It is important to take stock of the challenge the world faces in providing healthcare services to a growing global population. Indeed, up to 3.5 billion people – almost half the world’s population – lacks access to the healthcare services they need, according to the World Health Organisation (WHO).
The causes of this uneven distribution of healthcare services are complex and multifaceted, and addressing it will involve herculean efforts from governments, research organisations, and the public and private sectors. No matter how daunting this task might appear, I am hugely optimistic about the role of AI which, I believe, will be key to improving access to healthcare in traditionally underserved communities.
As a Machine Learning specialist dedicated to applying AI to problems in medical image analysis, I previously worked with a team of researchers to develop a solution to help healthcare professionals assess fetal growth at a speed and accuracy that would not otherwise be possible.
At MBZUAI, we are in the early stages of research into how machine learning can be used to detect cardiovascular issues in CT scans, and even predict complications that could arise in the future. To this end, we are working on a large study in collaboration with the University of Oxford to find biomarkers that could predict possible heart problems before they manifest themselves through more recognisable symptoms. The team has already recruited 100,000 participants and will perform CT scans on them. We will then teach the AI system to detect the difference between healthy cardiovascular systems and those with – or likely to develop – issues.
This large Oxford CT scan study is exciting, but I also want to take it further and ensure that similar transformative technology reaches people in developing countries and remote areas, where problems such as cardiovascular disease and fetal abnormalities often go undetected – frequently with devastating consequences. But expanding the solution in this way would present some unique challenges, including how to perform scans on people without access to a modern clinic or hospital. This is where new technologies, in the form of handheld ultrasound devices, and the ability to push AI to the edge of the network, hold the potential to save countless lives.
Along with my team of researchers at MBZUAI, we are working on how to use ultrasound scans for predictive diagnosis. The potential benefits of combining portable ultrasound scanners with AI in developing countries are amplified because it could enable medical professionals to scan patients’ hearts and monitor the AI system’s interpretation of these scans. This newfound information could highlight serious problems that need further attention.
It is worth highlighting that the AI component of this medical treatment does not supplant the human clinician, rather it helps by doing the heavy lifting of interpreting scans and flagging those that need attention. In many parts of the world, this service could give millions of people their first access to reliable cardiovascular checks.
The same principle applies to assessing fetal growth and detecting abnormalities. We plan to conduct research aimed at using portable ultrasound scanners to perform the type of ‘anomaly scan’ that is
standard in developed countries, but performed on a much smaller scale in less developed regions, and can be considered rare in villages in much of Africa and South East Asia. Just as with cardiovascular scans, we will need to teach the AI system to learn how to assess the growth of a fetus and flag the ones that need further attention. The beauty of such a system is that it could be delivered to smaller clinics by nurses or midwives.
These scans are hugely important because, when anomalies such as congenital heart defects or spina bifida are detected early, they can be treated in-utero or soon after birth. Moreover, the use of AI to detect anomalies has the potential to reduce the time taken to assess each scan from a mean average of between 40 minutes and 1 hour to just 10-20 minutes.
All of this remains at a very early stage, but working alongside my gifted team at MBZUAI, I am optimistic that we can play our part in improving healthcare for millions of people who are currently underserved.
As the American writer William Gibson said: “The future is already here, it’s just not evenly distributed yet.” At MBZUAI we’re determined to bring the blessings of science – and particularly AI – to people in all parts of the world.
