SignalHF (IM008)

The SignalHF System is an adjunctive predictive cardiovascular indicator designed to assist healthcare professionals by calculating a 30-day risk score for heart failure worsening in patients with compatible cardiac implantable electronic devices. Its methodology relies on a machine learning algorithm that analyzes a combination of physiological data from the CIED and patient demographic information, with its performance validated through a large-scale retrospective clinical study.

The SignalHF System is a software as a medical device (SaMD) intended for use by healthcare professionals managing adult patients (over 18 years old) undergoing heart failure (HF) surveillance and implanted with compatible Cardiac Implantable Electronic Devices (CIEDs), such as pacemakers, ICDs, and CRTs. It employs a proprietary machine learning algorithm, the SignalHF HF Score, to analyze a diverse set of physiological measures from the patient's CIED (including activity, atrial burden, heart rate variability, heart rate, heart rate at rest, thoracic impedance for fluid retention, and premature ventricular contractions per hour) along with personal health records like demographics. The system's purpose is to provide an adjunctive score indicating the risk of a patient experiencing a worsening HF condition or requiring hospitalization within the next 30 days, aiming to detect chronic worsening with alerts typically raised a median of 30 days before a predicted event, rather than providing real-time crisis alerts. The performance of SignalHF was evaluated in the FORESEE-HF study, a non-interventional retrospective clinical study involving 17,974 patients, utilizing data from CIEDs and the French national health database (SNDS). For patients with ICD/CRT-D devices, the system demonstrated a sensitivity of 59.8% for detecting HF hospitalizations, an unexplained alert rate per patient-year (UAR PPY) of 0.654, and a lower quartile on alerting time of 35 days before hospitalization. For pacemaker/CRT-P devices, the sensitivity was 45.9%, UAR PPY was 0.470, and the lower quartile alerting time was 37 days. Performance varied by device manufacturer (Medtronic, Boston Scientific, Biotronik) and specific device configurations, with most groups meeting predefined endpoints. However, data was insufficient for Boston Scientific pacemaker/CRT-P devices, and Medtronic 1-lead pacemakers did not meet all endpoint requirements. The device labeling includes a warning that in patient populations unlikely to have acute HF decompensation, the algorithm's positive and negative predictive values may be reduced.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the SignalHF algorithm weigh the various physiological inputs (e.g., thoracic impedance, activity levels, heart rate variability) and demographic data in calculating the HF Score, and is this weighting adaptable or fixed?; What is the performance of the SignalHF system in diverse patient populations not extensively represented in the FORESEE-HF study, particularly concerning different ethnicities, stages of heart failure, or specific comorbidities?; Given the median 30-day advance alert for potential HF hospitalization, what specific clinical intervention pathways are envisioned or have been studied in conjunction with SignalHF alerts to optimize patient outcomes?; How does the 'unexplained alert rate' vary with patient baseline HF severity, frequency of remote data transmission, or the presence of non-cardiac conditions that might influence CIED-measured parameters?; What are the specific data quality thresholds and requirements for the CIED-derived physiological measures and personal health records to ensure the optimal performance and reliability of the SignalHF algorithm, and how does the system handle missing or anomalous data points?