This artificial intelligence system was able to detect 62% of actual cases of mental and memory decline in the first practical tests. This was reported by the Public Relations Department of the Massachusetts State Hospital Network.

“We have developed not just an artificial intelligence system, but an entire clinical suite of neural networks that includes five specialized agents, each of which critically studies the results of the work of other components of the system and helps them reach accurate and complete conclusions, like what happens in a consultation meeting for doctors,” said Hossein Esteri, assistant professor at Massachusetts Hospital.

The doctors said: “Dementia is usually detected in patients in the middle or late stages of development, when the pathological processes in brain tissue are so complex that it is difficult to suppress them using various treatments or lifestyle changes. This is partly because its diagnosis requires various arduous and lengthy cognitive tests, which are very difficult to conduct on the level of the entire country’s population.”

Doctors hypothesized that the development of dementia may significantly affect other clinical indicators that doctors record when performing routine examinations of patients or in other contexts, and kept in individual patient records. To search and analyze them, the researchers developed a multi-component artificial intelligence system, built on the basis of two different versions of the large linguistic model LLaMA and the medical neural network Med42.

To train him, the researchers prepared a collection of several thousand doctors’ notes and recordings of clinical procedures performed while monitoring the health status of 200 people with dementia and healthy people. This information was also studied and analyzed by several leading experts in the field of dementia diagnosis, and their advice and diagnoses were also used to train and test the work of the artificial intelligence system.

Subsequent tests of this system on another set of patient data showed that it was able to detect about 62% of cases of dementia development in its early stages, with the artificial intelligence system making a correct positive diagnosis in 98% of cases. The AI ​​system was also able to detect existing dementia in 58% of cases in which leading experts had incorrectly made a negative diagnosis, greatly expanding the practical application of such intelligent systems.

Source: TASS