[ad_1]
A new model of artificial intelligence (AI) showed a precision comparable to that of experienced doctors.
Artificial intelligence has developed considerably in recent years. In the medical sector, the IA is the subject of intense discussions, but it is more focused on image recognition or badysis than on the diagnosis. For example, one of these algorithms was taught to badess a person's age and blood pressure simply by looking at a picture of his eye, while another was able to detect Alzheimer's disease from brain scanners even before doctors can do it. At present, a team of researchers has expanded the range of AI capabilities, developing an algorithm to diagnose common childhood diseases.
The diagnosis was considered a strictly human activity, especially in modern medicine, where the number of pathogens, diagnostic tests and biomarkers has increased dramatically in recent years. Subsequently, clinical decision-making has also become more complex and demanding – to be left in the hands of competent physicians.
However, in the current digital age, the electronic medical record has become a mbadive repository of data, data that can be used to mimic the way doctors think. To make a diagnosis, doctors often use a logical approach to make a diagnosis. They start from the main complaint and ask specific questions about this complaint and other relevant aspects. Then they check the background, background and any other pieces of useful information and provide a diagnosis. Of course, experienced doctors do it almost intuitively, without mentally breaking down all the steps, but in a sense, the whole process is very logical.
<! – Username: zmescience_300x250_InContent
->
Could this approach be imitated on a computer? These researchers think so.
Kang Zhang and his colleagues have developed an AI-based model that applies an automatic natural language processing (NLP) system, using in-depth learning techniques to identify clinically relevant information from health records. e. The model looks for medical records, symptom reports, lab results, as well as a library of best practice guidelines.
They trained and calibrated the model during 1.3 million patient visits to a major health center in Guangzhou, China. They had a total of 101.6 million data points.
After that, the AI was able to identify common childhood diseases with a precision comparable to that of a doctor. In addition, he was able to divide them into two categories: common (and less dangerous) conditions such as influenza and foot-and-mouth disease, and dangerous or life-threatening conditions, such as seizures and seizures. acute asthma and meningitis.
The researchers point out that the machine is not meant to replace the doctor's diagnosis, but to provide a tool for streamlining health practice. It could, for example, sort patients according to the potential severity of the disease and serve as a diagnostic aid in complicated cases.
"While this impact may be particularly evident in areas where there are few health care providers relative to the population, such as China, health care resources are in high demand around the world and the benefits of health care are high. Such a system will probably be universal. "
The study was published in Nature Medicine.
Publish views: 41
Appreciated this article? Join more than 40,000 subscribers to the ZME Science newsletter. Subscribe now!
Source link
