A heart condition is not necessarily a death sentence—unless it goes unnoticed. While Electrocardiogram (ECG) machines are increasingly commonplace, the specialised knowledge required to interpret them isn’t. The interpretation gap between human and machine can often mean that even when the data is captured, life-saving diagnoses are missed. And with heart defects, this is no joke.
What is an ECG?
According to the British National Health Service (NHS), an ECG is a test that records the electrical activity of your heart, including the rate and rhythm. If there is an abnormality, a doctor can notice it and diagnose a potential issue with the heart.
You’ll usually have an ECG (electrocardiogram) if a doctor or healthcare professional thinks you’re having symptoms of:
- a heart attack
- coronary heart disease
- problems with how quickly or regularly your heart beats (arrhythmia)
Find more information here.
Medical care is a geographical lottery
This disparity in cardiac care is, sadly, starkly illustrated in Ethiopia. Its population of more than 120 million people is served by approximately 30 fully trained cardiologists. That means per cardiologist, there are four million people potentially in need of their care. In world leader Greece, on the other hand, there are 2,800. When you compare the two nations, a person in Greece is approximately 1,400 times more likely to have access to a cardiologist than someone in Ethiopia.
On the one hand, there’s a clear functional problem. Even if a more general practitioner in Ethiopia can physically perform an ECG, without proper cardiologist training, they likely lack the knowledge to identify subtle abnormalities among the complex wave forms. Without being able to read the signs, diagnosis and treatment are delayed at best, or, at worst, don’t take place at all.
Particularly in cardiology, a delay in treatment can have dire consequences. Even short delays can increase the risk of irreversible heart damage, permanent heart failure or death. On the other hand, the problem seems to be systemic. If being born in Greece means you’re thousands of times more likely to receive adequate medical care than in Ethiopia, then there’s a fundamental failure of medical access that disproportionately affects vast populations in, as usual, the Global South.
Sustain AI supports physicians with AI
Non-profit organisation MI4People, alongside an international team of academic and clinical partners, has developed SUSTAIN-AI ECG. This pilot project is leveraging artificial intelligence to act as a second pair of eyes for overworked general practitioners in Ethiopia. And, it works: it’s already capable of detecting 10 distinct cardiac conditions from standard 12-lead ECGs within seconds.
The developers have ensured that the platform has a low barrier to entry. It’s completely web-based, runs in a standard browser, and requires only a basic digital ECG device and an internet connection. Its aim is not to act as a stand-in for a real doctor, but to simply provide diagnostic suggestions that a local practitioner can utilise in their practice. Most importantly, it operates on a non-profit model and is provided free of charge to the clinics most in need.
The project is an example of what can happen when respectful cross-continental partnership goes well. The AI itself was developed by the research group of Prof. Dr. Michael Guckert and Prof. Dr. Jennifer Hannig at the Technische Hochschule Mittelhessen. Its clinical implementation is led by the Armauer Hansen Research Institute, a biomedical research Institute in Ethiopia. Vitally, this ensures that the technology is adapted to the specific needs of Ethiopian clinical environments. To date, 30 healthcare professionals have been trained, and the system has supported nearly 6,000 clinical cases.
Improving diagnostic accuracy
As we write about frequently at RESET, AI systems are known to inordinately privilege the Global North while subjugating, and often exploiting, the Global South. Indeed, most global AI models are trained on populations from Europe or North America, and a significant barrier to medical AI in Africa has been the lack of representative data. However, SUSTAIN-AI ECG is trying to tackle this too.
They collect anonymised data to create the first clinical dataset of its kind, focused specifically on African populations. This dataset is made publicly available to ensure that diagnostic accuracy is maintained across diverse, local demographics and to enable other researchers in similar regions to build localised tools.
Currently, the SUSTAIN-AI ECG pilot is operational in nine Ethiopian hospitals, but the project has ambitions for further expansion. The technological foundation is proven and live. However, scaling to hundreds of rural clinics will require sustained infrastructure and training resources. But with 6,000 clinical cases already under their belt, SUSTAIN-AI ECG has already demonstrated that when AI is stripped of profit-first motives and redirected toward public good, it can be a vital tool in ensuring that life-saving innovation belongs to everyone.
