Rescue inhaler usage prediction in smart asthma management systems using joint mixed effects logistic regression model

Asthma is a very common and chronic lung disease that impacts a large portion of population and all ethnic groups. Driven by developments in sensor and mobile communication technology, novel Smart Asthma Management (SAM) systems have been recently established. In SAM systems, patients can create a detailed temporal event log regarding their key health indicators through easy access to a website or their smartphone. Thus, this detailed event log can be obtained inexpensively and aggregated for a large number of patients to form a centralized database for SAM systems. Taking advantage of the data available in SAM systems, we propose an individualized prognostic model based on the unique rescue inhaler usage profile of each individual patient. The model jointly combines two statistical models into a unified prognostic framework. The application of the proposed model to SAM is illustrated in this article and the effectiveness of the method is shown by both a numerical study and a case study that uses real-world data.