The “Watson sensation” that hit Korea’s medical industry seems to be waning. The domestic hospitals competed to introduce the artificial intelligence program of IBM last year, but no more joined the boom this year.

Starting with Gachon University Gil Medical Center which introduced the AI system in December 2016, eight hospitals jumped on the Watson bandwagon last year --Pusan National University Hospital, Konyang University Hospital, Keimyung University Dongsan Medical Center, Daegu Catholic University Medical Center, Chosun University Hospital, Chonnam National University Hospital and VHS Medical Center.

Some attribute the lack of followers to skepticism about the AI system, which had grown since the U.S. MD Anderson terminated its contract with IBM Watson last year.

More point out, however, that it is the systemic limitations that cooled down the Watson fervor in Korea. In other words, Watson-recommended remedies often turn out “useless,” in an environment where physicians have to treat patients according to the reimbursement criteria of the Health Insurance Review & Assessment Service (HIRA).

And that explains why the experts point to “HIRA medicine” as the reason for the low agreement ratio of a mere 50 percent on average between Watson and local medical workers.

Watson suggests “recommended,” “for consideration” or “not recommended” treatment methods after analyzing documents and recent research papers according to the patient’s condition. The “recommended” treatments are marked green, “for consideration,” orange and “not recommended,” pink.

The proportion of treatment methods agreed between the medical staff and Watson’s “recommended” ones for 118 colorectal cancer patients was 55.9 percent, according to Gachon University Gil Medical Center in December. The hospital announced its medical performance results for the first year anniversary of the introduction of Watson.

The rate increased to.78.8 percent when including the “for consideration” methods. Konyang University Hospital also analyzed the Watson application results on 100 breast cancer patients in time for the first year anniversary of the AI system, and the rate of treatment methods agreed between the medical staff, and Watson’s “recommended” ones was 48 percent.

A medical doctor explains Watson's diagnostic results to his patient at Gachon University Gil Medical Center.

Watson-recommended methods unfit for reimbursement criteria

Another university hospital has also experienced similar cases since Watson came into its medical system.

Watson recommended Olaparib, a PARP inhibitor, for a patient with breast cancer caused by BRCA mutation. The medical staff, however, suggested the prescription of Gemcitabine, an anticancer drug, to meet reimbursement standards.

The medical staff and Watson also disagreed on the prescription for a patient with lung cancer due to EGFR exon No. 19 genetic mutation. Watson recommended Tagrisso (Osimertinib) but the medical workers, following the reimbursement criteria, went with Iressa (gefitinib). Opinions also split for a patient with hormone refractory prostate cancer; Watson recommended GnRH agoinist + Xtandi (enzalutamide) while the medical staff suggested Paclitaxel.

“Watson recommends the best and the cutting-edge treatment methods, but we cannot follow that, said a medical professor who is currently using Watson. “If we do not follow HIRA’s criteria, you will most likely get curtailed. We can only do what HIRA tells us to do,”

“Some say that it is due to the characteristics of Koreans that lead to such a low consensus rate (between Watson and the medical staff), but that is not the case except for the case of gastric cancer. The influence of HIRA’s reimbursement criteria is much greater.”

Another professor who is currently using Watson also pointed out that in case of gastric cancer, it is only natural that opinions split since there are more patients in Asia, including Korea and Japan, than in the U.S., and the treatment methods differ.

“For cases other than gastric cancer, the discordance between Watson and the medical staff is due to HIRA’s criteria,” the professor said.

One of the provincial university hospitals which have introduced Watson to attract patients who leave for the metropolitan area has even requested IBM to reflect HIRA’s reimbursement standards to increase the agreement ratio. IBM reportedly replied that it is possible to reflect the opinion but that the request to modify the best-recommended treatment based on medical grounds just to meet Korean standards is absurd.

“It is meaningless to focus on the consensus rate. The AI is no more than a navigation system,” said Professor Lee Un, a neurosurgeon who is leading Gil Medical Center’s AI Hospital Formation Team. “Watson is one of the numerous solutions suggested by the AI. It is meaningful to use the AI in a way to produce better results.”

soo331@docdocdoc.co.kr

<© Korea Biomedical Review, All rights reserved.>