with AId for
By Stephanie Kreml,
With its deep neural network AI models, Cardiologs cloud-based platform gives all clinicians access to powerful ECG analysis.
Yann Fleureau co-founded Cardiologs in 2014 with the goal of assisting physicians so they could provide care more efficiently. He also wanted to create products to expand clinical access to more patients.
“We believe that expertise in cardiology should be scalable and accessible anywhere, anytime,” says Fleureau, who is also Cardiologs’ CEO. “We leverage machine learning (AI) and cloud computing technologies to fulfill this vision.” The French company has grown rapidly since its beginnings, and it now has a global reach with additional operations in Boston, Massachusetts.
Cardiologs received FDA 510(k) clearance for its ECG analysis platform as a Class II medical device in June 2017, making it the first-to-market AI-based clinical decision support solution for screening ECG recordings. Now Cardiologs’ AI model can detect more than 100 different ECG abnormalities.
Starting with the Holter monitor recordings from the MIT BIH ECG database, Cardiologs published its initial results for detecting atrial fibrillation with its deep neural network based AI model in 2016. This allowed the platform to achieve a positive predictive value of 91%, with a sensitivity of 97% and a specificity of 99% in detecting atrial fibrillation in its studies submitted for FDA clearance.
And this statistical strength continues to progress. Today the company iterates and improves its AI model with each new ECG it analyzes and has a growing database of more than 1.4 million ECGs.
Last September, Cardiologs published results for a head-to-head retrospective study comparing its AI-based model to the conventional Veritas algorithm. The study included over 24,000 consecutive 12-lead ECG in a six-month period.
Cardiologs deep neural network model showed superior accuracy and specificity in detecting atrial fibrillation as well as other atrial dysrhythmias, including atrial tachycardia and atrial flutter. This was true compared to both the Veritas algorithm by itself and with physician overread.
Strong Clinical Validation
The idea is not to replace [doctors], rather to help them so that they can make the best possible decision
Cardiologs complies with all regulatory standards, and it goes beyond the baseline FDA requirements with biannual software audits and monthly software updates to continuously improve performance and usability.
Cardiologs uses Microsoft Azure for its cloud-based platform with US-based servers. The HIPAA-compliant platform minimizes capital outlay for practices, allowing clinicians access from any internet-connected computer. The company implements industry best-practices for encryption to ensure data security.
Cardiologs’ platform generates reports within minutes, with the ability to customize these reports for the needs of different cardiology subspecialties. It also provides a broad array of tools for raw ECG signal analysis.
The platform’s interface provides easy navigation of ECG tracings and visual recognition of arrhythmias. It highlights AI-detected abnormalities and walks clinicians through a structured analysis of findings. Clinicians can confirm these findings and add them to the final report as needed.
Cardiologs’ platform not only improves the efficiency of cardiologists; it enables less skilled clinicians to have access to more accurate ECG interpretations. “Our vision is to create a technology that can help democratize access to expert level healthcare and eliminate the growing socioeconomic medical divide around the globe,” states Fleureau.
“The electrocardiogram, in itself, is the most basic cardiac sensor. It’s widely available all across the world, and in a lot of different areas of the world, especially in emerging countries,” Fleureau said during an interview with Forbes. “So, if there is no expertise to read properly the [ECG], well, the examination itself is pointless. By democratizing access to expertise in [ECG] analysis, we do raise the standards of cardiac care at scale globally.”
And while some may be concerned about the growing influence of AI-based technology in medicine, Fleureau seeks to ease clinicians’ minds. In an interview with MIT Technology Review last year he explains, “The idea is not to replace [doctors], rather to help them so that they can make the best possible decision, especially in delicate settings.”
However, Cardiologs’ AI-based technology is really geared to boost clinicians’ ability in the developed world. With its cloud-based platform, the company’s AI-powered models enable cardiology practices to increase their capacity for analyzing ECG data. Technicians using Cardiologs’ AI-powered solution take on average 5 minutes to analyze a standard Holter recording and only about 15 minutes to analyze the same seven-day Holter reading.. This increase in speed allows patients to receive a diagnosis and treatment plan during one post-study visit.
This sense of urgency drives Fleureau and the Cardiologs team. “There are thousands of patients being monitored at home with traditional medical devices that register palpitations for which there is no explanation. With AI we can monitor and diagnose arrhythmia conditions before it’s too late,” he told MIT Technology Review.
The Cardiologs platform is device-agnostic, so physicians can upload digital ECG data from any compatible cardiac monitor. This allows physicians to choose the monitors they prefer, and it prevents them from being locked-in to a particular device.
Covering All Bases
By democratizing access to expertise in [ECG] analysis, we do raise the standards of cardiac care at scale globally.
With Cardiologs success over the last few years, the company successfully raised $15 million in Series A funding in January. This capital infusion will allow Cardiologs to increase its market reach and empower more clinicians in North America and Europe. The company will also use these investment proceeds to continue platform improvements and add new integrations and expanded applications.
“We have developed a new category of heart disease diagnostic products powered by AI that promise to revolutionize healthcare by delivering accurate, cost-effective, and timely expert-level diagnostics,” Fleureau says. No doubt, Cardiologs is set to democratize the access to instant, reliable and affordable expertise for every patient, every test, everywhere.
Poised to Grow
We believe that expertise in cardiology should be scalable and accessible anywhere, anytime.