Machine Learning for Prediction of Patients on Hemodialysis with an Undetected SARS-CoV-2 Infection

Abstract

Background: We developed a machine learning (ML) model that predicts the risk of a hemodialysis (HD) patient having an undetected SARS-CoV-2 infection that is identified after the following 3 or more days. Methods: As part of a healthcare operations effort we used patient data from a national network of dialysis clinics (February-September 2020) to develop a ML model (XGBoost) that uses 81 variables to predict the likelihood of an adult HD patient having an undetected SARS-CoV-2 infection that is identified in the subsequent ≥3 days. We used a 60:20:20% randomized split of COVID-19 positive samples for the training, validation, and testing datasets. Results: We used a select cohort of 40,490 HD patients to build the ML model (11,166 COVID-19 positive cases and 29,324 unaffected (control) patients). The prevalence of COVID-19 in the cohort (28% COVID-19 positive) was by design higher than the HD population. The prevalence of COVID-19 was set to 10% in the testing dataset to estimate the prevalence observed in the national HD population. The threshold for classifying observations as positive or negative was set at 0.80 to minimize false positives. Precision for the model was 0.52, the recall was 0.07, and the lift was 5.3 in the testing dataset. Area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC) for the model was 0.68 and 0.24 in the testing dataset, respectively. Top predictors of an HD patient having a SARS-CoV-2 infection were the change in interdialytic weight gain from the previous month, mean pre-HD body temperature in the prior week, and the change in post-HD heart rate from the previous month. Conclusions: The developed ML model appears suitable for predicting HD patients at risk of having COVID-19 at least three days before there would be a clinical suspicion of the disease.

Trajectories of clinical and laboratory characteristics associated with COVID‐19 in hemodialysis patients by survival

“…The model developed showed changes in IDWG had the highest variable feature importance (reflecting the impact of the value to the prediction). 11 Our present analysis adds to previous findings by showing quantitative trends in key parameters during the period before and after presentation with COVID-19. One of the novel findings, we identified was the inverse trends in pre-HD SBP before presentation between COVID-19 positive versus negative patients.…”

“…In general, absolute differences in trends, although significant, were small and we therefore suggest using multiple markers in combination for risk prediction, as shown in our previous paper regarding a machine learning prediction model developed for early detection of patients with COVID-19. 11 Moreover, it is important to realize that the trends do not show the mean of individual trajectories, but an aggregate of the cohort groups. We cannot rule out that there might be some minimal temporal bias secondary to the definition of the reference date/Day 0 for suspicion/testing that may have an impact on the trajectories.…”

Trajectories of Clinical and Laboratory Characteristics Associated with COVID-19 in Hemodialysis Patients by Survival

“…Although most patients were likely tested for clinical reasons, which is known for most of the COVID-19 positive patients, this is a limitation of the analysis. In general, absolute differences in trends, although significant, were small and we therefore suggest using multiple markers in combination for risk prediction, as shown in our previous paper regarding a machine learning prediction model developed for early detection of patients with COVID-19 11 . Moreover, it is important to realize that the trends do not show the mean of individual trajectories, but an aggregate of the cohort groups.…”

“…We recently showed how a machine learning predictive model that uses changes in clinical parameters before diagnosis had reasonable performance in classification of patients with a SARS-CoV-2 infection three days before symptoms onset (area under the curve is the testing dataset was 0.68). The model developed showed changes in IDWG had the highest variable feature importance (reflecting the impact of the value to the prediction) 11 . Our present analysis adds to previous findings by showing quantitative trends in key parameters during the period before and after presentation with COVID-19.…”

Predictors of Shorter- and Longer-Term Mortality after COVID-19 Presentation among Dialysis Patients: Parallel Use of Machine Learning Models in Latin and North American Countries

“…Prior efforts have leveraged machine learning modeling to assist with early detection of SARS-CoV-2 infection in HD patients, [25] and these models add another set of resources to be considered in the clinician's toolbox by providing a method to suitably assist with the prognosis of HD patients who contract COVID-19. Amidst the time of SARS-CoV-2 vaccines being more and more of an option in the world, the predictors of mortality will need to be established speci cally in vaccinated dialysis patients considering regional differences in the world in patient populations and vaccine types.…”