This paper addresses a major issue with patient's homecare systems which aim at improving the quality of care and the assistance services. To ensure effectiveness and efficiency, these systems have to provide physiological and environmental data acquired by sensors in a continual basis and transferred them to central middleware to be processed. This process has for consequence to generate a large amount of data in real time which need to be handled immediately in order to detect alarm situations. These data can also involve normal situations. We are interesting in this problematic and are developing an application which distinguish between normal and abnormal cases in order to take into account critical conditions in real time, filter irrelevant data and store only the ones needed for use in the next processes. This approach will be implemented and performed locally at patient's home. It takes care of sending only useful data to central middleware and detecting alarm situations. In this work, we take into account the patient's context and the home environment by proposing a sensors-based context model. The data processing is based on using alarm and data storage rules.