Findings based on smartphone home videos from children younger than 2 years
By Lori Solomon HealthDay Reporter
WEDNESDAY, Dec. 18, 2024 (HealthDay News) — Deep learning analysis can be used to detect epileptic spasms (ES) from smartphone home videos of young children, according to a study presented at the annual meeting of the American Epilepsy Society, held from Dec. 6 to 10 in Los Angeles.
Gadi Miron, M.D., from Charitè-Universitätsmedizin Berlin, and colleagues evaluated if artificial intelligence (AI)-based analysis of videos could be used for automated detection of ES from home videos. The analysis included smartphone videos (991 seizures and 597 nonseizure five-second video segments) from 152 infants (younger than 2 years) with ES.
The researchers reported detection of ES with an area under the receiver operating characteristic curve (AUC) of 0.94, sensitivity of 78 percent, specificity of 85 percent, and accuracy of 81 percent. Estimation of false alarm rate (FAR) also included videos from 67 normal children (666 video segments) and yielded an FAR of 5.6 percent, corresponding to 37 of 666 segments wrongly classified. With a threshold of 90 percent sensitivity on the validation data, the AUC was 0.94, sensitivity 88 percent, specificity 66 percent, accuracy 79 percent, and FAR 7.3 percent on the out-of-sample test data.
“This technique could also aid doctors by offering additional data for patient evaluations and potentially expedite referrals for diagnostic tests, leading to quicker treatment decisions,” principal investigator Christian Meisel, M.D., Ph.D., also from Charité-Universitätsmedizin Berlin, said in a statement.
Copyright © 2024 HealthDay. All rights reserved.
