My concerns about our current approach to cardiology imaging

I've been unfortunate enough to spend a lot of time recently with my son on the cardiac ward of our local hospital. So I've experienced first hand some of the pain experienced by our healthcare partners on a daily basis, caused by technology and equipment that doesn't meet the needs of clinicians or patients.

Thankfully some of these problems Loadbalancer.org can actually do something about. Some of which we can't. So I'm sharing my personal account here in the hope that, in time, it might prevent another family from experiencing the same frustrations.

Takeaway 1: High-end tech isn't always a good thing

My son has experienced a lot of chest pain recently, resulting in us spending hours upon hours over the last few weeks hooked up to various ECG machines. What I hadn't fully appreciated (prior to our many admissions) is the complexity of some of these machines, and the range of clinical applications they're used for.

The ECG machines used by paramedics to assess the condition of my son were different to the ECG machines used by clinicians to diagnose him, which were, in turn, different from the ECG machines used to monitor him on the ward. On each occasion the machines were used to monitor cardiac rhythms, but the complexity and functionality of these machines varied wildly.

What I observed was that with the high-end machines clinicians struggled to get their respective ECG machine to work to their own ends — with the complexity of the equipment working against, not for, them. My son, for example, had to be monitored on the ward while he was sleeping. No mean feat for the high-end, ECG machine that was incredibly sensitive to movement, meaning he had to be woken multiple times in the night to have the machine reset.

All of this led to:

• Clinicians having to repeat ECG readings until they were satisfied they had managed to take a reliable reading.

My takeaway? Complex technology and high-end solutions are less important than their ability to facilitate the clinical intervention they're being used for. Instead, clinicians need technology with better task-technology fit. The end goal should therefore not be to simply buy technology with the longest list of features. Technology needs to make people's lives easier, not harder.

Loadbalancer.org CAN help in a small way with this by offering load balancers with task-technology fit technology at their core. Meaning we put emphasis not just on functionality, but also on the way people need to use and deploy our software, making it easier to use. Because (quite frankly) the IT infrastructure challenges faced by hospitals and clinicians on the front line are complex enough.

Takeaway 2: Hard copy ECGs delay interpretation and data sharing amongst teams

ECG data is still printed hard copy in many hospitals, and this was very much our experience. The ambulance ECGs have to be printed off and physically handed over to the A&E department, who then did their own hard copy printouts, before the children's ward then had to do their own. At each stage the physical scanning and uploading of these readings required the clinicians to step away from their sick patients.

This results in:

• Hours of additional time spent by clinicians on admin each week. And it's not just ECGs that still exist in hard copy but need to be shared electronically — chest x-rays, echocardiograms, CTs and MRI scans also need to be uploaded and shared with consultants in other departments and hospitals. All of this has a significant knock-on effect. Especially if one of their scanners or printers go down...
• Furthermore, if the scans are not uploaded and integrated correctly with the patient's own medical records, this can further jeopardize the clinician's ability to provide their patients with the best care.

My thoughts? We have the technology out there to do things differently, but the complex spaghetti of healthcare IT infrastructure means change is not always easy. Loadbalancer.org can't sadly do anything about the continued reliance on hard copy medical records in hospitals. But we CAN help by providing high availability for the medical imaging equipment itself, as well as for the many hospital scanners, printers, and DICOM viewers — helping them avoid unnecessary delays and downtime that might otherwise impact patient care.

Takeaway 3: Being able to accurately interpret ECG readings is critical for maximum ROI!

ECG readings are famously difficult to interpret, so it's no wonder many of my son's ECG readings were deemed to be 'inconclusive'.  Ultimately, were weren't given an official diagnosis until his ECGs had been scanned and reviewed by a paediatric cardiologist at another hospital, despite being reviewed by the Paediatric Registrar and numerous consultants. All of this places significant pressure on the most senior doctors to interpret large numbers of ECG readings.

This can lead to:

• Errors in diagnosis, as well as unmanageable workloads

The takeaway? ECG interpretation is an incredibly complex task that requires knowledge of anatomy, electrophysiology, pathophysiology, visual pattern recognition, and experience of diagnostic reasoning. What one consultant told us was indicative of a cardiac issue, another told us was 'within the realms of normal'. Given such diverse interpretations of the same ECG reading, surely the adoption of AI applications that have a higher accuracy rate should be accelerated? Especially when research published in Nature found that Deep Neural Network (DNN) machine learning algorithms demonstrated a superior ability to classify ECGs (with a predictive value of 0.84, compared to the 0.78 predictive value of human cardiologists).

Loadbalancer.org CAN help in a small way with this by load balancing medical imaging applications, as well as the object storage data that underpins AI software. This ensures these applications remain not only highly available, but are also able to provide the high throughputs required for image analysis.

The way forward?

Given the amount of time Loadbalancer.org spends every day working with US and UK medical imaging vendors, we believe that honesty about the challenges faced is important to help us move forward. Hence this blog. After all, honesty is how we roll — even if that means occasionally having to say 'no' to a customer!

There is no more vulnerable individual than a sick patient, and clinicians are on their knees trying do the best they can while drowning in high workloads, limited resources, and complex patient workflows. So why not start a conversation with us about what 'doing things differently' might look like? Not because we'll always be the right fit for your specific use case (hospital IT infrastructure is complex, remember...), but because we care about your outcome, not our time.