Accelerating discovery collaboratively with patients as partners: Q&A with Shannon McWeeney

Shannon McWeeney, Ph.D., works at the intersection of computer science, biostatistics, genetics and medicine, finding ways to speed cancer research past the bottlenecks stalling progress. She’s representing the OHSU Knight Cancer Institute in the Cascadia Data Discovery Initiative, or CDDI, a data-sharing project started by Microsoft and the Fred Hutchinson Cancer Research Center. On Tuesday, she was among the panelists on stage at the Geekwire Summit in Seattle, joining some of the biggest names in technology and biomedicine in the Pacific Northwest. McWeeney is the associate director for computational biomedicine in the OHSU Knight Cancer Institute. She is a professor of biostatistics and bioinformatics and head of the Division of Bioinformatics and Computational Biology in the OHSU School of Medicine. She talked with Cancer Translated about her career-long commitment to data sharing and patient engagement.

Cancer Translated: Why is there a need for data sharing efforts such as CDDI? Aren’t cancer researchers already sharing their findings with each other?

Shannon McWeeney: You know, in science we tend to think of just the researchers, and my world has never been about just the researchers. We have to keep in mind that patients are our partners. We are allowed to use their data and share it with permission from them. We owe it to them to maximize the knowledge gained from their efforts and their sacrifices. There are three cornerstones to this effort: the patients, the researchers, and the clinicians. If we just remove barriers for researchers and we exclude patients, we’ve failed. If we make data inaccessible to oncologists, we’ve failed. It’s all three. That means we have to rethink a lot of the current models.

CT: What are the biggest barriers preventing the kind of broad data sharing that you are talking about?

SM: We have a lot of barriers. There is the proprietary aspect, the notion that data is too valuable to share freely. Researchers will delay data sharing because they don’t want to get scooped by their competitors. Patients have legal rights to access their data, but too often we make it incredibly hard for them to get it. A patient can’t go from one cancer center to another and use the same form or enjoy the same ease of access. At some places you have to go in person, or you have to fax, it requires a signature. Other places you can use an online app and get your records. And even if we streamline that process, is the data they get back actually informative and useful? They may be asking for a summary because they want to go to another doctor, or they just want to understand for themselves, or they are interested in creating a personal health record that they can use for things like clinical trials matching. Or they want to share their data with a research study. Often the data that they get back is not meaningful to them and may not even be usable for research. This data should be easy for patients to obtain in a usable form. It’s their data. All of these barriers prevent us from doing what is really the most obvious thing: truly collaborating so that we can accelerate progress. We win when we are all talking with each other.

Related story: OHSU-led effort results in largest cancer dataset of its kind

CT: Can you give some examples?

SM: The Beat AML initiative with the Leukemia & Lymphoma Society is a trusted partnership where multiple academic institutions came together to pool resources, create a cohort and build something we could have never done on our own. CDDI with Microsoft is another example. We’re starting out with five trusted academic collaborators [Fred Hutch, BC Cancer, the University of British Columbia, the University of Washington eScience Institute and the OHSU Knight Cancer Institute] but we’re trying to come up with workable platforms for much broader sharing, such as systems for differential privacy.

CT: Can you explain how differential privacy works?

SM: You can think of it as a mathematical  guarantee that patient privacy is not violated. Say that I have a patient data set that you’d like to analyze. How do we handle that, especially if it comes from the electronic health record? With algorithms for differential privacy, an outside collaborator can still analyze the data but you are making sure a certain component of it is masked. You can get the answer back that you need, without actually impacting protected privacy. We have a responsibility to the patients who share their data to ensure that we are compliant with their wishes and that we keep the data safe. But most patients, especially in the case of rare diseases and cancer, want their data to be shared, they want it to help advance research. They don’t want their data locked up and never used.

CT: What kind of work are you and your OHSU colleagues doing with CDDI?

SM: We’re contributing a very realistic simulated patient data set for CDDI to use to practice testing these algorithms. We are having discussions among the five institutions around what it will take for each to trust the systems we are developing, what type of validation do we need to prove this is something we can actually implement. If we find commonality – here are set of  parameters that all five institutions agree on – I think the potential for that to scale is incredible.

CT: We started this conversation talking about including patients in these efforts. Why is that so important?

SM: We know that when researchers take the time to really engage with patients, they can achieve nearly 100% compliance with notoriously difficult things like getting research participants to send samples back from home. But patient engagement is something we also have to embrace because we owe it to them. Some will say, oh, we can’t give this data to them because its research data they won’t understand. That’s assuming a lot. I’m not saying we do anything stupid to allow patients to potentially misinterpret it and make harmful decisions. But they should be informed about how their data is being used. They should be informed of the discoveries we are making. They definitely should have the ability to negotiate about how their data is used so they can change their mind about things. The patient has to be part of these discussions. We can’t leave them out.