An ambient assisted living (AAL) system consists of heterogeneous sensors and devices which generate huge amounts of patient-specific unstructured raw data every day. An important feature of remote monitoring applications is to identify the abnormal conditions of a patient accurately and so send appropriate alerts to the care givers. In traditional systems, situations are classified by generalized medical rules or fuzzy rules which are not always applicable for every kind of patient. These systems cannot sense the future at an early stage. In some monitoring systems, when a patient feels unwell he/she needs to press a wearable panic button to notify a response centre about the emergency. In this work, we have presented BDCaM, a generalized framework for personalized healthcare, which leverages the advantages of context-aware computing, remote-monitoring, cloud computing, machine learning and big data. Our solution provides a systematic approach to support the fast-growing communities of people with chronic illness who live alone and require assisted care. The system can accurately distinguish emergencies from normal conditions. The data used to validate the model are obtained via artificial data generation based on data derived from real patients, preserving the correlation of a patient’s vital signs with different activities and symptoms.