Calibration of AlgoRithm for Detection of Cardiac Decompensation Via Parametric Objects (CARDCOP)

Overview

The goal of the study is to calibrate the algorithmic model of the TakeCoeur AI device to detect early heart failure decompensation in patients with heart failure, using physiological data (clinical) actively and passively collected through connected medical devices (watch, blood pressure monitor, and scale).

In France, more than one million people suffer from chronic heart failure (HF) with an average age of 75 years. HF is a major public health issue, leading to 160,000 hospitalizations and 70,000 deaths each year, with an annual hospitalization cost estimated at €1.6 billion. Heart failure remains the third leading cause of cardiovascular mortality, with a 6 to 9 times higher risk of sudden cardiac death in HF patients compared to the general population. After the age of 80, approximately 25% of heart failure patients die within three months of a decompensation episode, and nearly 40% within a year. Despite significant progress, the therapeutic management of patients remains complex, and hospitalizations are difficult to anticipate. It is even estimated that nearly 400,000 to 700,000 people with HF remain undiagnosed in France. Avoidable hospitalizations contribute to a deterioration in the quality of life for these patients and sometimes result in death. Heart failure is a condition characterized by gradual worsening. Initially, the patient may be asymptomatic but progressively start to experience several, if not all, of the following symptoms: marked shortness of breath (dyspnea), edema, difficulty breathing when lying down (orthopnea), rapid weight gain, chronic fatigue, heart palpitations, and a drop in blood pressure upon standing. Monitoring the evolution of these symptoms is crucial to identifying the risk of heart failure decompensation and allowing for early and appropriate intervention to avoid hospitalization. The progression of these symptoms can be tracked through the monitoring of physiological (clinical) variables such as weight, blood pressure, heart rate, and oxygen saturation. These are widely recognized parameters as indicators of health status and are frequently used in heart failure patient monitoring algorithms, as shown by systematic reviews. Weight is one of the mandatory parameters to be collected according to HAS recommendations. The American Heart Failure Association (HFSA) and the guidelines of the European Society of Cardiology (ESC) also recommend daily weight monitoring. Blood pressure is also among the recommended clinical symptoms to monitor, as it is a precursor sign of heart failure decompensation. The four warning signals-shortness of breath, rapid weight gain, lower limb edema, and fatigue (EPOF)-must be monitored, especially after the age of 60, to promote early diagnosis and prevent hospitalizations. Resting and exercise heart rates can predict the risk of cardiovascular disease. A high resting heart rate is associated with an increased risk of coronary heart disease and all-cause mortality and is also considered a predictive factor for decompensation in HF patients. Respiratory rate has been found to be significantly lower in patients with heart failure decompensation. However, physical activity is linked to better cardiovascular health and reduced mortality . Recommended by the American Heart Association (AHA) as one of the "8 Simple Measures for a Healthy Life", physical activity helps promote heart health. Continuous and real-time collection of all these physiological (clinical) parameters in heart failure patients at risk of decompensation could improve patient follow-up and help predict the risk of acute heart failure decompensation. Indeed, a better understanding of these parameters, their variations, and their correlations would allow for better characterization of heart failure patients and early identification of decompensation. This study aims to identify predictive factors of heart failure decompensation to develop and train an early detection algorithm that will be used in telemonitoring after the algorithm's calibration. To collect this data stream, heart failure patients will be equipped with three connected devices (watch, scale, blood pressure monitor) linked to the TakeCoeur AI device. The early detection algorithm for heart failure decompensation will be built based on variations in the physiological parameters collected and the occurrence of heart failure decompensation during the study period. The patient's clinical data will be collected by the cardiologist at inclusion, and the physiological parameters will be passively and actively collected through connected devices. A baseline for each patient will be established at inclusion, reflecting their initial status. Throughout the study, any deviation or variation from this baseline will be detected. In case of a heart failure decompensation, a link between this decompensation and the variations in digitally collected parameters will be established. The investigators hypothesize that the early detection of heart failure decompensation by the TakeCoeur AI device will align with the actual occurrence of heart failure decompensation, as recorded through healthcare utilization during this study, with good sensitivity and a low rate of false negatives.

Conditions

Eligibility

Locations (3)

Outcomes

Central Contacts