Anticipated AI: 80% of Medical Institutions
Not Yet Adopted by Unsure Cost-effectiveness

• Healthcare

Survey on Medical Information System Implementation

Nikkei Research conducted a survey in March 2023, targeting registered members of Nikkei Medical Online, a healthcare professionals' community operated by Nikkei BP, on the implementation status of medical information systems at medical institutions.
In the previous article, the survey results showed that most physicians referred to diagnostic imaging for the use of AI (artificial intelligence) as their anticipated technology.
So, is AI already spreading in the medical field? We asked physicians about the actual and forthcoming plans for their implementation of AI.

A total of 2,031 respondents at various medical institutions who are involved in purchasing were asked about their status of AI medical devices installment. Respondents were shown options of six ("genome treatment," "support on diagnostic imaging," "support on diagnosis and treatment," "drug development," "nursing care and dementia," and "support on surgery"), key priority areas that a roundtable group promoting AI applications at Japan's Ministry of Health, Labour and Welfare have proposed.
In conclusion, AI medical devices are not adopted widely. 79.4% of the total said they have not implemented any, and even "support on diagnostic imaging," which had the highest adoption rate, stood at 10%. "Genome treatment" was 7.5% and "support on diagnosis and treatment" was 7.1%.

"Diagnostic Imaging" and "Genome" popular among universities and public hospitals

Looking at the results by facility, both university hospitals and public hospitals, that are handling cutting-edge advanced healthcare, had the highest adoption rates. Particularly, university hospitals showed a significantly high rate, with 24.4% implementing genome treatment and 24.1% using AI diagnostic imaging. On the other hand, 94.3% of small clinics have not installed any of the above.

Why do they not adopt? The most common reason was "unsure about cost-effectiveness" (52.8%), by far from the second, "low cost-effectiveness" (24.8%). "Not sure of my use case" (22.7%) was also high. This indicates that they cannot imagine how the implementation will (positively) impact both the improvement of treatment quality and hospital management. Further to the context, the tech industry may be lacking enough appeals of ROI when implementing AI in the healthcare industry.

75.7% of the total said they have no further plans to introduce AI. However, there was a small portion of medical institutions who are considering the adoption of "support on diagnostic imaging" (12.1%), "support on diagnosis and treatment" (8.5%), "support on surgery" (6.6%), and "genome treatment" (6.4%).

Understanding the specific needs in the medical frontline is the key to further implementation

Although they are "unsure about the cost-effectiveness," they still consider the adoption of AI because they are nervous about being left behind in this wave of technological innovations. AI, which can instantly extract necessary information from enormous amounts of data and analyze it, is a good fit with healthcare, where handling massive data while interacting with patients. Presumably, healthcare professionals somewhat sense that and feel that AI is promising.
Further beyond, AI will likely spread from areas where "cost-effectiveness" is clear and easy to convince. In the areas of "support on diagnostic imaging," where CT scans and MRIs are installed widely, investment toward fast and accurate diagnosis on a large scale is necessary. The growing genome treatment, which optimizes individual patient care, needs AI that can analyze big data quickly and precisely.

From the open-ended responses in the survey, some physicians in their 20s and 30s expressed expectations of utilizing ChatGPT. Where interactive AI begins to fill in our daily life, the demand of "want to use AI for work, immediately" is rising among busy medical professionals. Inarguably, we should take cautious steps for using AI in healthcare, it is now time that the tech industry should have more discussions with the healthcare industry to better propose systems that are more fit to the industry's needs.

Despite the great alliance expected between healthcare and AI, our survey revealed that the implementation is not yet advancing. We feel that both the tech industry and the healthcare industry should examine the reasons for this low adoption and the surrounding issues, now.