I-Corps: Machine Learning-Based Burn Injury Diagnosis and Care

Texas Tech University Health Science Center, Lubbock TX

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the development of machine learning technology to evaluate patients with skin burns. The proposed software tool is designed to determine burn severities and surface area percentages quickly and accurately. Currently, medical professionals use a visualized method to look at wounds and determine burn severities and to estimate burn percentage versus unburned skin. This may result in a wait of 3-7 days during which time doctors determine the burn degree by comparing pictures from the first day and following days of continuous monitoring. The proposed image processing and machine learning algorithm may provide clinicians with the ability to predict total burn surface area, expected resuscitation amount, delineate second- and third-degree burns, as well as prognosticate indeterminate-degree burns. The proposed technology may reduce ICU cost for both patients and hospitals, and may allow for better, more efficient patient care, better survival, and better functional outcomes. This I-Corps project is based on the development of a machine learning algorithm to process imaging from burn patients. The proposed assist tool is designed to determine the total body surface area of burns using an image capture technology. Image capture technology may be used to calculate total body surface area based on image recognition of burned areas, providing physicians and other medical personnel with more objective data on which to base resuscitation. In addition, the proposed technology may be used to delineate between full thickness and partial thickness of burn injuries. Also, it may be used to delineate between operative and non-operative indeterminate thickness burn injuries. Burn wounds are always evolving, and based on the depth of a burn injury, an operation may be required. The lengthy observation period with current visual methods exposes patients to infection, hospital-related comorbidity, and life-altering hospital cost. Using image capture technology and machine learning to analyze a standardized image of burn injuries to delineate between operative and non-operative burns may mitigate these risks. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.