Digital therapeutics startup Biofourmis reported on Tuesday the receipt of clearance from the US Food and Drug Administration (FDA) for its RhythmAnalytics cloud-based software for automated interpretation of cardiac arrhythmias.

The company said the RhythmAnalytics platform uses an enhanced deep-learning technique to detect over 15 types of cardiac arrhythmias along with beat-by-beat morphology computation which include ventricular arrhythmias, ventricular ectopic beats and all non-paced arrhythmias including Atrial Fibrillation (AFib). The AI-models were trained using over a million ECG recordings collected using various US FDA-cleared ECG monitoring wearables.

In conjunction with the US FDA approval, the company said it has signed partnership with Mayo Clinic, Brigham and Women's Hospital, among others, for the RhythmAnalytics, which continuously monitor for cardiac arrhythmias and manage patients at home. The patient numbers have doubled on a month-on-month basis.

Biofourmis added that it intends to offer RhythmAnalytics cloud-based API as Software-as-a-Service (SaaS) to cardiac monitoring organisations to improve accuracy and scalability of their ECG analysis, thereby improving throughput and efficiencies in their cardiac monitoring centres.

According to the company, it demonstrated cardiologist-level arrhythmia detection and classification using a novel-deep learning architecture, which was submitted to the US FDA. Over 120,000 single-lead ECG episodes were used to train the neural network with over 30 distinctive cardiac arrhythmias.

Additionally, the RhythmAnalytics outperformed two other similar deep learning systems as well as a panel of cardiologists with Sensitivity of 90.8%, Specificity of 98.2% and an overall F1 score of 0.834, concluded the company.