Skip to content

EduNetArchive/Burkin_Personalized-medicine

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Personalized Medicine Project

This repo contains the copy of "Personalized Medicine" projects' scripts. This project is aimed at developing of risk prediction models for patients with cardiovascular diseases.

Keywords

Machine learning, cardiovascular deseases, biomarkers

Abstract

Cardiovascular diseases are the leading cause of death worldwide, and accurately identifying individuals at risk is crucial for prevention and appropriate treatment. However, there is a need to identify specific risk factors tailored to the Russian population, which has been underrepresented in international research. In this study, predictive models were developed using machine learning on a Russian population dataset to predict future complications in patients. Eight key clinical parameters were identified as leading to increased risk. The addition of biomarkers, specifically PCSK9, improved the predictive accuracy, resulting in a Random Forest model with an AUROC of 0.957 in the train subset and 0.859 in the test subset for predicting combined cardiovascular death, myocardial infarction, or stroke within 1-2 years.

Description

We are aiming:

  • to analyze the medical data of patients with CVDs from Surgut Cardio Hospital;
  • to identify biomarkers correlating to the risk factor increase;
  • to compare primary risk biomarkers to previously described risk factors;
  • to develop predictive machine learning models;
  • to create guidelines for cardio medical community.

Publications

  1. Russian Cardiology Bulletin 2023
  2. Computational Biology and Artificial Intelligence in Medicine 2022 DOI:10.14341/CBAI-2022-14

About

Prognostic ML-models and key-feature extraction for analysis of cardiovascular complications

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 96.9%
  • Python 3.1%