Researchers at Microsoft have developed an artificial intelligence-based tool to diagnose complex medical cases through a series of steps including gathering patient information, determining necessary tests, analyzing results, and proposing an appropriate diagnosis.
Accuracy Exceeding Doctors
Reuters reported on Friday that the tool achieved accuracy more than double that of general practitioners in diagnosing over 300 complex and unfamiliar medical cases, and outperformed them in tests based on emergency department admission data.
Simulating Expert Work
The tool relies on a model using a system to coordinate diagnostic steps, simulating the work of a team of medical experts. It identifies questions to ask the patient about their medical history, aspects to focus on during clinical examination, and then decides on required tests and analyses based on the collected information.
The model can continuously update its diagnostic hypotheses and plans, requesting additional tests when needed, rather than analyzing all symptoms and test results at once, as some current medical AI models do.
Evaluations and Expectations
Dominic King, Executive Vice President of Microsoft’s AI Health Unit, stated that evaluating medical systems using multiple-choice questions does not necessarily reflect the reality of clinical practice, as patients usually do not arrive at doctors with organized and complete information about their conditions.
The results also showed that using the system to organize diagnostic steps could reduce costs compared to methods relying on doctors' experience to determine diagnostic procedures.
The results reflect the tool's performance in the tests conducted on it and do not necessarily mean its superiority over doctors in all clinical circumstances, requiring further evaluation before widespread adoption in medical practice.
Source: Syrian Arab News Agency (SANA)
Comments (0)