The study in Sweden of an AI mobile app is the first to be tested for melanoma in a primary care setting.
According to a recent article in Science Daily, one standout feature of the AI-mobile app is that it appears to be capable of diagnosing melanoma with high precision. The study was led from Sweden’s Linköping University. Studies have previously been conducted on skin lesion images, but the new app offers the highest degree of precision.
Dr. Magnus Falk, the senior associate at Linköping University’s Health Department, led the study. There is still no approved AI-based method to assess skin lesions in Swedish healthcare.
Experienced physicians find it difficult to distinguish melanoma from other skin issues, but since it is a serious form of skin cancer, it must be identified early. One of the goals of the Swedish study was to compare the performance of the new AI tool against the performance of primary care physicians who must make decisions with limited resources.
Dr. Panos Papachristou is affiliated with the Karolinska Institute and lead study author. The doctor commented that the results are very encouraging as the app accurately identified each melanoma. Papachristou emphasized that it is of equal importance that the app recognize harmless lesions, and it fulfilled this mission.
About the Process
Primary care physicians followed their normal procedures for the study to diagnose suspected skin lesions. If melanoma was suspected the physician would refer the patient to a specialist or cut the lesion, then diagnose the tissue.
Once the physician made a decision that the tissue needed further diagnosis, they would use the AI-based mobile app.
A picture is taken using enlargement lens, then analyzed. Using the image as a guide, the physician is then better informed as to whether the lesion is in fact melanoma. At this point the researchers were able to compare the app’s performance to the normal diagnostic methods.
250 skin lesions were examined. 11 melanomas and 10 precancerous lesions were found. The mobile app found every melanoma and only missed one precancerous lesion. One of the requisites of the study would preclude physicians to allow their decision to be persuaded by the response of the app.
In his opinion of the study results, Dr. Falk said that there is still work to be done due to the uncertainty surrounding the AI tool. He said that more studies should be conducted for further evaluation of its usefulness.
A large follow-up study is being planned for several countries comparing the use of the app to not having it involved in the study.
Editor’s Note: Get Involved
Cancer doesn’t discriminate. Patient Worthy and its partners are interested in amplifying the voices of those from all identities and backgrounds. If you have a cancer journey to share, reach out here to learn more about how your voice can help spread awareness and inspire individuals from all walks of life.
AI artificial intelligence cancer diagnosis melanoma melanoma diagnosis research
Last modified: May 1, 2024
