New Artificial Intelligence-Based App Detects Mpox Rashes

Researchers from Stanford Medicine and other organizations have developed an artificial intelligence-based app that analyzes images of skin lesions to identify mpox rashes with 90 percent accuracy:  Researchers from Stanford University and other institutions have created an artificial intelligence (AI)-based app, called PoxApp, that can capture photos of skin lesions and provide insight as to whether they were caused by mpox.

Previously known as monkeypox, mpox is a rare disease that results from an mpox virus infection, according to the Centers for Disease Control and Prevention(CDC).  The CDC also noted that in 2022, there were 30,225 recorded cases of mpox in the US and 38 deaths.

To enable early detection of mpox, researchers created PoxApp, using a dataset of 130,000 images of various skin conditions to train the AI the app uses.  This app allows patients to capture photos of suspicious skin lesions, answer some questions, and receive a risk score along with suggestions for testing or vaccination.  Researchers noted that this process could take place within five minutes, according to the press release.  “It’s a quick, easy and anonymous way to get a first assessment,” said Alexander Thieme, MD, the Lead Developer of the App and a Visiting Scholar in the Department of Medicine from Charité at the Universitätsmedizin Berlin and Berlin Institute of Health, in a press release.  “We are hoping to increase the likelihood that someone sees a doctor due to their skin lesions rather than ignore it.”

When calculating the risk score, the app takes into account several factors, including whether the patient has a skin lesion, as well as their symptoms and if they were in contact with a person exposed to the disease.  Researchers also found that the app could detect mpox at various stages:  flat lesions without fluid, raised lesions with clear fluid, deep rashes filled with white fluid, and prior to the lesion scabbing over, which usually takes place within two (2) or three (3) weeks.

The app provides five different levels of advice, from seeing a doctor immediately to taking no action.

Despite these capabilities, researchers noted that the app is not intended to serve as a substitute for physicians, as it may provide false negatives at times.  Researchers noted that the goal of developing the app is to benefit those with limited care access.  “Many people seek out medical information on the internet, and much of that may be inaccurate,” said Thieme in a press release. “With this app, developed with guidance from the Centers for Disease Control and Prevention and the World Health Organization, we hope to encourage people to seek out care.”

AI is increasingly being used to support disease detection.  A Mayo Clinic study from September 2022 involved testing an AI-based screening strategy intended to detect new cases of atrial fibrillation following a review of electrocardiograms.

With 1,003 patients who participated in continuous monitoring, labeled as the intervention group, and 1,003 who received usual care, the algorithm detected atrial fibrillation in six (6) of 370 low-risk patients and 48 of 633 high-risk patients.  These findings led researchers to the conclusion that the AI-guided screening strategy enhanced the detection of atrial fibrillation.

REFERENCE:  Health IT Analytics; 07 MAR 2023; Mark Melchionna