Cardiomatics Secures $3.2M Funding for ECG-Reading AI

Cardiomatics Secures $3.2 Million Seed Funding for ECG Automation

Cardiomatics, a Polish health tech startup specializing in artificial intelligence, has announced a successful seed funding round of $3.2 million. This investment will be used to broaden the application of its technology automating the interpretation of electrocardiograms (ECGs).

Investment Details

The funding round was spearheaded by Kaya, a venture capital firm focused on Central and Eastern Europe. Additional participation came from Nina Capital, Nova Capital, and Innovation Nest.

Furthermore, Cardiomatics received a $1 million non-equity grant from the Polish National Centre of Research and Development, supplementing the seed raise.

Technology and Functionality

Founded in 2017, Cardiomatics provides a cloud-based tool designed to accelerate diagnosis and enhance efficiency for healthcare professionals.

The software automates the detection and analysis of approximately 20 different heart abnormalities and disorders. It generates reports on ECG scans in a matter of minutes, significantly faster than traditional manual analysis by a specialist.

Democratizing Healthcare Access

Cardiomatics emphasizes its role in expanding access to healthcare. The tool empowers cardiologists to optimize their workflows, enabling them to treat a greater number of patients.

It also allows general practitioners and smaller clinics to offer ECG analysis without the need for referrals to specialized hospitals.

Current Usage and Integration

To date, the AI tool has analyzed over 3 million hours of ECG signals commercially. The company reports a customer base exceeding 700, spread across more than 10 countries, including Switzerland, Denmark, Germany, and Poland.

The software is currently compatible with over 25 different ECG monitoring devices. Cardiomatics highlights its modern cloud interface as a key advantage over older, legacy medical software systems.

Accuracy Validation

Regarding the validation of its AI’s ECG reading accuracy, Cardiomatics stated: “Our algorithm development utilizes a dataset containing more than 10 billion heartbeats from approximately 100,000 patients, and this dataset is continually expanding.”

The company builds the majority of its datasets internally, supplementing them with publicly available databases.

Data-centric AI is a core principle, with 90% of the data dedicated to training and 10% reserved for validation and testing. The startup prioritizes representative test sets reflecting real-world client signals.

Accuracy is assessed experimentally on a monthly basis during ongoing algorithm and data development. Clients also evaluate accuracy daily in clinical practice.

Future Plans and Expansion

The seed funding will be allocated to product development, expansion within existing markets, and preparation for entry into new territories.

“These funds will support our rapid expansion across Europe, scaling our leading AI technology and ensuring an optimal experience for physicians,” a company representative explained. “We are also preparing for launches in new markets, including pursuing FDA certification for the US market.”

Regulatory Compliance

Cardiomatics received European medical device certification in 2018. However, the regulatory environment for medical devices and AI within the European Union is evolving.

The EU Medical Device Regulation, which came into effect earlier this year, and the forthcoming Artificial Intelligence Act will likely increase compliance requirements for AI health tech companies like Cardiomatics.

Adapting to the Regulatory Landscape

The company stated: “We were among the first AI-based solutions approved as a medical device in Europe in 2018. We closely monitor developments in European regulations regarding AI.”

“We will promptly implement any new standards or requirements, extending documentation and algorithm validation to ensure the reliability and safety of our product.”

Challenges in Assessing Algorithm Efficacy

Cardiomatics acknowledged the difficulties in objectively measuring the effectiveness of ECG reading algorithms.

“Objective assessment is often limited to narrow datasets from specific patient groups and single devices. We receive signals from diverse patient groups using various recorders, and we are actively developing methods to address this challenge.”

The company is working on algorithms that can reliably evaluate performance regardless of factors like recording device or patient demographics.

The Intersection of AI and Medical Expertise

“ECG interpretation by a physician involves experience, established rules, and a degree of artistry. A human specialist visually interprets the ECG curve, while an algorithm processes a stream of numbers, framing the task as a mathematical problem.”

“However, effective algorithms require domain knowledge. The knowledge and experience of our medical team are crucial to Cardiomatics. Algorithms are also trained on data generated by cardiologists, highlighting the strong correlation between medical expertise and machine learning.”