Privacy-Preserving Text Summarization Using Semantic Similarity With Biobert And Clinicalbert For Multiple Medical Documents Leveraging Parallelized High-Performance Computing

Abstract

The enormous volume of textual data produced by medical documents in the healthcare industry provides insightful information, but it also presents serious privacy, data security, and computational complexity issues. Through the use of parallelized high-performance computing (HPC), this research presents a unique framework for the privacy-preserving text summarization of various medical records utilizing semantic similarity algorithms driven by modified BioBERT and ClinicalBERT. In order to maximize productivity, the framework uses distributed computing environments and secure computation approaches to satisfy the demand for summarizing sensitive medical data while maintaining anonymity. This study shows that the method offers quick and privacy-compliant summarization, protecting patient privacy without sacrificing the information's relevance and semantic accuracy.