The present disclosure relates to a method for monitoring behaviour of a patient in real-time. The method comprises receiving, by a patient monitoring device, data related to the patient from one or more sources. Thereafter, the patient monitoring device classifies the received data into one or more categories based on one or more rules. Further, the patient monitoring device correlates the categorized data to identify one or more activity patterns corresponding to the patient, wherein each of the one or more activity patterns are associated with an activity performed by the patient at predefined time intervals. The patient monitoring devices compares the activity pattern with predefined activity patterns and detects abnormal behaviour of the patient if the identified activity pattern is different from one or more predefined activity patterns corresponding to the patient.