Search

Objective: To investigate the feasibility and accuracy of artificial intelligence (AI) methods of facial deidentification in hospital-recorded epileptic seizure videos, for improved patient privacy protection while preserving clinically important features of seizure semiology. Patients and Methods: Videos of epileptic seizures displaying seizure-related involuntary facial changes were selected from recordings at Taipei Veterans General Hospital Epilepsy Unit (between August 1, 2020 and February 28, 2023), and a single representative video frame was prepared per seizure. We tested 3 AI transformation models: (1) morphing the original facial image with a different male face; (2) substitution with a female face; and (3) cartoonization. Facial deidentification and preservation of clinically relevant facial detail were calculated based on: (1) scoring by 5 independent expert clinicians and (2) objective computation. Results: According to the clinician scoring of 26 facial frames in 16 patients, the best compromise between deidentification and preservation of facial semiology was the cartoonization model. A male facial morphing model was superior to the cartoonization model for deidentification, but clinical detail was sacrificed. Objective similarity testing of video data reported deidentification scores in agreement with the clinicians’ scores; however, preservation of semiology gave mixed results likely due to inadequate existing comparative databases. Conclusion: Artificial intelligence-based face transformation of medical seizure videos is feasible and may be useful for patient privacy protection. In our study, the cartoonization approach provided the best compromise between deidentification and preservation of seizure semiology.

Objectives: The Cribari Matrix Method (CMM) is the current standard to identify over/undertriage but requires manual trauma triage reviews to address its inadequacies. The Standardized Triage Assessment Tool (STAT) partially emulates triage review by combining CMM with the Need For Trauma Intervention, an indicator of major trauma. This study aimed to validate STAT in a multicenter sample., Methods: Thirty-eight adult and pediatric US trauma centers submitted data for 97,282 encounters. Mixed models estimated the effects of overtriage and undertriage versus appropriate triage on the odds of complication, odds of discharge to a continuing care facility, and differences in length of stay for both CMM and STAT. Significance was assessed at p <0.005., Results: Overtriage (53.49% vs. 30.79%) and undertriage (17.19% vs. 3.55%) rates were notably lower with STAT than with CMM. CMM and STAT had significant associations with all outcomes, with overtriages demonstrating lower injury burdens and undertriages showing higher injury burdens than appropriately triaged patients. STAT indicated significantly stronger associations with outcomes than CMM, except in odds of discharge to continuing care facility among patients who received a full trauma team activation where STAT and CMM were similar., Conclusions: This multicenter study strongly indicates STAT safely and accurately flags fewer cases for triage reviews, thereby reducing the subjectivity introduced by manual triage determinations. This may enable better refinement of activation criteria and reduced workload., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Author KWS declares that he is funded by the University of Arkansas for Medical Sciences Clinician Scientist Program. All other authors report no conflicts of interest or competing interests exist., (Copyright © 2020. Published by Elsevier Ltd.)