2026-27 Project (Tarroni & Varela)

LR AI ECGs: Improving cardiac care in low-resource settings using Artificial Intelligence tools for electrocardiogram analysis

SUPERVISORY TEAM

Supervisor

Dr Giacomo Tarroni at City St George’s
School of Science & Technology, Department of Computer Science
Email: giacomo.tarroni@citystgeorges.ac.uk

Co-Supervisor

Dr Marta Varela at City St George’s
School of Health & Medical Sciences, Department of Medicine
Email: mvarela@sgul.ac.uk

PROJECT SUMMARY

Project Summary

Can artificial intelligence (AI) transform heart care in resource-constrained settings? This project will harness foundational AI models for electrocardiogram (ECG) analysis (pre-trained on millions of ECGs) to diagnose cardiovascular disease in LMICs with unprecedented accuracy. Partnering with cardiologists in LSHTM and City St George’s, as well as collaborators in Kenya, Ghana, and Thailand, the student will leverage local clinical databases of ECGs, to adapt and miniaturise these foundational models. At the press of a computer or mobile phone button, they will be able to accurately diagnose the heart conditions most relevant to the region from ECG traces. With cutting-edge AI, real-world data, and direct links to clinical teams, this project offers the chance to deliver genuine global health impact while shaping the future of cardiovascular diagnostics.

Project Key Words

Artificial Intelligence, Electrocardiogram, AI, ECG, LMICs

MRC LID Themes

  1. Health Data Science

Skills

MRC Core Skills

  1. Quantitative skills
  2. Interdisciplinary skills

Skills we expect a student to develop/acquire whilst pursuing this project:

We expect the students to acquire strong programming skills and the ability to design, train, adapt and minimise AI models. They will also improve their communication and problem-solving skills, which will be necessary to design suitable prototypes for low-resource settings.

Routes

Which route/s are available with this project?

  • 1+4 = No
  • +4 = Yes

Possible Master’s programme options identified by supervisory team for 1+4 applicants:

  • Not applicable

Full-time/Part-time Study

Is this project available for full-time study? Yes
Is this project available for part-time study? Yes

Location & Travel

Students funded through MRC LID are expected to work on site at their primary institution. At a minimum, all students must meet the institutional research degree regulations and expectations about onsite working and under this scheme they may be expected to work onsite (in-person) more frequently. Students may also be required to travel for conferences (up to 3 over the duration of the studentship), and for any required training for research degree study and training. Other travel expectations and opportunities highlighted by the supervisory team are noted below.

Day-to-day work (primary location) for the duration of this research degree project will be at: City St George’s – Tooting campus, London

Travel requirements for this project: We envisage two visits to partner sites in LMICs, one to present the project to local partners and jointly design working prototypes. This will be followed by a subsequent visit to test prototypes and provide training on it.

Eligibility/Requirements

Particular prior educational requirements for a student undertaking this project

  • Minimum standard institutional eligibility criteria for doctoral study at City St George’s
  • We hope to recruit a student with a strong quantitative background, in Computer Science, Maths, Engineering, Physics or similar. Prior experience with programming is highly desirable, although not strictly necessary.

Other useful information

  • Potential Industrial CASE (iCASE) conversion? = No
  • CSG change of supervisory role = The supervisory team may switch roles over the course of the studentship award.

PROJECT IN MORE DETAIL

Scientific description of this research project

Techniques and Database Availability:

Artificial intelligence (AI) has shown great promise in electrocardiogram (ECG) analysis, accurately identifying arrhythmias, myocardial infarction, cardiomyopathies, and other cardiovascular diseases (CVD). Beyond replicating expert diagnoses, AI ECG can reveal prognostic signals not detectable by human readers, such as 5-year risk of type-II diabetes or new-onset atrial fibrillation. To date, AI ECG research has been concentrated in high-income countries and on common CVDs, largely because traditional approaches require well-curated datasets of >10,000 ECGs paired with expert labels. Recent advances in foundational AI ECG (fAI ECG) models-pretrained on millions of ECGs in a manner analogous to large language models-now allow meaningful fine-tuning with as few as ~500 expert-annotated ECGs. This opens the possibility of addressing LMIC-specific diseases and rare cardiac conditions. The project’s main aim is to pioneer low-resource deployment of fAI ECG models to deliver clinical impact in LMICs. It will be carried out in collaboration with clinical cardiologists Dr Modou Jobe (LSHTM), Dr Anoop Shah (LSHTM) and Prof Elijah Behr (SGUL), who have extensive networks and already have access to large databases of clinical 12-lead ECGs from Kenya, Ghana and Thailand.

Project Objectives:

  1. Apply pre-trained open-source fAI ECG models to local LMIC databases, extending to conditions of regional importance such as rheumatic heart disease, Brugada syndrome, and HIV-related CVD. Benchmark performance against existing clinical workflows.   
  2. Develop and evaluate a lightweight prototype deployable on low-specification computers or mobile devices, leveraging existing connections with experts in AIMS Cameroon (https://aims-cameroon.org/).    
  3. Extend to single-lead ECG models to facilitate wide adoption and explore integration of tabular data (age, sex, comorbidities) to enhance model performance.

At all steps, we will benchmark the proposed solution against existing clinical workflows or other relevant techniques and provide suitable training to local professionals. We will also release all code and models in an open-source framework to further boost research in these topics. Steps 2-3 will be refined with local collaborators and the PhD student to maximise clinical relevance and ensure ownership. These findings will inform broader AI-based interventions in LMICs beyond ECG.

Risks and Mitigation:

  • Data heterogeneity: The collected ECGs will vary in format; mitigation through existing collaborations with Imperial and MIT, who can provide robust harmonisation pipelines.   
  • Stakeholder adoption: Mitigated by early involvement of local clinicians, engineers, and patients via PPI events.   
  • Training bias: Pretrained models may reflect high-income populations; mitigated by retraining model layers on LMIC datasets. Generalisability across ethnic groups will be monitored, with initial deployment targeted to the originating clinics.

The supervisory team has extensive expertise in AI for clinical translation, which will be complemented by the clinical parterns and existing networks in LMIC cardiology. This will ensure early access to data, technical training, and mentorship for the student.

Further reading

Relevant preprints and/or open access articles:
(DOI = Digital Object Identifier)

Other pre-application materials: https://github.com/PKUDigitalHealth/ECGFounder

Additional information from the supervisory team

The supervisory team has provided a recording for prospective applicants who are interested in their project. This recording should be watched before any discussions begin with the supervisory team.

Tarroni & Varela Recording

MRC LID LINKS

To apply for a studentship: MRC LID How to Apply
Full list of available projects: MRC LID Projects
For more information about the DTP: MRC LID About Us

Comments are closed.