This is a real question. Will Artificial Intelligence (AI) replace doctors and solve diagnostic challenges more efficiently? Microsoft might have an answer for it.
Recently, Microsoft's AI team conducted research that showed how its AI tool was able to investigate and solve complex cases that have put physicians in a dilemma.
Microsoft AI Diagnostic Orchestrator (MAI-DxO) correctly diagnosed up to 85 per cent of the case proceedings in the New England Journal of Medicine (NEJM), a rate more than four times higher than a group of experienced physicians. The team also mentioned that the AI tool also gets to the correct diagnosis more cost-effectively than physicians.
Underscoring the importance of generative AI, Microsoft launched the AI tool with the help of clinicians, designers and AI scientists.
How was the AI tool evaluated?
United States Medical Licensing Examination (USMLE) questions were among the earliest benchmarks used to evaluate AI systems in medicine.
Apart from this, researchers presented the tool with 304 interactive case challenges drawn from the case series published in NEJM. These cases were among the most diagnostically complex and intellectually demanding in clinical medicine, said the team. Both the AI tool and physicians had to solve the cases in a stepwise manner.
The team observed that for the same tasks, the experts achieved a mean accuracy of 20 per cent across completed cases. The medical practitioners who participated in the study worked without access to colleagues, textbooks, or even generative AI.
The results of the research are transformational for the healthcare industry and could empower patients to self-manage certain healthcare aspects and help clinicians with advanced decision support for complex cases.
"Our findings also suggest that AI reduces unnecessary healthcare costs. U.S. health spending is nearing 20% of US GDP, with up to 25% of that estimated to be wasted – per having little influence on patient outcomes," said the Microsoft AI team.
The company also mentioned that while AI can be helpful in solving complex diagnostic challenges, further testing is needed to assess its performance on more common, everyday presentations.
