How valuable is the application value of artificial intelligence large models in the medical and health field? Recently, SenseTime released an upgraded version of the big language model "Big Doctor", which contains 13 preset scenarios and supports functions such as prompt project customization, long-range memory access, and medical knowledge base query and summary, helping industry organizations create exclusive medical and health models according to their own needs and promote the digital and intelligent transformation of the industry. At present, SenseTime has cooperated with Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine, Xinhua Hospital affiliated to Shanghai Jiao Tong University School of Medicine, the First Affiliated Hospital of Zhengzhou University and other institutions to realize the application of large models.
"Big Doctor" is open to upstream and downstream institutions in the medical and health industry chain.
"Hello, I am a medication assistant at the National Telemedicine Center, is there anything that can help you?" After logging in to the large language model of medication consultation developed by SenseTime in cooperation with the National Telemedicine Center of the First Affiliated Hospital of Zhengzhou University, users can receive answers to various questions.
"What is warfarin?" "Warfarin is a drug that is mainly used to prevent and treat thrombotic diseases such as deep vein thrombosis, pulmonary embolism, myocardial infarction, etc." "I started taking medicine after heart stent surgery last year, and recently I have a little blood in my stool, and I want to make an appointment for colonoscopy, does it matter?" "In the case of warfarin, having a colonoscopy may have some effects. Warfarin is an anticoagulant that may increase the risk of bleeding. During colonoscopy, there may be some slight bleeding, but this usually does not cause serious problems. However, if you are taking warfarin and you develop blood in your stool, this may be a concern. ”
SenseTime cooperated with the National Telemedicine Center of the First Affiliated Hospital of Zhengzhou University to develop a large language model for medication consultation
Although the answers provided by this large model based on the "big doctor" are highly professional, they will also tell the user: "The following information is not a substitute for the doctor's opinion, and it is recommended to use it under the guidance of the doctor." According to reports, the "big doctor" function covers four areas: smart health, smart patient service, smart clinical and digital intelligence construction, including 13 subdivided medical and health scenarios such as intelligent self-diagnosis, medication consultation, post-diagnosis follow-up management, smart medical records, and image report structuring.
In the face of daily health consultation, it can provide health management services such as disease risk prediction, physical examination consultation, health Q&A, etc., and give health consultation suggestions after multiple rounds of dialogue. Based on this capability, this large model can help enterprises in the field of health to realize the digital and intelligent innovation of service models, allow personalized health management services into the home, and improve the convenience of user health management.
In the face of patients' medical needs, it can efficiently realize patient service functions such as intelligent guidance and pre-consultation. In the outpatient link, the "doctor" will record, sort out and identify the content of doctor-patient consultation dialogue in real time, and upload the extracted and summarized medical record information to the electronic medical record system, freeing doctors' hands and improving clinical efficiency. In clinical diagnosis and treatment scenarios, it can provide clinical assisted decision support. After the diagnosis, the large model can provide medication guidance for patients, and help patients establish follow-up files, formulate follow-up plans, and improve the quality of medical services and patient acquisition.
In addition, with the ability of information summary, the large model can generate phased summary and discharge summary according to the information of the patient's medical process, and generate structured reports according to the free text in medical imaging reports and clinical medical records, which is convenient for doctors to conduct medical statistics and improve scientific research efficiency.
The "Great Doctor" function covers 13 subdivided medical and health scenarios in four major fields.
In Ruijin Hospital, "big doctor" has become a powerful assistant of doctors by learning the experience of gastrointestinal experts and a large number of real medical records, combined with the medical record structuring and reasoning decision-making ability of large models, and solved the problems of subjective input bias, lack of real decision-making variables, and insufficient important decision-making ability in the past in specialized auxiliary decision-making. It provides accurate recommendations for systemic drug therapy (chemotherapy, targeted, immunotherapy) and topical therapy (surgery, radiotherapy, etc.) based on the medical records of patients with gastrointestinal tumors entered.
In Xinhua Hospital, SenseTime and the hospital plan to use large models to empower patient service scenarios such as follow-up, analyze the speech content spoken by patients, extract follow-up data such as basic information of patients, changes in their condition, and medication conditions, and give professional health suggestions to optimize patient follow-up experience.
"The scope of services of Dayi is not only hospitals, we hope to cover the entire medical and health industry chain." Zhang Shaoting, vice president and head of smart medical business of SenseTime, said that this large model has been opened to upstream and downstream institutions in the medical and health industry chain, and users can log in to the "Discuss" language large model website (https://chat.sensetime.com/) to apply for a trial account. In the future, this artificial intelligence company will explore cooperation with enterprises and institutions in the fields of nutrition and health management to further release the potential of large models in the medical and health industry.
Column editor-in-chief: Huang Haihua Title picture source: Shangguan title picture Photo editor: Yong Kai
Source: Author: Yu Taoran