AI to help choose best cancer treatments

The Yomiuri ShimbunA Japanese scientist conducting research in the United States and an information analysis company based in Tokyo have decided to jointly develop an artificial intelligence (AI) service capable of selecting suitable treatment methods for cancer patients and helping ease patients’ worries over treatment.

Prof. Yusuke Nakamura of the University of Chicago, world-famous for his research into cancer genes, and FRONTEO Inc. plan to develop the AI service, which will incorporate data taken from the experience and diagnostic criteria of Nakamura and other experts. It will then learn from huge volumes of medical research reports and propose the most suitable treatment methods based on each patient’s genetic information and other factors.

FRONTEO already has an AI program capable of learning from humans’ experiences and senses. Nakamura and the company aim to begin the service within two years.

The new AI service will, among other things, provide explanations to patients and serve as a reference to help doctors in medical institutions choose treatment methods. It will also allow patients to access free, reliable and up-to-date information about diseases and medicines from home. 

The AI service will include functions to detect patients’ interests and concerns through analyzing their written input, enabling it to accurately provide the information patients require.

The number of medical research papers registered in U.S. official databases alone exceeds 26 million. Academic studies just on cancer are published at the rate of 200,000 per year, and even doctors who specialize in cancer have difficulty staying abreast of the latest knowledge. Information accessible over the internet is of mixed quality, and patients often have difficulty separating the wheat from the chaff to find accurate information.

The University of Tokyo and other entities are also using IBM Corp.’s Watson computer to study how to select the most appropriate treatment methods from research papers. Watson’s main feature is its ability to learn efficiently by incorporating experts’ judgments and choosing only high-quality information. 

Nakamura said: “There are still many patients who are unable to receive proper treatment since they can’t obtain correct information. We would like to design a system in which patients can be provided treatment without feeling regret.”