In our everyday lives, we are exposed to IoT devices that sense the environment and produce raw events. Especially, in smart home scenarios such raw events can be used to detect and monitor human activities and behaviors. This could be of a great help in context that involve fragile people such as elderly or patients, to monitor their condition and support their daily life activities. With our contribution we aim at defining an approach that can be applied on smart-home IoT event logs and support the discovery and monitoring of human routines. In this work we present our approach that relies on the application of community detection algorithms for the discovery of the routines and the adoption of process mining techniques for their inspection. Especially, we report on a first implementation and validation of the approach with respect to a well known IoT event log.