Truace
Subscribe

Blood test and AI model help predict heart disease in people on peritoneal dialysis

Study title: Predicting cardiovascular disease in peritoneal dialysis patients using glycated proteomics and machine learning: A prospective multicentre study

Researchers built a prediction tool for cardiovascular disease in people receiving peritoneal dialysis, a group in which heart disease is the leading cause of death. The tool combines two glycated peptide biomarkers found by mass spectrometry with six routine clinical measures, using elastic-net logistic regression.

Read the study abstract

Background: Cardiovascular disease (CVD) is the leading cause of death among peritoneal dialysis (PD) patients, yet PD-specific biomarkers and prediction models remain limited. This study aimed to identify glycated peptide biomarkers associated with CVD in PD patients and develop a prediction model integrating glycated proteomics and clinical indicators using machine learning. Methods: In Phase I, a matched discovery cohort (48 PD patients with CVD and 48 without CVD) was used to identify glycated peptides by li…

By the Truvace desk1 min read1 sourceEvidence: Moderate · 68/100 — Controlled or validated comparison · n = 438 · 1 independent sourcePeer-reviewedTRV-2026-2301· cite & verify
Predicting cardiovascular disease in peritoneal dialysis patients using glycated proteomics and machine learning: A prospective multicentre study
Improvised peritoneal dialysis in an 18-month-old child with severe acute malnutrition and acute kidney injury a case report P01 by Francis Fredrick, Gudila Valentine. CC BY 2.5 · https://creativecommons.org/licenses/by/2.5

In brief

The model could allow earlier identification of high-risk dialysis patients and more targeted prevention. Its accuracy is moderate, with AUCs around 0.78 to 0.82, and the study does not show that acting on the predictions improves patient outcomes.

Main points

  1. The discovery phase compared 48 dialysis patients with cardiovascular disease against 48 without, and found 257 differentially abundant glycated peptides.
  2. The final model kept two peptides plus phosphate, diabetes, total Kt/V, albumin, cardiac troponin T and glycated albumin.
  3. AUC was 0.784 in training, 0.781 in internal testing and 0.822 in the external validation cohort.
  4. High-risk patients, defined by a training-derived cutoff of 0.373, had higher 1-year cardiovascular disease incidence in both validation cohorts.

The gain

A model combining two glycated peptides with six clinical indicators separated high- and low-risk dialysis patients and scored an AUC of 0.822 in an external cohort, outperforming three published risk models.

The rundown

The work ran in two phases. Phase I used a matched discovery cohort and liquid chromatography-tandem mass spectrometry to identify glycated peptides, with glmnet used to prioritise four candidate biomarkers.

In Phase II, candidate biomarkers were quantified in a prospective internal derivation cohort of 240 patients and a multicentre external validation cohort of 102 patients. Features were selected with an F-test before elastic-net logistic regression was fitted.

The authors compared the combined model with three published cardiovascular risk models and reported higher ROC-AUCs for their model. They conclude it may facilitate earlier risk stratification and targeted prevention.

Sources

  1. Peer-reviewedClinical and Translational Medicine2026-11-01

The debate