| Autor: |
Fernandes FT; School of Public Health, University of São Paulo, São Paulo, SP, Brazil. fernando.fernandes@fundacentro.gov.br.; Fundacentro, São Paulo, SP, Brazil. fernando.fernandes@fundacentro.gov.br., de Oliveira TA; School of Public Health, University of São Paulo, São Paulo, SP, Brazil.; Statistics Department, Paraíba State University, Paraíba, PB, Brazil., Teixeira CE; School of Public Health, University of São Paulo, São Paulo, SP, Brazil.; Bioinformatics and Computational Biology Lab, Brazilian National Cancer Institute, Rio de Janeiro, RJ, Brazil., Batista AFM; School of Public Health, University of São Paulo, São Paulo, SP, Brazil., Dalla Costa G; BP-A Beneficência Portuguesa de São Paulo, São Paulo, SP, Brazil., Chiavegatto Filho ADP; School of Public Health, University of São Paulo, São Paulo, SP, Brazil. |
| Abstrakt: |
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. |
