Artificial intelligence is becoming a bigger part of everyday life. It can write text, answer questions, and even help people create pictures and computer programs.
Now, scientists are exploring another important possibility: Could artificial intelligence help doctors take care of patients?
New research published in the journal Nature suggests that advanced AI systems may be able to assist doctors with many parts of patient care.
The studies describe two new AI systems called MIRA and AMIE. Both systems can perform a wide range of medical tasks, from collecting information from patients and making diagnoses to suggesting treatments and planning follow-up care.
The findings are important because healthcare systems around the world are facing growing challenges. Many countries are dealing with physician shortages, aging populations, and rising numbers of people with chronic illnesses.
Doctors often have heavy workloads and limited time with each patient. Researchers hope that AI tools may eventually help reduce some of these pressures.
Most artificial intelligence systems developed for healthcare have focused on narrow tasks. For example, one system might look for signs of cancer in medical scans, while another might summarize patient notes. Real medical care, however, is much more complicated.
When doctors treat patients, they must gather information about symptoms, ask about medical history, order tests, interpret results, decide on treatment options, prescribe medications, and continue monitoring patients over time.
They also need to adjust treatments when conditions change. Researchers wanted to know whether AI systems could perform this type of complex reasoning.
The first study focused on MIRA, which stands for Medical Intelligence for Reasoning and Action. The system was developed by researchers led by Jakob Kather. MIRA was designed to work within an isolated electronic health record system and was tested using real-world data from more than 500 emergency department cases.
The researchers created virtual patients whose responses matched information recorded in actual clinical notes. MIRA communicated with these virtual patients by asking questions and gathering information, much like a doctor speaking with a patient.
The system then had access to more than 85,000 possible actions. It could order diagnostic tests, review test results, recommend treatments, prescribe medications, arrange medical procedures, and decide whether patients should be admitted to the hospital.
The results were impressive. MIRA achieved an average diagnostic accuracy of nearly 88 percent. In comparison, a panel of six physicians from different medical specialties achieved an average accuracy of about 78 percent.
The second study examined a system called AMIE, which stands for Articulate Medical Intelligence Explorer. AMIE was developed by researchers at Google and built using the Gemini family of artificial intelligence models.
Unlike many earlier systems, AMIE was designed to reason across multiple medical visits. This means it could track how a patient’s disease changed over time and evaluate whether treatments were working.
Researchers compared AMIE with 21 primary care physicians using 100 virtual patient scenarios covering five medical specialties. The cases were designed according to established medical guidelines used in the United Kingdom.
The results showed that AMIE performed as well as doctors in overall medical reasoning. In some areas, it performed even better. The system was more precise in recommending tests and treatments and showed strong alignment with established clinical guidelines.
Researchers also tested AMIE using a new benchmark called RxQA, which focuses on medication reasoning. Again, the system outperformed physicians on particularly difficult cases.
Although these findings are exciting, the researchers caution that neither system is ready to replace doctors. Medicine involves much more than making decisions based on information. Doctors must communicate with patients, understand personal preferences, recognize unusual situations, and make judgments in complex real-world environments.
The studies also have limitations. Most testing occurred in controlled virtual settings rather than in busy hospitals and clinics. Future studies will need to determine whether these systems perform equally well with real patients and unexpected situations.
An analysis of the findings suggests that conversational AI tools may become valuable assistants rather than replacements for physicians. These systems could help gather information, suggest possible diagnoses, ensure treatments follow clinical guidelines, and reduce routine workloads.
If future research confirms their safety and effectiveness, AI could become an important partner in healthcare, helping doctors provide better and more efficient care while addressing growing shortages of medical professionals.
If you care about medicine, please read studies that vitamin D could help lower the risk of autoimmune diseases, and drug for inflammation may stop spread of cancer.
For more health information, please see recent studies about which drug can harm your liver most, and results showing this drug can give your immune system a double boost against cancer.
Source: Nature.
