ChatGPT shows “impressive” accuracy in making medical decisions

Can artificial intelligence replace doctors? Although it is still too early to know, the fact is that the growing popularity of these systems currently reflects the presence Great potential in the field of medicine and healthcare.

Now, a recent study conducted by a group of researchers from General Brigham’s Mass Efficiency has been verified ChatGPT, artificial intelligence developed by OpenAIWhen making clinical decisions.

Specifically, this AI model shows a 72% accuracy in clinical decisions ranging from diagnosis to medical treatment suggestions.

The AI-based chatbot has shown the same success in both Primary and emergency care, as well as in all medical specialties. The results of the American research team at Mass General Brigham, a health system based in Boston, were published in the journal Journal of Medical Internet Research.

“We valued the help ChatGPT provided us from the moment we started with the patient and throughout the process, through diagnosis and treatment. “His performance is at the same level as someone who just graduated from medical school as an intern or resident,” he admitted. Mark Suchiphysician to General Brigham’s Mass.

Although AI technology is advancing rapidly, the extent to which these models can help in healthcare has not yet been studied.

In this sense, this study reveals that ChatGPT is able to “collaborate” with healthcare professionals to achieve this Treating patients, recommending a specific test, determining how to proceed with medical care, and ultimately making the final diagnosis.

To verify the reliability of this tool, researchers analyzed their responses to a series of possible diagnoses based on the patient’s initial information (including age, gender, symptoms, and level of urgency of the condition). In addition, they also tested their ability to make different decisions, thus simulating the entire process of caring for a real patient.

The results were a “surprise” for the team: Chat GPT showed an average accuracy of 72%. In addition, When the level of accuracy in the final diagnosis was evaluated, the success rate reached 77%. However, when he had to make a decision about the type of clinical treatment, such as deciding which medication to give a patient, he got only a 68% correct rate, according to the researchers.

“Chat GPT had difficulties with differential diagnosis, which is a fundamental issue in medicine because it is When the doctor has to decide what to do“, according to Suchi, adding: “This is important because it adds more value and tells us where doctors are truly experts: In the early stages of patient care, at times when information is scarce and a list of possible diagnoses is needed“.

Support tool

So, these scientists believe that before tools like ChatGPT are integrated into clinical care, More research and legislative guidance are needed. That’s why this team of experts is now studying whether AI tools can improve patient care and outcomes in hospitals with limited resources.

Lively arrived Adam Landmand“We are currently working on solutions that help with health documentation and patient responses, with a focus on accuracy, reliability and equity,” added one of the research’s co-authors.

The emergence of AI tools in the health field has been groundbreaking and has the potential to positively transform ongoing care. AI systems can analyze Large amounts of medical data, such as images from X-rays, MRIs, CT scans, and laboratory data This is to help doctors diagnose diseases.

In addition, they can analyze clinical and epidemiological data Predicting disease progression, risk of complications, and treatment outcomes. As if that weren’t enough, these systems also have the ability to optimize treatments based on genetics or medical history, manage electronic records, as well as provide assistance with surgery, monitoring or customer service.

However, although AI has great potential in this area, it also raises some possibilities Ethical and privacy challenges, which, according to experts, should be treated with caution. Furthermore, clinical decision making based on artificial intelligence is considered a A support tool for doctors It does not (at this time) replace your experience and clinical judgement.

Overall, collaboration between health professionals and AI systems in the future could lead to… More effective and personalized medical care.

References: