Since the advent of medical artificial intelligence, it has been really integrated into clinical application. There is a big gap. Where is the crux of “big thunder and small rain”? Recently, the “Ai + medical” seminar jointly sponsored by ozmca and medical research and education group invited professionals from the medical, scientific research and investment circles to discuss the hot spots, analyze the reasons and give countermeasures for the three problems concerned by the industry.
Q1: can medical artificial intelligence really make mistakes?
The digitization of artificial intelligence, which took the lead in testing the water in the UK, encountered embarrassment: those interested are young, healthy and wealthy people, but the elderly who really need services do not “catch a cold”. This leads to the deviation between supply and demand: offline medical services are embedded in artificial intelligence, and the elderly are not used to it; The budget allocation is directly diverted online, and the resources do not match.
Use the common auto driving navigation system as an example: drivers trust the system too much and watch videos while driving automatically, resulting in car accidents. He said, “automation is convenient for mankind to see, but the mistakes caused by too much trust deserve attention.” since artificial intelligence can also make mistakes, how to minimize the probability of mistakes?
The key concerns are put forward by clinicians, and artificial intelligence scientists design systems and solve problems, but their emphases are different. Scientists pay attention to specific methods, and doctors pay more attention to whether they have high clinical value. The gap between the two needs to be run in and filled. It is guided by doctors and assisted by AI scientists to clarify the “role”, which is conducive to achieving the maximum accuracy of medical AI.
Q2: how should AI build trust with human beings?
Over the past decade, the investment in medical artificial intelligence has gradually cooled down from the boom. Previous data showed that more than 90% of medical artificial intelligence investment failed. Looking at the market, there are few AI products that have really been approved for listing. Can medical artificial intelligence build trust with human beings like intelligent voice system? Technicians say bluntly: the basis of trust lies in artificial intelligence doing better, cheaper and more effective than human beings.
Artificial intelligence algorithms need to be more systematically verified. Zheng Zhe, vice president of Fuwai Hospital, said that scientists are familiar with traditional treatment technologies from animal experiments, human experiments to mechanism research. As for the mechanism of artificial intelligence algorithm, human beings are not clear. He said: “we input one and output one. What happened? Is it feasible and available? These need to be sorted out so that more people can understand, understand and trust artificial intelligence.”
Up to now, the algorithms of artificial intelligence are imposed by human beings. If artificial intelligence changes experience into autonomous learning, it can achieve leapfrog development. Artificial intelligence will gradually master the ability of autonomous learning and improve, and human beings will gradually build trust.
Q3: can artificial intelligence endow medical ethics?
Artificial intelligence has a strong social media attribute. How to endow medical ethics? Participating experts believe that the concept of “concept remodeling” needs to be emphasized in artificial intelligence medicine.
For example, during the period of COVID-19, many people knew that vaccination could prevent disease, but “vaccine hesitation” still existed. Artificial intelligence can design communication software to understand people’s reluctance to vaccinate, improve disease home management, popularize treatment standardization, reshape the concept of vaccine and realize more universal vaccination. Ozmca believes that AI intervention can greatly increase service accessibility, benefit more people and create great medical benefits, which is also a part of medical ethics.