Skip to Main Content
In the present study we show how an algorithm for generating heart failure alerts can be improved by retrospectively evaluating the available data. We built on a previous study on home based monitoring of heart failure patients after an episode of acute decompensations using mobile phones. Data from patients monitored in the years 2003 to 2008 were analyzed. For the analysis of historic treatment and measurement tables, GNU-R statistical software was used. A data processing algorithm was implemented to optimize the alarm-management of the heart failure home-monitoring system. The improvement was achieved by reducing the number of generated false alarms and by introducing a risk parameter, which could be used to indicate the patient's status. The algorithm provided adequate sensitivity and specificity and showed a significant improvement to the previous model, reducing the number of false alarms.