MAI-DxO: Microsoft’s New AI diagnoses complex medical cases with 85% accuracy
MAI-DxO uses a ‘chain of debate’ process and a panel of AI agents trained on rare case studies.
Microsoft has introduced a new AI-powered diagnostic tool capable of tackling complex medical cases that often baffle expert clinicians. Called MAI-DxO (Microsoft AI Diagnostic Orchestrator), the system has been developed by Microsoft’s AI health unit, founded by DeepMind co-founder Mustafa Suleyman.
When tested on complex real-world cases published in the New England Journal of Medicine, the AI tool correctly diagnosed 85.5%. For comparison, experienced doctors managed to solve only 20% of the same cases without external help.
The tool uses five virtual AI agents, each simulating a medical expert with unique roles, such as choosing tests or proposing hypotheses. The approach, dubbed the ‘chain of debate’, allows for step-by-step reasoning in arriving at diagnoses.
Microsoft trained MAI-DxO using 304 case studies and large language models from leading AI companies, including OpenAI, Google, Meta, and xAI. The AI panel mimics a real-world diagnostic team with significantly faster and more accurate outcomes.
Despite the promising results, Microsoft acknowledges that more validation and regulatory clarity are needed before such tools can be used in clinical practice. The company is currently working with health organisations to test the system further.
The aim is not to replace doctors but to ease their workload by offering a reliable assistant for the most challenging cases. Microsoft says MAI-DxO could represent a significant step toward what it calls ‘medical superintelligence’.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
