A multipurpose machine learning approach to predict COVID-19 negative prognosis in São Paulo, Brazil.
A multipurpose machine learning approach to predict COVID-19 negative prognosis in São Paulo, Brazil.
复制标题
DOI：
10.1038/s41598-021-82885-y
复制
发表时间：
2021-02-08
影响因子：
4.6
通讯作者：
Chiavegatto Filho ADP
中科院分区：
文献类型：
--
作者：
Fernandes FT;de Oliveira TA;Teixeira CE;Batista AFM;Dalla Costa G;Chiavegatto Filho ADP
新型冠状病毒病(新冠肺炎)对临床决策和医疗资源的有效配置提出了挑战。准确的预后评估对于提高患者的存活率是必要的，尤其是在发展中国家。这项研究建议通过训练多用途算法来预测新冠肺炎患者发生危重状况的风险。我们于2020年3月至6月从巴西S的一家大型医院跟踪调查了1,040例新冠肺炎RT-PCR诊断为阳性的患者，其中288例(28%)患者预后严重，即入院、使用机械通气或死亡。我们使用常规收集的实验室、临床和人口统计数据来训练五种机器学习算法(人工神经网络、额外的树、随机森林、CatBoost和极端梯度增强)。我们随机抽样了70%的患者来训练算法，30%的患者被留下来进行性能评估，模拟新的未知数据。为了评估算法是否能够捕获一般的严重预后模式，每个模型都通过组合三个结果中的两个来预测另一个进行训练。所有算法都表现出很高的预测性能(平均AUROC为0.92，灵敏度为0.92，特异度为0.82)。多用途算法的三个最重要的变量是淋巴细胞与C反应蛋白的比率、C反应蛋白和Braden评分。这一结果突显了机器学习算法能够从常规收集的数据中预测非特定的负面新冠肺炎结果的可能性。
The new coronavirus disease (COVID-19) is a challenge for clinical decision-making and the effective allocation of healthcare resources. An accurate prognostic assessment is necessary to improve survival of patients, especially in developing countries. This study proposes to predict the risk of developing critical conditions in COVID-19 patients by training multipurpose algorithms. We followed a total of 1040 patients with a positive RT-PCR diagnosis for COVID-19 from a large hospital from São Paulo, Brazil, from March to June 2020, of which 288 (28%) presented a severe prognosis, i.e. Intensive Care Unit (ICU) admission, use of mechanical ventilation or death. We used routinely-collected laboratory, clinical and demographic data to train five machine learning algorithms (artificial neural networks, extra trees, random forests, catboost, and extreme gradient boosting). We used a random sample of 70% of patients to train the algorithms and 30% were left for performance assessment, simulating new unseen data. In order to assess if the algorithms could capture general severe prognostic patterns, each model was trained by combining two out of three outcomes to predict the other. All algorithms presented very high predictive performance (average AUROC of 0.92, sensitivity of 0.92, and specificity of 0.82). The three most important variables for the multipurpose algorithms were ratio of lymphocyte per C-reactive protein, C-reactive protein and Braden Scale. The results highlight the possibility that machine learning algorithms are able to predict unspecific negative COVID-19 outcomes from routinely-collected data.
登录
查看更多内容
DOI：
10.1186/s12941-020-00362-2
发表时间：
2020-05-15
影响因子：
5.7
作者：
Chen, Wei;Zheng, Kenneth I.;Qiao, Zengpei
通讯作者：
Qiao, Zengpei
影响因子：
9.6
作者：
Bajwa, Ednan K.;Khan, Uzma A.;Christiani, David C.
通讯作者：
Christiani, David C.
影响因子：
5.7
作者：
Bong, Choon-Looi;Brasher, Christopher;Enright, Angela
通讯作者：
Enright, Angela
基于机器学习的早期预警系统能够准确预测 COVID-19 的死亡风险
影响因子：
16.6
作者：
Gao Y;Cai GY;Fang W;Li HY;Wang SY;Chen L;Yu Y;Liu D;Xu S;Cui PF;Zeng SQ;Feng XX;Yu RD;Wang Y;Yuan Y;Jiao XF;Chi JH;Liu JH;Li RY;Zheng X;Song CY;Jin N;Gong WJ;Liu XY;Huang L;Tian X;Li L;Xing H;Ma D;Li CR;Ye F;Gao QL
通讯作者：
Gao QL
影响因子：
7.2
作者：
Collins, Gary S.;Reitsma, Johannes B.;Moons, Karel G. M.
通讯作者：
Moons, Karel G. M.