A Groundbreaking Medical Model The Specific Scene Prompt Enhanced Conversation Robot

TapTechNews August 15th news, according to the report of Beijing Daily on the 14th, the guidance diagnosis large model - Specific Scene Prompt Enhanced Conversation Robot developed by the team of the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences, has recently passed clinical trials. This model has both professionalism and empathy, and can not only be applied to guidance diagnosis work, but also is expected to be used in more complex scenarios such as patient health education and preoperative conversations in the future.

It is known that the construction of this large model is based on the real guidance diagnosis conversations collected comprehensively in different medical scenarios, and uses the knowledge base formed by extracting conversation information as training data, which can solve various specific problems that occur in different medical scenarios.

The research team established a Real Medical Conversation Corpus Alliance to collect original doctor-patient conversation records, and established a standardized process of voice collection > automatic text conversion > manual proofreading. As of June 2024, the team has collected and sorted out more than 1.2 million real doctor-patient conversations in 24 different medical scenarios.

In the prospective randomized controlled trial, the guidance diagnosis large model shows significant advantages over humans in factualness, safety, and empathy. In the real application scenario, this model can reduce 11.2% repeated communication and 5.4% doctor-patient conflict ratio.

The guidance diagnosis large model has recently obtained the highest level of evidence-based medical support evidence, and the research results have been published in the medical journal Nature Medicine. TapTechNews attaches the relevant achievement link as follows.

Reference