Comparison of deep learning with regression analysis in creating predictive models for SARS-CoV-2 outcomes

Abstract

Background: Accurately predicting patient outcomes in SARS-CoV-2 could aid patient management and allocation of healthcare resources. There are a variety of methods which can be used to develop prognostic models, ranging from logistic regression and survival analysis to more complex machine learning algorithms and deep learning. Despite several models having been created for SARS-CoV-2, most of these have been found to be highly susceptible to bias. We aimed to develop and compare two separate predictive models for death during admission with SARS-CoV-2. Method: Between March 1 - April 24, 2020, 398 patients were identified with laboratory confirmed SARS-CoV-2 in a London teaching hospital. Data from electronic health records were extracted and used to create two predictive models using: 1) a Cox regression model and 2) an artificial neural network (ANN). Model performance profiles were assessed by validation, discrimination, and calibration. Results: Both the Cox regression and ANN models achieved high accuracy (83.8%, 95% confidence interval (CI): 73.8 - 91.1 and 90.0%, 95% CI: 81.2 - 95.6, respectively). The area under the receiver operator curve (AUROC) for the ANN (92.6%, 95% CI: 91.1 - 94.1) was significantly greater than that of the Cox regression model (86.9%, 95% CI: 85.7 - 88.2), p=0.0136. Both models achieved acceptable calibration with Brier scores of 0.13 and 0.11 for the Cox model and ANN, respectively. Conclusion: We demonstrate an ANN which is non-inferior to a Cox regression model but with potential for further development such that it can learn as new data becomes available. Deep learning techniques are particularly suited to complex datasets with non-linear solutions, which make them appropriate for use in conditions with a paucity of prior knowledge. Accurate prognostic models for SARS-CoV-2 can provide benefits at the patient, departmental and organisational level.

Deep-Learning-Based Survival Prediction of Patients in Coronary Care Units

“…The C-index showed that the accuracy of the deep-learning model (0.822) was about 4% better than that of the CPH model (0.782). It was indicated that deep learning may be more suitable for handling large samples, multivariate and nonlinear survival analysis than the CPH model, which was consistent with the research results [27][28][29].…”

The use of predictive models to develop chromatography-based purification processes

“…In this context, a major difference between statistical analysis and ML is that the latter often sacrifices interpretability (or explainability) in favor of the model's predictive power (Song et al, 2021b), even though both ML and statistical analysis may perform equally well on some datasets. For example, area under the receiver operation characteristic (AUROC) curves have been compared for models predicting various diseases, revealing values of 0.736 (ML) vs. 0.748 (logistic regression) when predicting acute kidney disease (Song et al, 2021b), 0.837 (neural net models) vs. 0.836 (regression models) when predicting infraction mortality (Piros et al, 2019a), and 0.926 (ANN) vs. 0.869 (Cox regression) when predicting the outcome of COVID-19 (Abdulaal et al, 2020b).…”

Prognosis patients with COVID-19 using deep learning

“…Besides, most recent studies have been reported to understand and diagnose the patients with COVID-19 [ 37 – 42 ] such as temporal deep learning [ 43 ], data-driven based extreme gradient boosting (XGBoost) [ 44 ], deep learning with regression analysis [ 45 ], biomarkers-based [ 46 ], machine learning and clinical data based [ 35 , 47 , 48 ], statistical neural network (NN) and DL [ 49 – 51 ], boosted random forest [ 52 ], CNN-LSTM, CNN-RNN and CNN-ML based on X-ray images [ 53 – 57 ].…”