A local research team and an artificial intelligence firm have jointly developed an AI model to predict the treatment response and survival rate in prostate cancer patients.

Yonsei University College of Medicine’s Urology Department Professors Koo Kyo-chul, Lee Kwang-suk, and Chung Byung-ha said Monday they developed an “artificial neural network (ANN) modeling” to predict customized survival outcomes and propose the best therapy for patients, in collaboration with the AI developer, Selvas AI.

The research team retrospectively collected clinical and pathological data from 7,267 prostate cancer patients between January 1988 and December 2017 to analyze treatment responses and survival rates.

“The accuracy of the predictions was as high as 91 percent, which was higher than 80 percent by the existing prediction model that uses statistical analysis,” the research team said.

In particular, the team collected data reflecting the pathophysiology of each stage of diagnosis and treatment to analyze the effects of disease states on cancer progression.

By doing so, the AI model could predict the disease progression according to the condition at the time of the prostate cancer diagnosis, and suggest the optimal treatment, the team explained.

“Compared to the existing predictive model that did not reflect time series, this model can analyze the entire course of the disease to find the optimal treatment at the time of diagnosis,” Koo said. “Based on this, we have built a customized treatment selection system on an online platform. Gangnam Severance Hospital’s Prostate Cancer Center is using the platform for patient counseling and treatment policy establishment.”

The study was supported by the National Research Foundation of Korea under the Ministry of Science and ICT and published in the latest issue of the World Journal of Urology.