Artificial intelligence has already shown that it can be useful for analyzing medical images and has even shown that it can successfully pass medical student exams.
Now it is the turn of a new AI-based tool to demonstrate its ability to read the reports prepared by doctors and accurately anticipate the risks of death, hospital readmission and other possible complications.
Created by a team at New York’s Langone School of Medicine, Grossman School of Medicine, the software is now being tested at a number of the university’s partner hospitals, with the goal of making it common practice in the future in the community. medical.
Treat medical notes as data
A study on its possible interest was published Wednesday in the scientific journal Nature. Its lead author, Eric Oermann, a neurosurgeon and computer engineer at the New York School of Medicine, explains that while non-AI-based predictive models have been around for quite some time, they are rarely used in practice because they require a lot of effort. amount of data. input and formatting job.
But there is “one thing that is common to medicine everywhere, is that doctors take notes on what they see, what they talk to patients,” he said in an interview with AFP.
The predictive model, called NYUTron, was formed from millions of medical observations from the records of 387,000 patients treated between January 2011 and May 2020 at NYU-affiliated hospitals.
These observations included written reports from doctors, notes on the evolution of the patient’s condition, X-rays and medical images, or even recommendations given to patients upon hospital discharge, all forming a corpus of 4.1 billion words.
One of the main challenges of the software was to be able to interpret the language used by doctors, which varies a lot between professionals, especially in the abbreviations used.
95% of deceased patients identified
They also tested the tool in real world conditions, notably by training it to analyze reports from a hospital in Manhattan and then comparing the results to those from a hospital in Brooklyn, with different patients. By observing what happened to the patients, the researchers were able to measure the number of times the software’s predictions turned out to be correct.
Disturbing conclusion, the NYUTron software identified 95% of patients who died in partner hospitals before being discharged, and 80% of those who were readmitted less than a month after discharge. Results that exceeded the predictions of most clinicians, as do non-AI-based computer models currently in use.
But, to everyone’s surprise, a very experienced doctor, highly respected in the medical community, gave predictions “even better than those of the software,” Eric Oermann specified.
The software also successfully predicted 79% of patients’ length of stay in hospital, 87% of cases where patients were denied reimbursement for care by their insurance, and 89% of cases in which the patient suffered from additional pathologies.
Artificial intelligence will never replace the doctor-patient relationship, says Dr. Oermann. But it can allow “providing more information (…) to doctors so that they can make informed decisions.”
Source: BFM TV
