital with a stroke or stroke-like symptoms, an automated, computer-assisted screening tool quickly analyzes the patient’s information. If an algorithm predicts a high risk of stroke, a pop-up her alert is triggered for the emergency department team. Current ML methods, often focused on interpreting clinical notes and imaging findings, may not be available upon patient arrival, especially in rural and underserved communities. The technology is currently undergoing pilot testing in emergency departments of various well-known medical systems.
Min Chen, Associate Professor of Information Systems and Business Analytics at FIU Business and co-researcher, said: “Our algorithms can incorporate many variables to analyze and interpret complex patterns, so emergency department care teams can make better and faster decisions.”
Florida International University