Share data without sharing data with Multi-Party Computation
Sharing and analyzing data without revealing sensitive data and without violating privacy. Multi-Party Computation (MPC) makes that possible. A new collection of secure technologies with concrete applications for, among others, the healthcare and insurance sectors, developed as part of the Techruption program at the Brightlands Smart Services Campus in Heerlen.
Early December, during the Brightlands Techruption MPC Day on the Heerlen campus, researchers and the business community explored the possibilities of Multi-Party Computation. One good example is a planned pilot for an e-coaching app, implemented by the Zuyderland Medical Center, health insurer CZ, Statistics Netherlands and TNO. Accurate data analysis depends greatly on the wide availability of that data.
This is important, for example, in the healthcare sector, where physicians and researchers look for connections between people’s social backgrounds, education, lifestyle, and the region where they live. This is difficult because physicians, hospitals and health insurance companies are not allowed to freely share personal data of patients and clients due to privacy legislation. A new technology, Multi-Party Computation, offers possibilities. “MPC is a ‘toolbox’ of cryptographic technologies that allows multiple parties to calculate data together as if they had a shared database,” explains TNO’s Martine van de Gaar. “Because the data is encrypted and therefore unreadable, it can be shared without the parties ever being able to see other people’s data. MPC is a secure way to share data while maintaining privacy.”
At the Brightlands Smart Services Campus, various companies and researchers are working on this promising technology. “At TNO, we make the MPC technology practically applicable for companies and organizations,” continues Martine Van de Gaar. “For a long time, theoretical research into MPC technology has been performed by the academic world, and now the time is ripe for concrete applications. In recent years, we have noticed that the field has gained momentum, as have the protocols, so that they can be used in practice. The Brightlands campus in Heerlen houses many parties with the necessary knowledge and expertise in the field of data processing and analysis.”
Good news for the Zuyderland Medical Center and healthcare insurer CZ who are eager to find out exactly what the effectiveness of e-coaching is for patients. In collaboration with Statistics Netherlands (CBS) and under the supervision of TNO, they want to start a pilot next spring, in which they can use each other’s data for sharp analyses, without actually seeing this data. The pilot project uses the IDB Coach, an app for people with chronic intestinal infections. This app enables patients to carry out their own checks with their smartphone or tablet, communicate with their physician, and receive interactive tips on nutrition, lifestyle and medication. The patient will only actually visit the physician when necessary.
If the use of specialist hospital care decreases, the app is successful, right? “Well, not quite,” says Leon van de Weem, Project Manager at Zuyderland. “Suppose the patients then start consulting their family physician more often; this would mean the healthcare only shifted. We also do not know whether the coach is suitable for all patient groups. Can older people use it? Or people who cannot read well or who are computer-illiterate? To determine what the right care in the right place is, to really measure the effectiveness, we need more data, for instance from family physicians, local authorities and healthcare insurers. We would need data on education, social environment, etc. from Statistics Netherlands. We want to combine that data to get an exact picture of the pros and cons. At the moment, we cannot get that data, because it is confidential. With MPC, we can encrypt this data and still share the information, but only on an aggregated level. In other words, no one could trace the information back to an individual patient. At the moment, we are still legally testing whether we can use this data for this purpose. The use of different datasets is crucial in e-coaching. With current data from multiple sources, you can continuously update and adapt the app, bringing personalized care ever closer. This app can pave the way for multiple e-coaches.”
E-coaching is a promising development from a medical point of view, but which resources work best for which target groups, without stigmatizing? Financing parties, such as health insurers, are very interested in the actual effects. They want to know whether e-coaching can make healthcare more affordable, without compromising on quality. “That is correct,” says CZ’s Paul van Helvoort. “As a health insurance company, we are not only concerned with the reimbursement of bills; we also want to innovate, to keep healthcare affordable and accessible to everyone. We are convinced that e-coaching can be a good tool. However, it is important to find out exactly for which population groups an e-coach is useful. Does the use not lead to an extra burden for the family physician? If so, it would merely mean a shift, not a solution. For good research, you need data from hospitals, physicians and Statistics Netherlands, among others. Data that remains anonymous and yet is 100 percent reliable. MPC enables you to share that data without actually sharing. We have high expectations.”
This also applies to the Zuyderland Medical Center. “We invest in the app, in innovations,” says Leon van de Weem. “That should lead to better care, fewer patients and lower healthcare costs. That’s wonderful, but Zuyderland is also a company that pays for the innovation, only to see its sales decrease. If we can get proof that the demand for care is indeed declining, we’d also have good arguments for asking for compensation and for supporting further development.”
“A logical demand,” according to Paul van Helvoort. “Sure, from 2022 onwards, hospitals will no longer be allowed to grow in volume, which is why we need to innovate, because the demand for healthcare will only increase. As an insurer, we have an important role to play as a driver of innovation, which we also reimburse if it works. E-coaching is such an innovation. At the moment, people with chronic intestinal complaints visit the specialist once a month. With online coaching, this may be drastically reduced. This would be less stressful for the patient, and much cheaper. And that could apply to more e-coaching apps. That is why we think this pilot is very important.”
Statistics Netherlands (CBS) is also joining the project as a potential data supplier and as a party interested in applying MPC technology. “With this technology, we can make potential data available, enabling CZ and Zuyderland to refine their analyses further,” says Ted Stormen of Statistics Netherlands. “Physicians want to compare, use certain data from population groups, measure the effectiveness of treatments. They also want to learn more about people’s lifestyles and backgrounds. This data can never be traced back to people. We guarantee this by using a smart, innovative combination of MPC and blockchain technology. In addition, the data sets are encrypted throughout the entire process, complying with GDPR and other legislation. This is a great pilot. As Statistics Netherlands, we are constantly processing data and cooperating on initiatives with a social interest, such as in this case. MPC allows us, as well as the other parties, to do so without ever relinquishing our data.”