A few years ago, it was all about AI. Now, it’s all about generative AI. In fact, the UBS Truth Lab recorded almost 500 companies across 27 sectors making more than 3,500 references to generative AI in earnings calls during the first 5 months of 2023.

Healthcare has been pointed out as a particularly attractive setting for generative AI, where large amounts of unstructured data sets such as clinical notes, diagnostic images, medical charts and recordings often require vast amounts of manual brain power to decipher, analyse and correlate into actionable outcomes. The deep-learning algorithms that fuel generative AI can process these inputs into structured data that can one day hopefully be relied upon by doctors.

However, this is far from the only area of healthcare where efficiency needs to be further enhanced – and it’s not just about generative AI. The components mentioned above that form the swathes of unstructured data that generative AI could be of use to, each have their own shortcomings that innovative companies around the world are working tirelessly to solve – and some especially interesting ones are right here in Australia.

In order to drive efficiency in the healthcare system, while still striving for ever growing levels of accuracy in the face of declining amounts of physicians, medical practitioners must be armed with tools that they can rely on. The interpretation of diagnostic images, such as CT scans, MRI’s, X-rays and ultrasounds is a practice typically reserved for radiologists and specialists (in their particular area of medicine). It’s a practice that is, by the very nature of it – tremendously  vulnerable to human error.

The world’s biggest healthcare problem, cardiovascular disease (CVD),  claims around 18 million lives each year and has been the leading cause of death in the US since at least 1950. More than four out of five CVD deaths are due to heart attacks and strokes, and one third of these deaths occur prematurely in people under 70 years of age.

Computed axial tomography (CAT/CT) scanners virtually changed the face of medical sciences when they were invented by biomedical engineer Godfrey Hounsfield in 1972, but have since undergone rapid development that have significantly broadened their capabilities. The development of multislice CT scanners 20 years later, which started with two but now stretch all the way up to 640 slice counts, allowed these powerful machines to become a mainstay in cardiac imaging due to their newfound ability to detect and assess the degree of coronary stenosis.

Coronary artery calcification is a collection of calcium in the heart’s coronary arteries, which happens after plaque (fat and cholesterol) form in the arteries (atherosclerosis) for around five years. Traditionally dominated by invasive testing with a catheter that is inserted into the femoral artery and guided to the heart, Coronary CT Angiography (CCTA) has largely replaced this practice when assessing coronary artery disease (CAD) risk. 

Advancements in CCTA imaging have allowed it to identify plaque that is calcified, non-calcified or mixed. However, the interpretation of these images is a lengthy process that usually takes up to 45 minutes. Critically, soft plaque, which is highly unstable and is most liable to rupture, which can eventuate in a heart attack, is difficult to identify and rarely reported. In a time critical environment, cardiologists need to do the best they can in a reasonable amount of time so they can move on to the next patient.

In 50% of men and 64% of women, the first symptom of CAD is death. However, the degree of stenosis (narrowing of arteries caused by atherosclerosis) is not the sole indicator of cardiovascular risk, despite current testing having a focus on stenosis greater than 50%. In fact, a study found that more than two thirds of those who died from cardiovascular events had less than 50% stenosis.

Enter Artrya (ASX: AYA) and their Salix technology, which is able to turn this crucial 45 minute job into a process that takes less than 15 minutes from scan time to actionable report. Salix can identify stenosis greater than 50% with a 96% degree of accuracy, while bringing the same level of correlation to a human’s ability to deliver a calcification score. Importantly, its AI algorithms specialise in identifying vulnerable plaque. Artrya is launching their technology in Australia this year, where over 120,000 CCTA’s are performed each year. The company is also seeking FDA approval in the US, where 2.5 million CCTA’s were performed in 2021 – a number expected to blow out to 4.4 million by 2025.

The near term opportunity for AI in healthcare is to assist physicians in diagnosing and treating patients faster, not necessarily to replace doctors entirely.

A report prepared for the Association of American Medical Colleges estimates that the US will face a shortage of between 37,800 and 124,000 physicians by 2034. The American College of Cardiology (ACC) highlighted the growing concerns around the annual net reduction of cardiologists, which is currently 547, who are a part of the US workforce.

 According to the report, the specialist practice consists of 32,000 cardiologists, 8,480 of which are over the age of 61, and 2,000 of which are estimated to be lost every year – who are replaced by just 1,453 entering the workforce. The ACC paints a stark picture of growing demand for cardiology services that will have an insufficient supply response and ultimately deprive people of adequate healthcare.

While these statistics may seem alarming, solutions such as Artrya and many others that will hit the market before the end of the decade can certainly help fill the gap, by saving doctors precious time that could be used to treat ever-increasing amounts of patients. Heart disease costs the United States US$219 billion dollars every year. In Australia, someone dies of a heart attack every 13 minutes. Healthcare needs AI-driven solutions to overcome these problems.