Search for a command to run...
This study aims to explore the possibility of using fully homomorphic encryption (FHE) in conjunction with machine learning for private pathological evaluation. Specifically, we will look at how support vector machine (SVM) inference phases might be used to classify sensitive medical data. To make SVM inference on encrypted datasets easier to implement, a system is presented that makes use of the Cheon-Kim-Kim-Song (CKKS) FHE protocol. This architecture eliminates the need to decode data before analysis and guarantees the confidentiality of patient records. Further, a method for efficiently extracting features from medical images for use in vector representations is introduced. The system's performance and usefulness are supported by its examination on different datasets. Classification accuracy and performance are comparable to that of conventional, non-encrypted SVM inference; however, the suggested technique protects the CKKS scheme from known cryptographic assaults with a 128-bit security level. In a matter of seconds, the secure inference procedure is carried out. Cardiology, oncology, and medical imaging are just a few of the areas that could gain from FHE's improved security and efficiency in bioinformatics analyses, according to these results. The significant future implications of this research for privacy-preserving machine learning can pave the way for improvements in diagnostic processes, individualized medical treatments, and clinical investigations.