This document summarizes research on detecting emergency events using mobile phone data from Cote d'Ivoire. A Markov modulated Poisson process model is used to detect anomalous hourly activity patterns from time series data of calls between 970 cell towers over 7 weeks. Preliminary results show the model detected 8 of 19 known emergency events and 8 of 11 non-emergency events. Future work could analyze how events propagate by studying mobility and activity patterns during dissemination. The research aims to help governments and organizations respond faster to security incidents by predicting emergencies from mobile phone usage data.