Transforming heart health with smart data

It’s no secret that biomedical engineering has seen a lot of advancement in recent years, particularly in the development of methods to monitor heart health using data from wearable devices like smartwatches. However, the novel approach of the LHMF has been a major step forward in predicting the risks of heart disease and heart attacks more accurately.

Utilising the concept of "digital twins" to create a detailed 3D model of a patient’s blood flow, the LHMF model is based on data collected over a period of time, covering more than 700,000 heartbeats. This method is in contrast to the current standard in heart disease evaluation, which typically relies on snapshots of a single moment in time. Such snapshots can be inadequate for things like heart disease which develops gradually over months or years.

One of the primary challenges in creating these detailed simulations has been the immense computational power required. To address this, the Duke University team, led by Amanda Randles, developed a software package called HARVEY, which allows for more efficient processing of these complex simulations.

By simulating heartbeats in parallel rather than sequentially and breaking the task up among many different computing nodes, the team has reduced what previously took nearly a century of simulation time to just 24 hours. This is achieved by making reasonable assumptions about the lack of impact of certain time-specific coronary flows on others, thus allowing the team to simulate different time chunks simultaneously and later piece them back together.

This approach has several practical applications. For instance, it can be used to determine whether a patient requires a stent for treating arterial plaque or lesions, a method far less invasive than traditional approaches. More importantly, the technology aims to track pressure dynamics over extended periods, such as weeks or months after a patient leaves the hospital, providing insights into the risk of heart disease and symptom recurrence.

To validate their approach, the researchers tested the LHMF using both Duke's computer cluster and cloud computing systems like Amazon Web Services. They found that the errors in the LHMF simulations were negligible compared to traditional methods, demonstrating its effectiveness on a clinically relevant time scale.

The LHMF was further refined to track the frictional force of blood on vessel walls, a known biomarker of cardiovascular disease, allowing the researchers to create a personalised, longitudinal hemodynamic map for each individual. This map shows how the forces vary over time and identifies the percentage of time spent in various vulnerable states, offering insights into the progression of heart diseases like atherosclerosis.

This work represents a significant advancement in the use of wearable technology for health monitoring, potentially allowing for the early identification and more accurate tracking of heart disease.