Cardio-oncology
Cardiovascular events associated with cancer therapies, critical-care outcomes, and treatment safety.
Research and open science
From clinical questions to reproducible evidence.
The programme connects cardio-oncology, critical-care cardiology, pharmacovigilance, real-world evidence, echocardiography, clinical AI, and machine learning.
Research programme
Six connected themes
Acute and critical-care cardiology
Heart failure, acute myocardial infarction, arrhythmias, and longitudinal clinical outcomes.
Real-world evidence and pharmacovigilance
Clinical-outcome and safety research using large data resources including MIMIC-IV, eICU, and FAERS.
Clinical AI and machine learning
Explainable models, temporal validation, calibration, decision-curve analysis, and clinical utility.
Echocardiography and cardiac phenotyping
Ventricular volumes, valvular disease, filling pressure, and measurement discordance in relation to outcomes.
Interventional cardiology
Coronary intervention, no-reflow, procedural outcomes, and the domestic medical-device ecosystem.
Current projects
Projects in development or review
Principal investigator
CardioOnco-SAFE
An AI-supported clinical decision-support prototype for cardiovascular adverse events associated with cancer therapies.
The TRY 12,000,000 project proposal received preliminary approval under the TUSEB C4 programme; final scientific review and the funding decision remain pending.
Applicant and inventor
CardioMetric
A portable transparent educational device supporting ECG evaluation and basic measurement steps.
Patent application 2026/000366 is under examination.
The CardioOnco-SAFE budget refers to the submitted proposal. Funding has not been awarded. No financial conflict of interest is declared.
Reproducible research
Selected public repositories
GitHub
ECG-MAE-pipeline
Self-supervised 1D-MAE pretraining and multi-label ECG classification using PTB-XL.
View repository GitHubbrugada-triage-100hz
An open and reproducible deep-learning triage pipeline for Brugada syndrome.
View repository GitHubptbxl-identity-leakage
Analysis of patient-identity signals and leakage risks in ECG AI benchmarking.
View repository GitHubFAERS-CardioOncology-2015-2025
Reproducible code and intermediate outputs for a FAERS cardio-oncology safety analysis.
View repository GitHubsakarya-model-reproduction
Reproduction package for an explainable cardiorenal risk model in an acute-heart-failure phenotype.
View repository GitHublvesv-lvef-aki-mimic
MIMIC-IV echocardiography analysis of LVESV, LVEF, and acute kidney injury.
View repository