15 Aug. 2023. Data scientists and cancer specialists demonstrate a data encryption technique that makes possible real-world cancer data sharing while preserving individual privacy. A team from the company Duality Technologies Inc. in Hoboken, New Jersey with cancer specialists and data scientists from academic labs present their findings in the 7 Aug. issue of the Proceedings of the National Academy of Sciences.
The researchers are seeking a solution for sharing a wealth of real-world health data amassed in individuals’ electronic health and insurance claim records that can improve medical decision-making for individuals and communities. While clinical trials can answer questions of efficacy and safety for new therapies and medical devices under controlled conditions, real-world health data offer a large base of user experience to help validate clinical data and provide insights beyond the limited numbers of patients enrolled in trials. But, say the authors, real-world health data are often found in fragmented, incompatible databases where many patients are reluctant to share their records.
Duality Technologies is a seven year-old company providing data security solutions for artificial intelligence in a number of industries including health care. The company’s process is based on research by its founders in computer science and engineering labs at University of California in Berkeley, New Jersey Institute of Technology, and MIT that Duality Tech says makes possible the capture of data from a wide range of databases, particularly for training machine learning algorithms. The company says its technology ensures owners of the original data keep control of their assets, while still enabling the integration of previously separate databases and use of complex statistics.
Genomic data difficult to de-identify
One of Duality Tech’s prime applications is analysis of real-world medical data often generated from multiple sources. In the new PNAS paper, researchers in computer science, engineering, and oncology from UC-Berkeley, MIT, UC-San Diego, Tel Aviv Sourasky Medical Center in Israel, and Dana-Farber Cancer Institute, affiliated with Harvard Medical School, joined colleagues from Duality Tech to seek a solution for integrating real-world data from cancer patients while preserving the privacy of those records. The authors note that many current methods for de-identification of real-world medical records are labor intensive and difficult to scale, with genomic data particularly difficult to de-identify without losing important details.
The team applied a technique called fully homomorphic encryption or FHE that adds an encryption layer to already encrypted data, thus removing a need for a decryption key to the underlying data records. The authors enhanced FHE with a collaboration model where parties can temporarily contribute data to matched records for further analysis, and extended their privacy tools to statistics often employed in cancer research.
The researchers tested their techniques on real-world electronic health records of colorectal cancer patients at Sourasky Medical Center and data from two clinical trials of immunotherapy drugs among kidney cancer patients. The team found they could apply FHE with multi-party enhancements to process data with a number different statistical analytics from previously incompatible databases, without revealing the underlying raw records. And, say the authors, the techniques scale well to larger databases and can be applied to other collaborative health research tasks.
“Our joint study with Duality,” says Ravit Geva, head of clinical research in oncology at Sourasky Medical Center and the paper’s first author in a Duality Technologies statement released through Cision, “aimed and verified the accuracy of statistical oncology endpoints when done through encrypted data. The secure analysis yields accurate results compared with the currently used conventional data management and analysis methods on collaborative real-world oncological analyses without revealing patients’ protected health information.”
More from Science & Enterprise:
- A.I. Analytics Detect Heart Failure in Single Ultrasound View
- Automated Life Science A.I. Generation System Unveiled
- Primate Genome A.I. Model IDs Human Disease Variants
- Rare Disease Search Adds Large Language Model A.I.
- Real World Evidence Shown to Emulate Some Clinical Trials
We designed Science & Enterprise for busy readers including investors, researchers, entrepreneurs, and students. Except for a narrow cookies and privacy strip for first-time visitors, we have no pop-ups blocking the entire page, nor distracting animated GIF graphics. If you want to subscribe for daily email alerts, you can do that here, or find the link in the upper left-hand corner of the desktop page. The site is free, with no paywall. But, of course, donations are gratefully accepted.
[wpedon id=”42724″ align=”center”]
* * *
You must be logged in to post a comment.