Maintaining proper health records improves machine learning predictions for novel 2019-nCoV.

摘要: BACKGROUND: An ongoing outbreak of a novel coronavirus (2019-nCoV) pneumonia continues to affect the whole world including major countries such as China, USA, Italy, France and United Kingdom. We present outcome ('recovered', 'isolated' or 'death') risk estimates 2019-nCoV over 'early' datasets. A consideration is likelihood death for patients with 2019-nCoV. METHOD: Accounting impact variations in reporting rate 2019-nCoV, we used machine learning techniques (AdaBoost, bagging, extra-trees, decision trees k-nearest neighbour classifiers) on two datasets obtained from Kaggle March 30, 2020. 'country', 'age' 'gender' features predict both included patient's 'disease' history (only second dataset) dataset. RESULTS: The use improves prediction 'death' by more than sevenfold. models ignoring patent's performed poorly test predictions. CONCLUSION: Our findings indicate potential using part feature set improve This development can have positive effect predictive patient treatment result easing currently overburdened healthcare systems worldwide, especially increasing prevalence third wave re-infections some countries.