(Fighting against New Coronary Pneumonia) Zhong Nanshan team announced that the accuracy of New Coronary Pneumonia Critical Prediction Model reached 88%
China News Agency, Guangzhou, May 14 (Cai Minjie) The Guangzhou Institute of Respiratory Health announced on the 14th that the team of Zhong Nanshan, a member of the Chinese Academy of Engineering, represented the high-level expert group of the National Health and Health Commission of China to integrate clinical risk factors and establish a new critical illness for coronary pneumonia The prediction model has an accuracy of 88%. The model was initially used in many hospitals in Hubei, such as Raytheon Mountain Hospital, and was recognized by local peers. The relevant papers were recently published in the international authoritative magazine "JAMA Internal Medicine". The corresponding authors are Zhong Nanshan and He Jianxing, president of Guangzhou Respiratory Health Research Institute.
Severe neopulmonary pneumonia can be fatal, and 6% to 8% of patients diagnosed with new coronary pneumonia will develop critical illness, requiring intubation ventilation or ICU monitoring, and the mortality of these critical patients can be as high as 65%. Identifying patients with the risk of becoming critically ill early in the disease period and strengthening monitoring in time can help improve the patient's prognosis and also have important significance for the allocation of medical resources.
The model is based on 1590 patients with new coronary pneumonia in China. 72 clinical factors were screened by LASSO regression and 10 key independent risk factors were found, including: abnormal chest X-rays, age, hemoptysis, shortness of breath, impaired consciousness, The number of underlying diseases, previous cancer history, neutrophil to lymphocyte ratio, lactate dehydrogenase and direct bilirubin.
Subsequently, a multi-factor prediction model is constructed according to the weight of each risk factor, and the accuracy of internal verification reaches 88%. In addition, external verification was performed on a total of 710 patients from multiple hospitals in Wuhan, Daye, and Foshan, and the accuracy was between 88% and 98%. The research team also built a free-access Chinese and English version of the web prediction tool (http://118.126.104.170/) to facilitate doctors and patients around the world to log in to make predictions.
According to reports, the prediction model is easy to use and can provide each patient with an individualized prediction probability. The risk of subsequent development of critical illness in the low-risk group is only 0.7%, while that in the high-risk group is as high as 59.3%. To improve the efficiency of medical resource allocation.
Next, the artificial intelligence joint laboratory established by Guangzhou Respiratory Health Research Institute and Tencent will expand the big data research of respiratory diseases such as new coronary pneumonia and develop more accurate and practical clinical tools. (Finish)