| Popis: |
Aims Non-invasive remote patient monitoring is an increasingly popular technique to detect early clinical deterioration in heart failure patients. Previous studies have shown mixed results on the sensitivity and specificity of such systems. Therefore, we developed a novel individualized biometric monitoring algorithm, and evaluated it by comparison with simple rule-of-thumb algorithms and a moving average convergence-divergence algorithm. Methods and Results Transmitted daily bodyweight, heartrate, and systolic blood pressure were collected in 78 patients for a period of 279±128 days, during which 31 patients experienced 64 events of clinical deterioration of heart failure. The algorithm was informed by both cross sectional and patient-specific longitudinal information, predicting a dynamic patient-specific biometric value at each time point. Large deviations from the predicted measurement triggered an alarm. Sensitivity and area under the curve were higher (57% and 63% respectively) for the personalized monitoring approach when monitoring bodyweight only, in comparison to the rule-of-thumb (42% and 52%) and moving average convergence-divergence algorithms (12% and 52%). Similarly, when combining bodyweight with heart rate and systolic blood pressure, the personalized monitoring approach outperformed (70% and 64%) rule-of-thumb (47% and 55%) in terms of sensitivity and area under the curve. Source code for the algorithm is made publicly available. Conclusion The proposed personalized algorithm based on combining bodyweight, heart rate, and systolic blood pressure measurements is a promising tool for predicting early deterioration of heart failure. Heart rate monitoring performed better compared to bodyweight. Before clinical application, the algorithm should be tested in randomized controlled trials. This repository contains code and datageneratedfor the manuscript: "A personalized remote patient monitoring system based on daily measurements of body weight, heart rate, and blood pressure to early detect deterioration in heart failure patients". It comprehends: (1) PRECISION-HF Rcode file with an example,(2) MACD Rcode file with an example. User can download "bank.csv" and "non-bank.csv" files as an examplary input data sets. |
