Back

Can Artificial Intelligence Improve the Appropriate Use and Decrease the Misuse of REBOA?

Ma, Y.; Bokenkamp, M.; Dorken-Gallastegi, A.; Zamudio, J. P.; Gebran, A.; Velmahos, G. C.; Bertsimas, D.; Kaafarani, H.

2024-06-07 health informatics
10.1101/2024.06.06.24308557 medRxiv
Show abstract

BACKGROUNDThe use of resuscitative endovascular balloon occlusion of the aorta (REBOA) for control of noncompressible torso hemorrhage remains controversial. We aimed to utilize a novel and transparent/interpretable artificial intelligence (AI) method called Optimal Policy Trees (OPT), to improve the appropriate use and decrease the misuse of REBOA in hemodynamically unstable blunt trauma patients. METHODSWe trained then validated OPTs that "prescribe" REBOA in a 50:50 split on all hemorrhagic shock blunt trauma patients in the 2010-2019 ACS-TQIP database based on rates of survival. Hemorrhagic shock was defined as a systolic blood pressure [&le;] 90 on arrival or transfusion requirement of [&ge;] 4 units of blood in the first 4 hours of presentation. The expected 24-hour mortality rate following OPT prescription was compared to the observed 24-hour mortality rate in patients who were or were not treated with REBOA. RESULTSOut of 4.5 million patients, 100,615 were included and 803 underwent REBOA. REBOA patients had a higher rate of pelvic fracture, femur fracture, hemothorax, pneumothorax, and thoracic aorta injury (p<0.001). The 24-hour mortality rate for the REBOA vs. non-REBOA group was 47% vs. 21%, respectively (p<0.001). OPTs resulted in an 18% reduction in 24-hour mortality for REBOA and 0.8% reduction in non-REBOA patients. CONCLUSIONInterpretable AI models can improve mortality in unstable blunt trauma patients by optimizing the use and decreasing the misuse of REBOA. These models to date have been used to predict outcomes, but their groundbreaking use will be prescribing interventions and changing outcomes. LEVEL OF EVIDENCELevel IV, Prognostic

Published in Bioengineering · training set

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.