Enhancing Secure and Private Data Analysis with PENTHER
In healthcare data analysis, using artificial intelligence (AI) for diagnosis and disease prediction is crucial. However, AI needs a lot of data, and in places like Singapore, where data diversity matters, collaborating with global partners is key to accessing the necessary data for advanced data-driven methods. Due to the sensitive nature of such health data, and regulations like the Personal Data Privacy Act, strong protections must be put in place for data collaborations to happen.
Tackling Data Challenges with PENTHER
Cloud data sharing is popular due to its flexibility and scalability however, there remains strong security and privacy concerns regarding the use of confidential or sensitive data in the cloud. The PENTHER system is a technology that enables secure and privacy-preserving analysis of data in the cloud. Data remains encrypted in any computation, thereby allowing it to be securely analysed without being exposed in the cloud environment.
Key Parts of PENTHER
PENTHER is made up of two components, the Analyser and Encryptor. The Analyser processes data in encrypted form and works with Encryptor(s) to deliver the final results to the analysis requestor.
The Encryptor is crucial to the protection of all data throughout the outsourcing or collaboration process. It generates a set of public, evaluation and secret keys. The public key is for the Encryptor(s) to encrypt data, evaluation key for an Analyzer to operate on said encrypted data and the secret key is retained by the Encryptor to process encrypted results for release to the analysis requestor.
In collaboration settings, data contributors will each have their own Encryptor component and secret key to process encrypted results. Without the participation of all data contributors through their respective Encryptors, the encrypted results cannot be released to the requestor and deleting their secret key would render all encrypted data and results unusable to anyone.
Unlocking New Insights through Collaboration
All data remains encrypted throughout the analysis process and thus preserves their security and privacy while in external environments. Furthermore, in collaboration settings, data owners and custodians full retain control over the data they contribute to the collaboration through the secret keys they hold.
PENTHER can perform encrypted statistical analysis on datasets with up to ten thousand records in less than a minute. Its components were successfully applied as part of Singapore’s first real-world privacy-enhanced analysis of real-world health data.
Future Directions
With PENTHER, more collaborations in the healthcare sector can be established to generate data-driven insights from the data they possess. Besides that, PENTHER can enable a global data sharing network for healthcare researchers to pool more to answer important questions on the health and wellness of people.
While the current iteration of PENTHER supports statistical analysis, there is a wide range of tools used by the data science and artificial intelligence community such as regression analysis, machine and deep learning applications that we are working to enhance PENTHER to support for encrypted data. Improving the performance of the technology to scale up to even larger datasets is also important for wider adoption in the industry.
Besides that, we also aim to expand the capabilities to enhance collaborations with data among the public sector and public-private partnerships with PENTHER.
PENTHER is part of the Memorandum of Understanding signed by A*STAR’s Institute for Infocomm Research (I²R) and Bioinformatics Institute (BII) with Singapore Translational Cancer Consortium (STCC), University of Nottingham (UoN) and Nottingham University Hospital NHS Trust (NUH).