Wuhan team finds 3 new crown death markers, establishes prediction model accuracy of more than 90%

quickly and accurately assess the condition of patients is extremely important for the new crown pneumoniaLocal time on May 14, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology and other institutions in the academic journal "Nature - machine intelligence," issued a document that, using three biomarkers, machine learning algorithms can predict patients at least 10 days in advance of individuals with pneumonia new crown the mortality rate of over 90% accurateThese three biomarkers are lactate dehydrogenase, lymphocytes and high sensitive C-reactive protein levelscombination of these markers, paper presents a simple and easy-to-decision rules, you can quickly anticipate new crown pneumonia in patients with the highest risk so they can be prioritized, and may reduce mortalityabove-mentioned paper entitled "AninterpretablemortalitypredictionmodelforCOVID-19patients", the first author is deputy director of the emergency department, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology Li Yan, MDTongji Medical College, Tongji Hospital emergency department director Li Shusheng and Critical Care Medicine, Tongji Medical College, Tongji Hospital, Department of Anesthesiology and Xu Hui Huazhong University of Science professor of artificial intelligence and automation College of Yuan Ye co-corresponding author for the paperYan Li, who collected the blood samples in Wuhan, China 485 new crown patients with pneumonia and find reliable and meaningful risk of death markersAcquisition time of these samples is 2020 January 10 to February 18 2020, gathering place, Tongji Hospital, Tongji Medical College insample is divided into two parts, wherein a model 375 cases, 110 cases of further model for testing375 cases of patients for modeling, there are 201 cases after discharged, and the remaining 174 cases death110 cases for testing in 97 patients survived death in 13 casesOF machine learning algorithm is designed according to a mathematical modeling method for identifying the most predictive of mortality biomarkersClassification task of becoming a problem in the algorithm, the input data includes basic information, symptoms, blood samples, as well as common, severe, critically ill patients with laboratory test results, such as liver function, kidney function, blood coagulation, electrolytes and inflammation factorUltimately, the model lactate dehydrogenase, lymphocytes, high sensitive C-reactive protein preferably recognizes the new crown ill patients with pneumonia three key biomarkersThe author believes that this result is consistent with current medical knowledgeanalysts said the authors in the paper, the higher the level of lactate dehydrogenase itself and the tissue damage that appears in different diseases, including pneumonia and other lung diseasesFor new crown patients, severe pneumonia lactate dehydrogenase level rise shows that increasing the degree of lung injuryHigh-sensitivity C-reactive protein levels are important in acute respiratory distress syndrome, a sign of poor prognosis, reflecting the sustained inflammatory stateIn addition, the paper said, "Our results also show that lymphocytes or can be used as a potential therapeutic targetThis hypothesis is supported by clinical findings." Paper wrote, lymphopenia is a common feature of the new crown patients with pneumonia, may It is a key factor associated with disease severity and lethalityauthors conclude that their model can be accurately and easily through a simple intuitive interpretation of clinical examination, and quickly quantify the risk of death in patients with a new crownThe authors say the model involved three biomarkers can be easily collected at any hospital"In the crowded shortage of hospitals and medical resources, this simple model can help doctors to quickly prioritize patients, particularly during limited medical resources of the epidemic." (Bio Valley Bioon.com)< br />