An artificial intelligence (AI) would now be able to predict the reappearance of melanoma. It is the deadliest form of skin cancer.
Generally, patients who die of melanoma have been initially treated for early-stage melanoma. Afterwards, the cancer recidivism and is rarely detected before it spreads or reaches metastasis. To solve this problem, a group led by scientists at Massachusetts General Hospital (MGH) turned to AI. Thanks to the’artificial intelligence (IA), the team developed a technique to predict which patients might experience a recurrence.
A toxic treatment
Usually, patients undergo surgery to remove cancerous cells when the melanoma is still at an early stage. In more advanced casesthe use of immune checkpoint inhibitors is the most frequent. This treatment stimulates the anti-tumor immune reaction. However, it leads to significant side effects on the whole body.
Besides, early detection of these side effects is paramount in order to effect a suitable etiological assessment. Indeed, doctors can only use temporary immunosuppressive treatment after the latter. Nevertheless, making a multidisciplinary assessment is generally necessary for treat toxicity optimally.
“There is an urgent need to develop predictive tools to aid in the selection of high-risk patients for whom the benefits of immune checkpoint inhibitors would justify the high rate of morbid and potentially fatal immunological adverse events observed with this therapeutic class. »
Yevgeniy R. Semenov, lead author, researcher, Department of Dermatology at the MGH
Concretely, Yevgeniy Semenov and his colleagues wish reliably predict a recurrence of melanoma. Such a prediction would make it possible to more accurately select immunotherapy. In addition, it would slow down the evolution towards metastasis and increase the chances of survival from melanoma. Without forgetting that theexposure to treatment toxicity would be less substantial.
SEE ALSO: Top 10 Things Artificial Intelligence Does Better Than Humans
Artificial intelligence (AI) to the rescue
To this end, Yevgeniy Semenov and his team turned to machine learning algorithms. To power them, the scientists used information from the electronic medical records (EMRs) of various people with melanoma.
Specifically, they collected data on 1,720 cases of early-stage melanoma. About two-thirds of the information comes from the Mass General Brigham (MGB) health system. The rest comes from the Dana-Farber Cancer Institute (DFCI). Thanks to these EMRs, the group managed to extract 36 clinical and pathological characteristics of melanoma.
Ultimately, the researchers determined that the cancer cell division rate and tumor thickness are the characteristics the most predictive.
“Our results suggest that machine learning algorithms can extract predictive cues from clinicopathologic features for the prediction of early-stage melanoma recurrence, which will help identify patients who may benefit from adjuvant immunotherapy. . »
Yevgeny R. Semenov