Machine Learning Models Predict Kidney Disease Risks in Post-Pandemic Period

Strategic Insights -

Research conducted at the Penn State College of Medicine has revealed a strong link between COVID-19 and acute kidney injury (AKI) and chronic kidney disease (CKD). The study aimed to use large electronic health records (EHR) and machine learning algorithms to predict the incidence of AKI and CKD during the post-pandemic period. The researchers developed a practical webpage application for clinical use, leveraging eight machine learning models, including extreme gradient boosting (XGBoost), neural network, and random forest (RF). The study found that incorporating COVID-19 infection history as a predictor significantly improved model performance, with XGBoost demonstrating the best performance for predicting AKI and CKD risks.

Key Takeaways:

  • The study used national EHR data from TriNetX, covering a prospective cohort of 104,565 patients from July 1, 2022, to March 31, 2024.
  • A total of 69 baseline variables were included, with demographics, comorbidities, lab test results, vital signs, medication histories, hospitalization visits, and COVID-19-related variables.
  • Eight machine learning models were applied, with cross-validation and model tuning conducted during the training process.
  • The final models, incorporating 9 variables, were selected, with XGBoost demonstrating the best performance for predicting AKI and CKD risks.
  • The study demonstrated the applicability of using large national EHR data in developing high-performance machine learning models to predict AKI and CKD risks in the post-COVID-19 period.
  • Incorporating the number of COVID-19 infections in the past year showed improved prediction performance and should be considered in future models for kidney disease prediction.

Statistics:

  • 104,565 patients were included in the prospective cohort study.
  • 69 baseline variables were used in the study.
  • 8 machine learning models were applied, including extreme gradient boosting (XGBoost), neural network, and random forest (RF).
  • The study found that XGBoost demonstrated the best performance for predicting AKI and CKD risks, with AUROC = 0.803 for AKI in 1 month, 0.799 for AKI in 1 year, and 0.894 for CKD in 1 year.
  • Random Forest (RF) was selected for predicting the incidence of CKD in 1 month, with AUROC = 0.896.

Sources:

  • eBioMedicine: Prediction of acute and chronic kidney diseases during the post-covid-19 pandemic with machine learning models: utilizing national electronic health records in the US. (2025;115:105726)
  • Penn State College of Medicine: Artificial Intelligence and Biomedical Informatics Pilot Funding.