Dimitrios Doudesis

Emergency departments see a large number of patients with chest pain every day, but only a small proportion are actually having a heart attack. Current diagnostic pathways rely on high-sensitivity cardiac troponin thresholds and fixed-time algorithms, which do not account for patient-specific risk. As a result, many low-risk patients are admitted unnecessarily, while others are held for repeat testing over several hours. This leads to overcrowding, delays, unnecessary investigations, and missed opportunities for early discharge.

 

The problem is both clinical and operational. Clinicians need better tools to help them make safe, early decisions, and health systems need more efficient ways to manage chest pain pathways. The diagnostic process has not kept pace with available data or advances in machine learning.

The core idea is to create a University spin-out company that will license CoDE‑ACS, a machine learning algorithm developed to support clinical decision-making in patients with suspected acute coronary syndrome. CoDE‑ACS calculates a personalised risk probability using high-sensitivity cardiac troponin values, clinical features, and patient history. Unlike existing pathways, which rely on binary thresholds, CoDE‑ACS provides continuous risk scores to help clinicians rule in or rule out myocardial infarction more accurately and earlier.The algorithm has been externally validated in multinational cohorts and published in Nature Medicine, where we showed it outperforms current guideline-recommended strategies in terms of safety and efficiency.  

 

The startup will provide CoDE‑ACS as a stand-alone platform, accessible through a secure cloud-based API. This allows for flexible integration into hospital systems or existing diagnostic platforms, without requiring complete IT system overhauls. The tool can be licensed directly to hospitals or to diagnostic companies seeking to embed AI into their platforms.