Privacy preservation of medical data with improved association rule mining based data sanitization with optimal key generation for internet of medical things

Privacy preservation in Internet of Medical Things (IoMT) is critical for safeguarding sensitive health data. Traditional Association Rule Mining (ARM)-based models anonymize data but face scalability challenges when handling large IoMT datasets due to computational complexity. This research introduces an Improved Association Rule Mining (I-ARM) based privacy preservation model leveraging the Pufferfish Secretary Bird Optimization Algorithm (PSBOA) to address these challenges. The proposed framework involves a structured process of data sanitization and restoration. Initially, association rules are extracted using a Modified Apriori algorithm in the I-ARM process to identify sensitive patterns, followed by the generation of an optimal key via the PSBOA algorithm that integrates pufferfish optimization and the secretary bird optimization algorithm for secure data protection. The optimal key generation is conducted based on the constraints like modification degree, false rule generation and hiding ratio. Sensitive data is then sanitized using an XOR operation with the optimal key and securely stored in the cloud. The restoration process reverses the XOR operation to recover the original sensitive data when needed. Finally, a reverse process of the Modified Apriori algorithm is employed to reconstruct the original medical data from the sanitized dataset. Extensive experimental evaluations demonstrate that the proposed model effectively preserves privacy and ensures reliable data restoration. The experimental findings demonstrate that the suggested method achieved a higher privacy value of 0.034, greater restoration values of 0.970, and the lowest False Rule Generation value of 0.483, which is better than existing methods. Therefore, the hybrid PSBOA approach enhances key generation, offering robust security and resilience against unauthorized access, highlighting its potential for practical deployment.