Like any doctor, Jacques Fellay wants to give his patients the best care. Hidden inside our bodies are genetic(基因的) markers that can tell him which patients could probably have diseases such as AIDS and provide early treatments. However, there are worries for Fellay: they contain sensitive details that could lead to embarrassment, discrimination or even worse.
A new kind of encryption, fully homomorphic encryption (FHE) (全同态加密), is making it possible for data users to run multiple operations on genetic data without seeing the contents. This can help end big data's privacy problem, and Fellay's patients can be some of the first to benefit.
In 1978, the concept of homomorphic encryption (HE) was firstly proposed. According to it, one could encrypt and share data with others, who could analyze and perform calculations on the data with no idea what it means. After getting the data back, the data user could simply arrive at the result using the secret key and it will make total sense.
In 2009, Craig Gentry firstly provided a workable FHE program. Like HE, it rests on a mathematical idea called a homomorphism, which mostly relies on using algebra(代数) to map data from one form to another without changing its underlying structure. However, it supports multiple operations on encrypted data, rather than only one calculation in HE.
Later, Gentry went on to work at IBM, which now has complicated FHE tools to run encrypted data. In the medical field, for example, it detected signs of COVID-19 infection by applying FHE to analyze the encrypted CT scans of more than 1, 500 people's lungs. In a paper published in October 2021, the team used FHE to collect data from multiple sources and predicted the effect of cancer treatments or the process of HIV infection, as in Fellay's case.
Workable, but still slow: calculations on the FHE-encrypted data could take millions of times longer than those on raw data. But, as Goldwasser says," If you believe that security is not a plus, but it's a must," she says, "then in some sense there is no overhead."
