Microsoft researchers unveiled a new artificial intelligence (AI) system on Monday that can diagnose patients more accurately than human doctors. The name of Microsoft AI diagnostic arcter (MAI-DXO), including a number of AI models and a framework that allows patients to go to suggest symptoms and date-related tests. Based on the results, then it suggests a possible diagnosis. Redmund -based Tech Dev highlighted that in addition to the accuracy of the diagnosis, the system is also trained to be effective in costing tests.
Microsoft developed a benchmark to test MAI-DXO’s performance
In a post on X (previously known as Twitter), Microsoft AI CEO Mustafa Solomon posted about the MyDXO system. He called it a “big step towards medical superintendent”, adding that the AI system could solve some of the most difficult medical issues in the world with high accuracy and low costs, compared to traditional diagnostic measures.
The company said in a blog post, MyDXOs imitate a virtual panel of diverse diagnostic approaches that help in solving medical issues. The arkstrans include a multi -agent system where a speculation provides, picks up a test, provides two other checklists and stresses, and challenges the last speculation.
Mai-DXOWORK FLOW
Photo Credit: Microsoft
Once an assumption transmits this panel, the AI system can either ask a question, request a test, or provide a diagnosis if it feels that there is plenty of information. If it recommends a test, it analyzes the cost to ensure that the overall cost is reasonable. Interestingly, this system is a model agnostic, which means that it can perform with any third party’s AI model.
Microsoft claims that the system increases the diagnostic performance of every AI model that was tested. However, Open O3 performed well by resolving 85.5 % of the Benchmark cases of the New England Journal of Medicine (NEJM). The company said the same issues were also given to 21 practitioners from the United States and the UK, and all of them had a clinical experience between five and 20 years. The accuracy of human doctors was 20 %.
The company said that MAI-DXO can be created to operate in cost obstacles. Once the input budget is added, the system looks for the system cost of the system by making diagnostic decisions. This helps the AI system to order just necessary tests rather than all possible tests to reject all the causes of the symptoms.
To evaluate the AI system, Microsoft also developed a new benchmark called a Diagnostic Benchmark (SD Bench). Unlike ordinary medical benchmark tests that ask for multiple selection questions, this test assesses the AI system’s ability to ask the right questions and order the right tests. Then they compare the answers with the results published in the NEJM.
In particular, MAI-DXO has not yet been approved for medical use, and this means a preliminary research to promote AI’s capacity in diagnostic operations. Microsoft said its AI system can only be approved for strict safety tests, clinical verification, and clinical use after regulatory studies.
