Main information
- The Microsoft AI Diagnosis Orchestrateur has surpassed doctors in the precise diagnosis of medical cases.
- The AI system has reached an accuracy of 85.5 percent, while the average of the 21 doctors was 20 percent.
- AI has the potential to reduce unnecessary health care costs and improve diagnostic efficiency.
A recent Microsoft study has shown promising results for its new AI diagnostic system. AI, known as Microsoft AI Diagnostic Orchestrateur, has significantly surpassed a group of experienced doctors in the precise diagnosis of medical cases.
Experience, described in a Microsoft blog post, focused on 304 case studies from the New England Journal of Medicine. AI and human doctors were responsible for resolving these cases step by step, imitating real clinical scenarios. In particular, it was a question of prescribing tests, asking questions and refining potential diagnoses.
Promising results
The AI system has reached remarkable precision, correctly diagnosing 85.5 percent of cases when associated with the large O3 linguistic model of Openai. On the other hand, the 21 doctors practitioners who participated in the study, each with considerable experience ranging from five to twenty years, obtained on average a significantly lower rate of accuracy, of 20 percent. It should be noted that doctors have not been able to access the resources on which they usually rely to establish diagnoses, such as colleagues, medical literature or AI assistance.
Microsoft highlights the potential profitability of its AI system compared to human doctors. The company stresses that unnecessary health care costs in the United States represent a significant share of overall health expenses. His study suggests that AI could play a crucial role in reducing these expenses.
Complexity of the cases used
Mustafa Suleyman, CEO of Microsoft’s IA division, was enthusiastic about the results, declaring that they are an important step towards medical superintendent. He highlighted the complexity of the cases used in the study, which represent some of the most difficult diagnostic scenarios faced by doctors.
While welcoming these advances, Microsoft recognizes that AI must be considered a complement to human health professionals rather than a replacement. The company recognizes the wider clinical roles of doctors, which go beyond the simple diagnosis and include aspects such as the management of ambiguity, the establishment of a relationship of trust with patients and their families, and the supply of holistic care-of the areas in which AI is not yet up to it.
