by Microsoft has developed a AI system for artificial intelligence (AI), which claims that it is four times better than doctors in diagnosing complex diseases.
The Tech company’s AI diagnosis system was able to correctly identify diseases of up to 86 percent of the time compared to British and American doctors compared to only 20 percent.
Microsoft announced the results and claimed that it had laid the basis for “medical superintelligence”.
As a Wes Streeting, the health secretary, it is tries to bring AI into the widespread use in the NHS to improve efficiency. In April, the NHS list list rose for the first time in seven months and reached 7.42 m in one fell swoop to one of the most important commitments from the government to shorten waiting times.
Microsoft claimed that his system could solve problems cheaper than doctors – and beat doctors, even if he adhered to a budget for diagnostic tests.
The system, known as Microsoft AI Diagnostic Orchestrator (MAI-DXO), was tested on 304 cases of the New England Journal of Medicine, a medical magazine that is known for the publication of complex cases from the Massachusetts General Hospital.
The system comprised a virtual group of five different AI bots, each of which served different roles such as “DR Hypotheseer”, “DR-Test Chooser” and “Dr Challenger”, which deliberately asked further questions or order tests and provide a diagnosis.
In one case, the system diagnosed embryonic rhabdomyosarcoma, a rare form of cancer, which normally occurs in children, with a 29-year-old woman.
The system was able to diagnose 85.5 percent of the conditions if it was paired with the most advanced AI model, which was developed by Chatgpt Developer Openai and if it had no budget restriction for the order tests.
Even if it had to stick to a budget of $ 2,000 (1,458 GBP), it was more than 70 percent of the time correct.
The 21 human doctors, who had an average of 12 years of experience, spent an average of $ 2,963 for tests.
The average doctor had 19.9 percent of the time correct, although they could not use textbooks or software to follow information and were more of generalists than specialists.
Microsoft’s AI tool diagnosed the condition more than half of the time, even if no tests can be ordered, said Microsoft.
The researchers said that the journal, which had made the cases available, was behind a paywall and that many were published after the AI system was trained. This ensured that the cases could not be included in the data records used to create the AI, and the system had to get to the diagnoses itself.
Microsoft’s AI Health Division is headed by Mustafa Sulyman, the British entrepreneur, who co -founded Deepmind before the laboratory was acquired. Who moved to Microsoft last year.
