Providing intelligent guidance, pathological analysis and other services for patients and doctors, AI big models enter big hospital models | clinical | doctors
The rise of artificial intelligence big models has given "artificial intelligence+healthcare" more powerful functions. Recently, Ruijin Hospital affiliated with Shanghai Jiao Tong University School of Medicine showcased the latest achievements in the construction of "Future Hospital", including AI health consultation empowered by large models, "Cloud Ruijin" smart services, and a smart operation platform for full process control. Among them, the multi hospital intelligent imaging cloud platform has been empowered by large models, and on the basis of interconnectivity of medical images in multiple hospitals, it has achieved AI assisted diagnosis and treatment of radiation and pathology covering multiple parts and diseases, improving clinical diagnosis and treatment efficiency.
The Research Report on China's Artificial Intelligence Large Model Map shows that as of now, 79 large models with a parameter scale of over 1 billion have been released in China, and related application industries are rapidly expanding from office, life, entertainment and other fields to medical, industrial, education and other fields. At the recently held "Love Unlimited · Daily News" Artificial Intelligence Forum in Shanghai, Dimitris Metaxas, a distinguished professor at the Department of Computer Science at Rutgers University in the United States, stated that medical big models have demonstrated remarkable capabilities in solving multimodal data analysis and clinical long tail problems in healthcare. For example, in some clinical scenarios with limited data, big models can bring significant improvements in artificial intelligence effects.
"I feel unwell in my stomach." "Do you have any other symptoms?" "Vomiting, nausea, diarrhea." "Based on your symptoms, filling in the correct birthday and gender can provide more accurate department recommendations." "On April 3, 1993, male." "We suggest that you go to the Department of Gastroenterology for treatment. Please click on" Outpatient Appointment "."
This is a conversation between a "smart companion assistant" and a patient. For the medical treatment process, Shangtang Technology has created an intelligent medical platform based on the medical language model "Da Yi", which has multiple functions such as intelligent guidance, intelligent pre consultation, pre examination, intelligent follow-up, and hospital navigation. It has achieved digital and intelligent upgrades in multiple stages of pre diagnosis, during diagnosis, and after diagnosis, which can shorten the patient's stay in the hospital and optimize the patient's medical experience.
Zhang Shaoting, Vice President of Shangtang Technology and Head of Smart Medical Business, introduced that SenseCore, a Shangtang large device, has 5000 Petaflops computing power and supports simultaneous training of 20 super large models with billions of parameters. With the help of super computing power and medical infrastructure model groups, the company can efficiently train models for clinical long tail problems based on the specific needs of medical institutions, and even assist medical institutions in autonomous model training. "This model breaks through the bottleneck of limited data samples and high annotation difficulty in the medical long tail problem. It can carry out small data, weak annotation, and efficient training for different tasks, while significantly reducing the deployment cost of large models and meeting the personalized and diversified clinical diagnosis and treatment needs of different medical institutions."
With this "medical big model factory" capability, Shangtang Technology has collaborated with industry partners to launch various basic models such as medical language big model, medical imaging big model, and biological information big model, covering different medical data modalities such as CT, MRI, ultrasound, endoscopy, pathology, medical text, and biological information.
For example, Ruijin Hospital has deployed AI assisted diagnosis and treatment applications that cover multiple parts and organs throughout the body. The applications on these multi hospital intelligent imaging cloud platforms cover more than ten clinical directions, including lungs, coronary arteries, head and neck blood vessels, liver, orthopedics, pathology, etc. With the empowerment of multimodal medical basic models, they can provide intelligent services that meet the entire process of diagnosis, treatment, and recovery for different clinical departments, effectively improving the diagnosis and treatment efficiency and accuracy of doctors.
In addition to radiological imaging, the pathology department of Ruijin Hospital has also achieved digital and intelligent transformation. Based on high-resolution full-size digital pathological slice images, the intelligent pathology assisted diagnostic system can automatically perform AI analysis on digital pathological slices, quickly locate suspicious lesion areas, and detect abnormal cells. At present, relevant algorithms have covered a series of scenarios such as gastrointestinal biopsy pathological examination, TCT cervical cytology screening, and breast immunohistochemistry quantitative analysis, alleviating the problem of pathological doctors' workload overload and significantly improving the efficiency of film reading.
"We will continue to deepen the construction of the 'Future Hospital', cooperate with artificial intelligence leading enterprises such as Shangtang Technology, apply cutting-edge technologies such as big models and metaverse to more clinical directions, further extend service boundaries, and provide patients with continuously improving medical services." said Zhu Lifeng, full-time deputy director of the Shanghai Digital Medical Innovation Center at Ruijin Hospital.
