In its latest research on “Amie”, “Speech Medical Intelligence Explorer”, Google is working to expand the capabilities of diagnostic artificial intelligence to include understanding visible medical content. Amie is a research artificial intelligence system based on a huge and improved linguistic model specifically for medical diagnostic thinking and conversations.
The company has updated the system to be able to process visual medical information or analyzing clinical images, which allows it to conduct conclusions more comprehensively, as the trainer works.
This step represents a new development in bringing artificial intelligence closer to the way doctors diagnose medical conditions through medical images. Imagine to speak with artificial intelligence about your health problems, and instead of being satisfied with analyzing the symptoms of your description of them, it can also explain and analyze your medical images.
Doctors rely heavily on what is visible, such as skin conditions, devices readings, and laboratory analysis reports, and the Google team properly indicated, even instant messaging applications provide multimedia information (such as images and documents) to enrich the conversation.
How did Google developed Amie’s capabilities in medical reasoning?
To enhance Amie’s medical thinking capabilities, Google merged the Gemini 2.0 Flash model and applied the so -called “status -based inference framework”.
This framework allows Amie to be dying dynamically with questions and answers during the conversation, as the real doctor works. The system can determine the gaps in the information, order additional inputs such as images or test results, and improve its diagnostic conclusion in actual time.
For example, if Amie discovers the existence of incomplete information, he can order a picture of a skin condition or electrical heart planning (ECG), analyze that visual data, and merge the results into clinical dialogue.
To train Amie without endangering patients, Google created a hypothetical simulation laboratory.
Restrictions: Despite the promising results, Google has made clear a number of restrictions that should be taken into account when interpreting the search results. First, the study was conducted in a controlled research environment that does not fully reflect the reality of health care in the real world.
Also, the conversation interface cannot capture all the details of the personal or video consultation, in addition to the fact that the woven patients, although they are trained representatives, cannot represent all aspects of the real clinical interaction.
Conclusion: Google’s work on the Amie “speaking medical intelligence explorer” shows the advanced possibilities of artificial intelligence in understanding the symptoms of patients not only through texts, but also through the analysis of complex visual data, just as the specialist doctor does. This multi -media approach enhances the accuracy of the diagnosis and brings artificial intelligence an additional step towards supporting clinical decisions in an interactive time.
In a research simulation, the amie system showed impressive results, beating human doctors in several diagnostic tasks, and even receiving higher assessments regarding sympathy and credibility during text consultations.
