JMIR Publications

ABSTRACT

Background:

Burn injuries are challenging to treat due to the critical importance of accurately assessing burn depth for appropriate therapy. Traditional assessment largely depends on visual inspection by experienced surgeons. However, this method can be prone to errors. Inaccurate assessments can lead to inappropriate treatments, either delaying necessary care for severe burns or leading to unnecessary procedures for milder cases.

Objective:

The objective is to determine whether a multi-modal integrated AI system for accurate classification of burn depth, can preserve diagnostic accuracy and provide an important resource when utilized as part of the Electronic Medical Record (EMR).

Methods:

The study integrates digital photographs, and ultrasound Tissue Doppler Imaging (TDI) data from an EMR to predict burn depth leveraging vision-language model. Digital photographs were analyzed to rapidly classify burns as first-degree or superficial second-degree. For burns potentially classified as second-degree deep or third-degree, TDI data was incorporated for enhanced accuracy.

Results:

This study evaluated whether a classifier designed to identify 1st, 2nd, and 3rd degree burns can be used on data housed within the EMR. The classifier achieved an accuracy of 84.38%, surpassing human assessment of depth prediction, which has been reported at 76% for experts and 50% for non-expert surgeons.

Conclusions:

The storage of multi-modal imaging data within the EMR, along with the ability for post hoc analysis by AI algorithms, offers significant advancements in burn care, enabling real-time burn depth prediction on currently available data. Using digital photos for superficial burns, easily diagnosed through physical exams, reduces reliance on TDI, while TDI helps distinguish deep second- and third-degree burns, enhancing diagnostic efficiency. Clinical Trial: NCT05167461, https://clinicaltrials.gov/study/NCT05167461