Microsoft has taken A genuine footsteps towards medical supervisors, Mr. Mustafa Sleyman, CEO of the company Artificial intelligence arm. The technology giant indicates its strong new AI tools Diagnosis Four times more accurate and significantly lower than a group of human doctors.
The experiment has checked whether this tool can accurately diagnose the patient who is sick or not, imitating the job usually done by a doctor.
The Microsoft group has used 304 cases of cases derived from the New England Medical Journal to give a test called SDBENCH (SDBENCH). A language model has divided each case into a step -by -step process that the doctor will perform to achieve a diagnosis.
After that, Microsoft researchers built a system called Diagnostic Coordinator (Mai-Dxo), query a number of top AI models, including Openai's GPT, Google Gemini, Anthropic's Claude, Meta's Llama and Xai's Grok, in the way of imitating some of his experts.
In their experiments, Mai-Dxo surpasses the human doctors, achieving 80 percent accuracy compared to 20 percent of doctors. It also reduces 20 percent of cost by choosing less expensive tests and procedures.
“This coordination mechanism, diverse agents working together in this period of chain.
The company has poached some Google AI researchers to help other efforts, but another sign of A strong battle for AI expertise In the technology industry. Sleyman was formerly a CEO at Google to work on AI.
Who has been widely used in some areas of the US health care industry, including helping X -rays explain scan. The latest multimodal AI models are able to operate as more general diagnostic tools, although the use of AI in health care increases its own problems, especially related to the bias from the training of data deviations to specific demographics.
Microsoft has not decided whether they tried to commercialize technology, but the same executive director, who talked about an unnamed condition, said the company could integrate it into Bing to help users diagnose the disease. The company can also develop tools to help health experts improve or even automate patients to care for patients. What you will see in the next few years is that we are increasingly doing more to prove these systems in the real world.
This project is the latest organ in a developing study that shows how AI models can diagnose the disease. Over the past few years, both Microsoft and Google have published articles showing that large language models can accurately diagnose the disease when accessing medical records.
Microsoft's new research is different from the previous job in that it copies more accurately how doctors diagnose the disease by analyzing symptoms, ordering and further analysis until the diagnosis is achieved. Microsoft describes how it combines some AI models of Frontier as a path to super online health, in a blog post about the project today.
The project also suggests that AI can help reduce health care costs, an important issue, especially in the US. “Our model performed very well, both were diagnosed and received that diagnosis very effectively,” said Dominic King, Vice President at Microsoft, who involved in the project.