RESUMO
Aim This study aimed to develop a predictive model to predict patients' mortality with coronavirus disease 2019 (COVID-19) from the basic medical data on the first day of admission. Methods The medical data including the demographic, clinical, and laboratory features on the first day of admission of clinically diagnosed COVID-19 patients were documented. The outcome of patients was also recorded as discharge or death. Feature selection models were then implemented and different machine learning models were developed on top of the selected features to predict discharge or death. The trained models were then tested on the test dataset. Results A total of 520 patients were included in the training dataset. The feature selection demonstrated 22 features as the most powerful predictive features. Among different machine learning models, the naive Bayes demonstrated the best performance with an area under the curve of 0.85. The ensemble model of the naive Bayes and neural network combination had slightly better performance with an area under the curve of 0.86. The models had relatively the same performance on the test dataset. Conclusion Developing a predictive machine learning model based on the basic medical features on the first day of admission in COVID-19 infection is feasible with acceptable performance.
RESUMO
Introduction Ventricular shunting remains the standard of care for patients with idiopathic normal pressure hydrocephalus (iNPH); however, not all patients benefit from the shunting. Prediction of response in advance can result in improved patient selection for ventricular shunting. This study aims to develop a machine learning predictive model for treatment response after shunt placement using the clinical and radiomics features. Methods In this retrospective pilot study, the medical records of iNPH patients who underwent ventricular shunting were evaluated. In each patient, the "idiopathic normal pressure hydrocephalus grading scale" (iNPHGS) and a "Modified Rankin Scale" were calculated before and after surgery. The subsequent treatment response was calculated as the difference between the iNPHGS scores before and after surgery. iNPHGS score reduction of two or more than two were considered as treatment response. The presurgical MRI scans were evaluated by radiologists, the ventricular systems were segmented on the T2-weighted images, and the radiomics features were extracted from the segmented ventricular system. Using Orange data mining open-source platform, different machine learning models were then developed based on the presurgical clinical features and the selected radiomics features to predict treatment response after shunt placement. Results After the implementation of the inclusion criteria, 78 patients were included in this study. One hundred twenty radiomics features were extracted, and the 12 best predictive radiomics features were selected. Using only clinical data (iNPHGS and Modified Rankin Scale), the random forest model achieved the best performance in treatment prediction with an area under the curve (AUC) of 0.71. Adding the Radiomics analysis to the clinical data improved the prediction performance, with the support vector machine (SVM) achieving the highest rank in treatment prediction with an AUC of 0.8. Adding age and sex to the analysis did not improve the prediction. Conclusion Using machine learning models for treatment response prediction in patients with iNPH is feasible with acceptable accuracy. Adding the Radiomics analysis to the clinical features can further improve the predictive performance. SVM is likely the best model for this task.