Related Experiment Video
Updated: Jun 16, 2025
05:30
Author Spotlight: Advancing the Analysis of Plasma Extracellular Vesicle Proteome for Cardiovascular Biomarker Studies
Published on: January 31, 2025
350
Using Data-Driven Algorithms with Large-Scale Plasma Proteomic Data to Discover Novel Biomarkers for Diagnosing
Simeng Ma1, Ruiling Li1, Qian Gong1
1Department of Psychiatry, Renmin Hospital of Wuhan University, Wuhan 430060, China.
Journal of Proteome Research
|August 16, 2024
Summary
Machine learning identified 45 plasma proteins as potential depression biomarkers. Adding proteomic data improved depression diagnostic models, showing promise for early detection and treatment guidance.
Area of Science:
- Biomarker discovery
- Proteomics
- Machine learning in healthcare
Background:
- Technological advances enable plasma proteome quantification in large patient cohorts.
- Plasma proteome screening aids in identifying biomarkers for depression diagnosis and treatment.
Purpose of the Study:
- To model and discover depression biomarkers using machine learning.
- To enhance depression diagnostic models with proteomic data.
Main Methods:
- Utilized CatBoost machine learning for model construction on UK Biobank data.
- Employed Shapley Additive Explanations (SHAP) for model interpretation.
- Validated model performance using 5-fold cross-validation and AUC analysis.
Main Results:
- Identified 45 depression-related proteins based on top CatBoost features.
- The best diagnostic model, incorporating proteomic data, achieved an average AUC of 0.764.
- KEGG pathway analysis highlighted cytokine-cytokine receptor interaction as a significant pathway.
Conclusions:
- Data-driven machine learning and large-scale datasets are feasible for exploring depression diagnostic biomarkers.
- Proteomic data integration improves the performance of depression diagnostic models.
- Further validation is required for the identified biomarkers and diagnostic models.
