JMIR Publications

ABSTRACT

Background:

Skin disease affects 2.3 billion people globally. Due to scarcity of dermatologists, 2 in 3 cases are seen by primary care physicians (PCPs) with lower diagnostic accuracy. Published studies have shown that the diagnostic accuracy of a primary care physician or general practitioners is close to 50%.

Objective:

Our objective was to build AI classifiers across 26 and 54 common and urgent adult rashes that present in a primary care setting.

Methods:

We trained our AI models with approximately 50,000 total photos. The number of images within each disease or class ranged from 76 to 5,505. Additionally, we further tested narrowing the differential by adding in body part information to identify how this impacts Top-5 accuracy for 1 condition.

Results:

Overall, we trained an AI model to identify 26 classes on-par with the accuracy level of a dermatologist, who is on average 75% Top-3 accurate across 26 conditions. Additionally, we trained the AI model across 54 conditions and achieved 74.3% Top-5 accuracy across common conditions and 79.2% Top-5 accuracy across urgent conditions. In evaluating if body part information may increase Top-5 accuracy, we saw Top-5 accuracy for 1 condition increase from 67% Top-5 accuracy to 97% Top-5 accuracy.

Conclusions:

Overall, we concluded that including body part to down-select possible disease matches substantially increased the overall differential diagnosis accuracy for body region-specific conditions. We also concluded that AI may assist primary care physicians to identify the most likely skin conditions quickly in a clinical encounter, to improve overall diagnostic accuracy and inform the most appropriate next step for the patient. These promising findings highlight the need and potential of artificial intelligence and clinical decision support to augment the ability of primary care physicians to accurately and confidently evaluate patients with skin conditions.