Service Level Agreement Based Secured Data Analytics Framework for Healthcare Systems

Many physical objects are connected to the internet in this modern day to make things easier to work based on the convenience of the user, which reduces human involvement with the help of Internet of Things (IoT) technology.This aids in the capture of large amounts of data, the interchange of information via the internet, and the remote operation of machines. IoT health data is typically in the form of big data and is frequently coupled with the cloud for secure storage. Cloud technology provides a wide range of technological services via the internet, and it is a highly interoperable and on-demand network for a wide range of computing resources. The Service Level Agreement (SLA) is made between the cloud and the patient, and it outlines the services supplied as well as the level of security provided to the user. For fulfilling service, the deployed external cloud has challenges with load balancing and work scheduling. Furthermore, the gathered health data must be effectively processed by medical practitioners. To solve this issue, a Secure Cluster Naive Bayes (CNB) framework is proposed, both with and without Dimensionality Reduction. To preserve its anonymity, the obtained data is hashed and stored in the cloud using the blockchain technology. SLA sessions are organized to prioritize patient data for decryption and prediction. At the doctor's end, the decrypted data is first filtered and dimensionally reduced before being clustered using a dual K-means clustering technique and classified using the Naive Bayes algorithm. The web-based graphical user interface server is responsible for connecting the IoT device, the cloud, and the doctor. The security performance of the DRCNB and CNB frameworks is evaluated using block chain characteristics, the frequency of SLA violations, and processing and execution time. The DRCNB framework is 91.1% accurate, while the CNB model is 80.73% accurate, making it more accurate than previous models. The new models exceed the prior ones in terms of both security and prediction performance.