JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Predicting chronic kidney disease progression with artificial intelligence

  • 0Pathology and clinical laboratory. INPAC research group. Clinica Colsanitas. Keralty group, Fundación Universitaria Sanitas, Bogotá, Colombia.
Bmc Nephrology +

|

April 26, 2024

|

April 26, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning models predict chronic kidney disease (CKD) progression and the need for renal replacement therapy (RRT). A time-to-event model shows promise for clinical use in managing CKD patients.

Area Of Science

  • Nephrology
  • Medical Informatics
  • Data Science

Background

  • Chronic kidney disease (CKD) progression prediction tools lack clinical utility.
  • Developing accurate predictive models for advanced CKD is crucial.

Purpose Of The Study

  • To develop and validate machine learning models for predicting CKD progression.
  • To predict the need for renal replacement therapy (RRT) in patients with stage 3-5 CKD.

Main Methods

  • Retrospective, observational cohort study.
  • Utilized demographic, clinical, and laboratory data from CKD patients.
  • Developed and compared three machine learning prediction models.

Main Results

  • Model 1 (4.5-year risk) achieved F1 scores of 0.82 (RRT), 0.53 (stage progression), and 0.55 (eGFR reduction).
  • Model 2 (time-to-event) showed C-indices of 0.89 (RRT), 0.67 (stage progression), and 0.67 (eGFR reduction).
  • Model 3 (reduced Model 2) demonstrated C-indices of 0.68 (RRT), 0.68 (stage progression), and 0.88 (eGFR reduction).

Conclusions

  • The time-to-event model effectively predicted CKD progression outcomes over five years.
  • This model can aid in forecasting the onset and timing of adverse events in CKD patients.

Keywords: