Jimenez and Ali co-founded AccurKardia in 2019 with a vision for unlocking the value of the ECG signal, and the company currently markets one of the few FDA-cleared solutions for automated ECG interpretation and arrhythmia detection.
Here’s the rest of their story and vision for ECG automation.
Let’s start things off with some background on Accurkardia. Can you share a little about the company and your AccurECG solution?
Juan C. Jimenez: Accucardia is an ECG-led diagnostics company, leveraging ECG as a broad biomarker for cardiology and beyond.
Our first FDA cleared product is AccurECG, a device agnostic, fully automated ECG interpretation software platform that ingests ECG data from wet electrode devices such as Holter monitors, mobile cardiac telemetry (MCT) devices, and event recorders. We perform automated analysis of these recordings, similar to what a cardiovascular technician might perform for cardiologists or electrophysiologists.
AccurECG provides automated heart rate measurements, supraventricular and ventricular ectopic beat detection, and automated interpretation for 13 different arrhythmias such as atrial fibrillation, atrial flutter, and ventricular tachycardia.
It can be integrated with software from the different cardiac device makers via an API and integrates with the EHR or with telehealth and remote patient monitoring platforms in a HIPAA compliant manner.
We’re not adding steps to the ECG analysis process. AccurECG is meant to automate the interpretation work, reducing technicians’ time spent on analysis by 70%, so we can provide physicians a more consistent, faster, and affordable option for ECG interpretation at scale.
How is ECG analysis performed today?
Juan C. Jimenez: There are many use cases of how ECG analysis is performed. One example is long duration ECG analysis, which could involve producing a report of 24 or 48 hour Holter monitor recordings.
Currently, most of these recordings are analyzed by a combination of software and cardiac technicians. This process is largely manual and requires multiple steps.
Some cardiac monitoring companies have also outsourced these interpretations to cardiac technicians outside of the US in order to manage their increasing volumes and offset margin pressures from reimbursement rate reductions.
Even though most doctors might require an interpretation in two or three days, it often takes five to 15 days for doctors to get it back, and sometimes interpretations arrive too late.
AccurECG can help cardiac monitoring companies provide consistent results in a prompt manner while improving efficiencies.
How does AccurECG change this for patients and physicians?
AccurECG is changing ECG analysis in three areas.
The first is the speed of the analysis. Our platform works in what we call “near real-time,” and reduces the time it takes to analyze a long duration report to hours if not minutes. This gives doctors their ECG reports faster, so they can start treatments earlier, and improve patient outcomes.
The second is consistency. ECG signals are hard for the human eye to analyze. By automating and standardizing ECG analysis, AccurECG allows more consistent interpretation, leading to greater consistency in clinical treatments.
Third, we help cardiac monitoring companies better manage their cost structure by reducing the time their teams spend analyzing ECGs. We can help reduce the time a cardiac technician spends on a long duration ECG report by up to 70%. This significantly increases their throughput, and enables their technical interpretation staff to manage ever increasing ECG volumes.
Cardiology is an evidence-driven specialty. How have you proven AccurECG’s benefits?
Juan C. Jimenez: The clinical validation that we performed to support our FDA clearance consisted of two areas.
We performed our validation using a publicly available dataset required by the AAMI/ANSI EC57 standard, which is the gold standard, to benchmark ECG analysis for beat by beat measurements and arrhythmia detection.
We also conducted a 1,000+ patient clinical validation study comparing a three-cardiologist adjudication panel consensus versus the performance of our AccurECG software platform. At the time of our FDA clearance, this was one of the largest clinical validations for automated ECG arrhythmia detection performed for device agnostic interpretation platforms available in the US market.
How do you see patient and physician perspectives on AI evolving?
Mohamed Sadeq Ali: At this stage there is a lot of skepticism with AI in general, leave alone AI in healthcare.
There’s good reason for that skepticism. I think the biggest of all, particularly when it comes to healthcare, is the black box nature of AI, in particular the lack of explainability regarding why a specific diagnosis is made. One of the things that AI will have to address is explainability. It’s going to be critical to provide clinicians the rationale for a particular diagnosis.
Also I frankly don’t think AI will be exclusively used in clinical diagnosis – we will combine what AI provides alongside other inputs that don’t rely on AI and a clinician’s own experience and reasoning.
Transparency, the use of non-AI capabilities, combined with clinical factors, will start to make AI more accepted by both patients and physicians.
Eventually, I think not having AI in the mix won’t be an option, because as we move towards a world where there is so much more data than there used to be, you’ll absolutely need AI to filter the signal from the noise.
Without AI, clinicians and already stretched healthcare systems will simply be drowning in data, and having a lot of data is in some sense worse than having no data. It’s going to become an imperative to bring AI into the mix to filter through all of that data.
What’s the tipping point in healthcare AI adoption?
Mohamed Sadeq Ali: I think we’re already at that tipping point.
We are at a point right now where we are seeing healthcare systems around the world stretched. We’re seeing clinicians being pulled into things where AI can really help. It is crucial for healthcare providers that are at the front lines to be addressing a patient’s most critical needs as opposed to annotating an ECG.
We’re at a point where machines are operating at a high enough level of efficiency where they can do a lot of the grunt work, so that the clinician can focus more on serving the needs of the patient, analyzing the edge cases, looking through their lens of experience at things that a machine perhaps can’t.
We have health systems across the world, not just in the US, completely overloaded, and without the aid of AI we’re just not going to be able to solve the problems we currently face.
What advice would give an ECG monitoring team thinking about making the move to AI ECG analysis?
Juan C. Jimenez: Well, I would recommend that they partner with AccurKardia.
Other than that, they should understand that AI right now is not a replacement, it is a tool to assist in this particular use case.
They should understand that they should use automated ECG interpretation tools to handle repetitive manual tasks. That will allow physicians, clinicians, and technicians to dedicate more time to the most necessary areas, like direct patient care or more complicated analysis which requires clinical intuition built from years of experience.
Understanding what’s the best role for technology to play and what’s the best role for humans to play is essential to be able to adopt AI with minimal friction.
Mohamed Sadeq Ali: Anyone in healthcare looking to adopt AI should focus more on the quality of the data than the quantity of the data. I know people are extremely enamored with this whole idea of big data and that more is better. In healthcare, that’s not necessarily the only thing to concern yourself with. The quality of human annotation and the longitudinal nature of the data matters a lot as well.