Viral pneumonia (VP) is an acute respiratory infection caused by viral infection such as influenza viruses (H1N1, H5N1), rhinoviruses, coronaviruses (nCoV) et al. and characterized by dyspnoea and hypoxaemiaeven. Most patients with VP present with the prominent systemic symptoms (high fever, headache, myalgia, fatigue, chills, et al.) and the respiratory symptoms (cough, sore throat and so on), which has become a persistent and pervasive burden of disease in China and worldwide. It can be seen that timely and effective treatment is essential.

In the treatment of VP, antiviral and immunomodulatory therapies are the key therapeutic approaches. Recently, Traditional Chinese Medicine (TCM) brings light to VP patients due to remarkable clinical efficacy, and has been widely used in the treatment of VP patients in China. Cold or Hot Syndrome, one of the Eight Principle Syndromes of TCM, is an important components of TCM syndrome for identifying the nature of disease. However, the timely discriminative diagnosis of TCM Cold or Hot Syndrome in VP is often lacking. Consequently, It is vital to quickly, accurately and specificity identify the Cold or Hot Syndrome of TCM for diagnosis and treatment VP, so this study compared eight machine learning methods, namely the Random Forest algorithm (RF), Gradient Boosting Machine (GBM), Logistic regression algorithm (LR), multi-Grained cascade forest (GCForest), eXtreme Gradient Boosting (XGboost), Category boosting (CatBoost), Light Gradient Boosting Machine (lightgbm), stacking of XGboost, RF and GBM to combine the clinical features and hematological of viral pneumonia (coronavirus disease 2019, COVID-19) to construct a diagnosis model to differentiate the Cold or Hot Syndrome of TCM of VP.

And to facilitate clinical use and generalization, we selected ten medical indicators, namely Age, Peak body temperature of the day, Mean corpuscular volume (MCV), C-reactive protein (CRP), neutrophil granulocyte (NEU%), Red blood Cell distribution width (RDW-SD), Unconjugated bilirubin (IBLL), Serum Albumin (ALB), Creatinine (CREA), and UREG to build a discrimination diagnosis model of the Cold or Hot syndrome of TCM for VP, and named it: AI-assisted discrimination of the Cold or Hot Syndrome of Traditional Chinese Medicine for viral pneumonia based on clinical features.