An automated system for monitoring respiratory diseases, such as asthma, provides noninvasive, multimodal monitoring of respiratory signs and symptoms that can include wheeze and cough. Some embodiments employ a mobile device, such as a cell phone, in which raw data from a microphone and an accelerometer are processed, analyzed, and stored. Data can be collected continuously. Time domain and frequency domain analyses of signals to determine, e.g., energy, duration, and spectral content of candidate sounds can be employed to discriminate symptoms of interest from background sounds and to establish significance. Accelerometer signals are analyzed to determine activity levels. Analyses of a user's symptoms and activity level prior to, during, and after an event can provide meaningful determinations of disease severity and predict future respiratory events. The system can provide a summary of data, as well as an alarm when symptom severity reaches a threshold.