App using artificial intelligence is able to recognize suspected mpox rashes with 90% accuracy

Author: Dhruv Kapadia

Led by scientists at Stanford Medicine and other institutions, a team has developed a new app, known as PoxApp, that is able to detect with 90% accuracy skin lesions caused by mpox (previously known as monkeypox), in images. The artificial intelligence algorithms in the app were trained on a data set of around 130,000 images of various skin conditions to ensure high accuracy and a low risk of false negatives. Simply by submitting a few pictures along with answering a few questions, people can receive a risk score and recommendation in less than 5 minutes. Alexander Thieme, MD, the lead developer of the app, stated, “it’s a quick, easy and anonymous way to get a first assessment.” However, Thieme stresses that this is not a substitute for treatment. Instead, Thieme hopes that the risk score and recommendation will encourage people to visit a doctor.

To calculate the risk score, the algorithm considers symptoms, close contact with someone who may have been exposed, and whether you have a skin lesion. Even more remarkable, the app is able to detect mpox at the various 4 stages of the disease while also offering 5 different levels of advice, ranging from no action to seeing a doctor immediately. Additionally, the app is anonymous and grants users the opportunity to submit their results for research data. Thieme hopes that this will enable scientists and public health officials to one day predict future surges in mpox infection and employ an early warning system.