Privacy-Preserving Deep Learning Models for Analysis of Patient Data in Cloud Environment

A substantial amount of patient data is being generated every second by the healthcare sector. The medical data especially patient data could be used for analysis through advanced deep learning models, but the private nature of patient data limits the use. Massive volumes of diverse data must be collected, which is often only possible through multi-institutional collaborations. One way to create large central repositories is through multi-institutional studies. This method is limited to privacy issues, intellectual property, data identification, standards, and data storage when data sharing is done. As a result of these challenges, cloud data storage has become increasingly viable. The various models for exchanging medical records on the cloud while protecting privacy are discussed in this chapter. Furthermore, vertical partitioning of medical datasets that exploits attribute categories in health records is explained and analyzed in order to examine distinct areas of medical data with varying privacy issues. These methods can ease the strain on communication costs while minimizing the need to communicate sensitive patient information.