AI-Powered Healthcare Could Save Your Life

It was 1895 when Willem Einthoven, a Dutch doctor, made a discovery that would charge cardiology for the next century. Fiddling with wires and voltages, he traced the electric currents flowing through a beating heart. This became the first practical electrocardiogram (ECG), a device that visualizes the heart’s inner workings.

For decades, ECG technology remained much the same, with leads patching onto the skin to record the heart’s electrical impulses. But human perception has limits. Now, AI is taking medical analysis to profound new levels, unlocking lifesaving discoveries in terabytes of data. For example, 2,322,513 ECG records were used in a Nature research article to train deep neural networks to classify arrhythmias more accurately than cardiology specialists.

Let’s rewind for a moment. Heart disease has become the leading cause of death globally, causing around 51,000 deaths a day. Visualizing this data is striking.

Many show no prior symptoms, the first sign being their last. But what if we could detect threats early and prevent these deaths?

This is the promise of firms like HeartSciences and Anumana, medical companies harnessing AI to transform the humble ECG into a cutting-edge screening tool. HeartScience’s flagship product, MyoVista Wavelet ECG, incorporates sophisticated machine learning to uncover hidden heart problems. And a new partnership with Mount Sinai Hospital — home to pioneering AI research — could be the catalyst that saves lives worldwide.

Deep Neural Networks to Save Hearts

The standard 12-lead ECG has served cardiologists well, visualizing electrical signals as squiggly lines on graph paper. But it has limitations. The test often misses two major disease categories — ischemia (clogged blood vessels) and structural issues like weakened heart muscles. This is where MyoVista steps in. Its AI algorithms spotlight subtle ECG patterns missed by the human eye, acting as an early warning system.

One model uses a set of 51 ECH features, 26 of which come from Continuous Wavelet Transform (CWT) frequency features—the remaining come from traditional ECG parameters, along with clinical data like age and medical history.

This data makes it possible to train a DNN (deep neural network), boasting areas under the receiver operating characteristic curve (AUC) ranging between 0.80 and 0.87. In other words–these neural networks can accurately predict MACE (major adverse cardiovascular events).

These models have been trained and validated on patient data from institutions like the Icahn School of Medicine at Mount Sinai and West Virginia University. Such machine learning techniques are transforming medical imaging and diagnosis across specialties. But cardiology is particularly ripe for disruption as heart disease causes ever more deaths globally. MyoVista promises a quick, non-invasive test to uncover threats in routine doctor visits. And this early detection can make all the difference.

Algorithms Making Their Way Into Hospitals

ECG electrodes have become ubiquitous, deployed far beyond cardiology units. General practitioners, ER doctors, and specialists of all stripes snap ECG leads onto patients. But most lack specialized expertise to maximize this tool and miss subtle signs of cardiac dysfunction. MyoVista fills in these analytical gaps with AI acting as a digital cardiologist that spots what human doctors cannot.

In one validation study, MyoVista achieved 94% accuracy in prediction of LV diastolic dysfunction, as referenced in a recent report. But the algorithms powering MyoVista require hefty data resources to develop — far beyond most companies’ capacity. This is where HeartSciences’ recent partnership with New York’s Mount Sinai Hospital proves game-changing.

The Icahn School of Medicine at Mount Sinai has emerged as a global leader in AI-driven cardiology research. Their elite data science team has cultivated databases of curated heart patient records — an unparalleled asset. Now, HeartSciences has licensed exclusive access to this goldmine, including 13 Mount Sinai algorithms spanning major cardiac diseases. This agreement catalyzes their ambitions to transform ECG screening worldwide.

Like most AI software penetrating healthcare, HeartSciences needs to run a gauntlet of quality checks before commercializing MyoVista. Previously, novel digital diagnostic tools like theirs required lengthy De Novo FDA applications before clearance. But there’s a faster 510(k) pathway for HeartSciences’ model. This simplified process means that HeartSciences’ algorithms could soon be in your local hospital.

Additionally, HeartSciences has proactively engaged the FDA multiple times to optimize MyoVista for compliance. They also recently completed patient enrollment for their pivotal validation study spanning five US research sites and over 600 subjects. This data will demonstrate MyoVista’s capabilities on a wide demographic, paving the way for approval.

The Road Ahead

Heart disease deaths continue rising yearly, hastened by modern lifestyles and aging populations. Better prevention through earlier detection has become imperative. HeartSciences leads a growing field of companies centered on this mission — leveraging AI’s pattern recognition to uncover what standard tests miss.

The ECG is now well over a century old. But it has now outgrown the bulky monitors and needle graphs conceived by its Dutch inventor. AI is writing the next era of cardiac diagnostics with software smarts. In the right hands, this old tool may gain new power to prevent the #1 killer from claiming lives too soon.