Page 143 - AI for Good-Innovate for Impact Final Report 2024
P. 143

AI for Good-Innovate for Impact



               Use case – 32: Yuanjing Large Model (Medical)                                                        32-ChinaUnicom












               Country: China

               Organization: China Unicom

               Contact person: Dong Li, lid530@ chinaunicom .cn

               Haoran Wu, wuhr15@ chinaunicom .cn

               32�1� Use case summary table


                Domain                   Health, medical
                The problem to be        Patient medical selection
                addressed                Pre-diagnosis
                                         Interpretation of medical reports
                                         Medical research assistance

                Key aspects of the solution  Data governance
                                         Intelligent Patient medical guidance
                                         Online intelligent consultation
                                         Medical research assistance
                Technology keywords      Large models
                                         Comprehension
                                         generation

                Data availability        Private data,
                Metadata (type of data)   Natural language text, test results, multi-modal reports, patient
                                         queries (chat and video)

                Model Training and       Pretrain with Medical data
                fine-tuning              SFT, P-tuning

                Testbeds or pilot deploy-  pilot deployments case in a hospital in Guangdong Province
                ments


               32�2� Use case description


               32�2�1  Description

               Yuanjing Large Model (Medical) is committed to comprehensively empowering various
               application scenarios in the medical and health field, targeting scenarios such as patient
               medical selection and pre diagnosis information collection before seeking medical treatment,




                                                                                                    127
   138   139   140   141   142   143   144   145   146   147   148