The development in the field of miniature sensors and wireless communication enables the successful deployment of Internet of Things (IoT) in the healthcare sector. The healthcare data generated by the medical sensors is very sensitive and enough to qualify as an instance of big data. In this paper, a secure framework is designed to assist the diabetic patients. The sensors’ data of each patient is used to generate context aware correlation rules by using map-reduce apriori algorithm. From these rules another labeled dataset is created to build a classifier for predicting the present state of the patient. After the successful deployment of this classifier on the service provider side, secure communication has been provided between the patient and the service provider using Key-Policy Attribute Based Encryption. Hence, providing an IoT based secure ambient assisted living system for diabetic patient may be helpful to the healthcare sector.