Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 7, 2026Diabetes

Combined protein-based models substantially outperformed clinical models with consistent improvements in reclassification metrics (median delta C-index = 0.117; range: 0.060-0.213).

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Do proteomic and metabolomic predictive models improve risk prediction for major clinical outcomes in patients with type 2 diabetes compared to the SCORE2-Diabetes clinical model?

Population

Individuals with Type 2 Diabetes (T2D) from the UK Biobank

Comparison

LASSO-based predictive models using proteomic… vs SCORE2-Diabetes clinical model

Design

Cohort

Authors

HZHongqiang Zhang

Discussion

Loading...

Member takes

Overview

A simplified panel of 200 proteins significantly improves the prediction of multisystem complications in type 2 diabetes compared to standard clinical models like SCORE2-Diabetes.

Key Points

  • This study aims to construct a multi-omics atlas of type 2 diabetes outcomes and identify predictive molecular panels.
  • Integrated UK Biobank proteomic and metabolomic data.
  • Conducted cross-sectional and longitudinal analyses to identify associated proteins and metabolites.
  • Developed LASSO-based predictive models compared against SCORE2-Diabetes.
  • Identified molecular signals with consistent associations across multiple T2D outcomes.
  • Protein-based models outperformed clinical models with improved net reclassification metrics (median delta C-index = 0.117).
  • A simplified panel of 200 proteins was proposed, showing robust predictive performance.

Structured PICO

Do proteomic and metabolomic predictive models improve risk prediction for major clinical outcomes in patients with type 2 diabetes compared to the SCORE2-Diabetes clinical model?

P
Population
Individuals with Type 2 Diabetes (T2D) from the UK Biobank
I
Intervention
LASSO-based predictive models using proteomic and metabolomic profiles
C
Comparator
SCORE2-Diabetes clinical model
O
Outcome
Prediction of major T2D-related clinical outcomes (evaluated by Harrell’s C-index, net reclassification improvement, and integrated discrimination improvement)surrogate

A simplified panel of 200 proteins significantly improves the prediction of multisystem complications in type 2 diabetes compared to standard clinical models like SCORE2-Diabetes.

Cite This Study

Hongqiang Zhang (2026) studied this question.

synapsesocial.com/papers/6a250c957def13d035e1ccb1https://doi.org/10.2337/db26-2283-p
View Full Paper
Ask AI
Bookmark
Share