Democratizing Data in Canada
By Dana Barberio
June 27, 2016 | Much like the U.S. based Precision Medicine Initiative, the University of British Columbia’s Personalized Medicine Initiative (UBC’s PMI) is engaging a coalition of stakeholders to harness the power of personalized medicine to improve health outcomes. While the U.S. effort is larger in scope, both are recruiting large numbers of citizens to contribute personal ‘omics and other health data, as well as developing curated databases of public data, to enhance individual outcomes and to advance research and development. With this wealth of patient information entrusted to them, both initiatives are deeply dedicated to privacy. UBC’s PMI has a goal of securing a political commitment from the Canadian government for leadership of the initiative, to set priorities and provide financial resources across healthcare, innovation and education sectors.
Ten companies have already spun out of UBC’s PMI, including Molecular You (MYCo). This digital health company has a vision to decipher a wide variety of individual patient data to promote health, detect disease in its earliest stages, and provide insights on therapeutic efficacy and timing.
MYCo engages physicians to acquire a comprehensive array of patient data: Fitbit readings, dietary information, bloodwork, heart rate, sleep patterns, exome, proteome, microbiome, metabolome, epigenome and other medical records. They then integrate this information with their curated databases to glean a greater understanding of the individual’s risk for various diseases. This provides powerful insights into avoiding disease. If a person is already ill, the company suggests the best therapeutic options, and then profiles and evaluates post-treatment results, including any treatment changes.
Making Data Accessible to Patients and Physicians
There is a wealth of information available in health data, but right now that information is just not easily digested. Buried in the noise is the potential for tremendous insight into improved patient health outcomes, new innovations, and lowered costs.
Big data technologies are transforming the healthcare arena: lowering the cost of gathering and storing data for improved patient care, drug development, and implementing longitudinal disease studies. Yet another tier of benefits will come when researchers, physicians, and patients themselves can access the data in a meaningful way.
So how does this wealth of data become user-friendly? “A key part of what we’re doing at Molecular You is translating information into a usable format for physicians and patients” says Rob Fraser, COO and co-founder of PMI and President, CEO and co-founder of MYCo.
In addition to the deluge of data acquired from patients, MYCo is harnessing the wealth of public information available. “One of the challenges we face in the healthcare arena is that our body of knowledge is doubling about every 1.3 years,” and it is increasing, says Fraser. “To keep up with this information, we need to be curating much more frequently.” MYCo is starting to build algorithms that are automated, using machine learning techniques on curated data to help them tackle this incredible task.
The Power of Hadoop with Privacy Security
To manage these data, MYCo turned to PHEMI, a big data warehouse company based in Vancouver for help. PHEMI brings a background in supercomputing, telecommunications/secure data networking, and healthcare management. MYCo chose PHEMI’s Central Big Data Warehouse, a cloud-based software platform that combines data management, data mining, accessibility, and privacy protection.
For the PMI/MYCo endeavor, PHEMI Central aggregates the data across multiple data sources, manages time series data, and hunts for anomalies in individual patients and in patient populations. PHEMI uses a Hortonworks Hadoop software stack, which is based on the powerhouse technology that makes the Google and Yahoo search engines possible. PHEMI Central uses Hadoop to analyze longitudinal studies, generate cluster analyses to look for patterns in the data, and for natural language processing to identify information in pathology reports, for example. At PHEMI they’ve taken that technology, and managed the privacy security and governance of data to control access to the information.
“One of the big barriers to adoption of big data technology is that it requires a certain level of expertise,” says Adam Lorant, VP Product and Solutions, PHEMI. “There are a lot of moving parts with Hadoop, and a lot of specialized expertise required in the inner workings and administration of the Hadoop software stack, and consequently it’s very difficult for healthcare providers and researchers to build that expertise in house.”
PHEMI makes this a user-friendly experience for researchers exploring complex interactions. For example, one client is using PHEMI Central to do cluster analysis of microbiome, genotypic, and phenotypic data to uncover patterns that reveal where patients sit on the Autism Spectrum, Lorant says.
“At PHEMI, when we pull data in, we make sure it’s in a consistent, cataloged, indexed situation,” says Lorant. Imagine the tremendous trouble you would have in finding information in an unindexed library, and in particular if that information was contained in multiple books on multiple shelves in different parts of the library. “That’s what we do with PHEMI Central, link together information from diverse sources, find it fast and control access to that data on a need-to-know basis without compromising privacy. The other part of it is the ability to audit the data, who has seen it and what they’ve done with the data,” This is unique to PHEMI, he says.
In order to provide clients flexibility and control over data sharing, consent, and privacy, PHEMI Central provides a “Zero Trust Model” approach. This information security model, which was developed in 2009 by Forrester, has gained widespread adoption as a multi-layered security approach. The concept involves trusting nobody, verifying everybody at the networking layer. “PHEMI is the only vendor, to my knowledge, that has built Zero Trust into the data repository itself. If you can get this Zero Trust system into the data warehouse, then you have solved the data privacy versus data sharing problem that exists,” says Lorant.
Using PHEMI Central, MYCo is doing studies of healthy individuals as well as the super-healthy—the Canadian military’s Special Forces and athletes-in-training, to dive into the fine details of what hinders and enhances their performance. MYCo is looking at diseases that are not yet completely understood, such as polycystic ovarian syndrome.
“We’ll work directly with the physicians and patients to monitor them over a period of a year, as treatments change, so that we can begin to understand causative factors in their disease,” says Fraser. They are examining a large diabetes cohort as well as patients with rheumatoid arthritis and atrial fibrillation—conditions where typically symptoms are treated rather than the disease. “We see this as an opportunity to get at the source of the disease, and then monitor those changes,” says Fraser. A 20,000 person longitudinal study is in the works.
“This is an exciting time that we’re in,” says Lorant. “We’ve never been in the situation before where we can get this kind of information around an individual affordably. This is going to change the way we do most everything.”