17 Jul Interview with Eric Perakslis of Datavant
Eric is currently Chief Science Officer at Datavant, an innovative new company with the mission to organize the worlds healthcare data. Eric is also Visiting Scientist in the Department of Biomedical Informatics at Harvard Medical School and a frequent volunteer advisor to Médecins Sans Frontières and other international aid and relief NGOs. Most recently Eric was Senior Vice President and Head of the Takeda R&D Data Science Institute where he built an integrated institute of more than 165 multi-disciplinary data scientists serving all aspects of biopharmaceutical R&D and digital health. Read his full bio.
Interview with Eric Perakslis of Datavant
Q: What need is Datavant addressing?
A: While the volume of health data grows precipitously year over year, our ability to make use of that data is not keeping pace. There are many aspects to the inherent challenges and opportunities of this data explosion but of particular concern is the increased fragmentation. Each new clinical study, consortium and/or large federal grant inevitably creates a tailored knowledge base that, in effect, is the next data silo and the fragmentation compounds. We have a paper out this month in Nature Drug Discovery that calls for a much ‘bigger’ data approach to the massive unmet medical need of Alzheimers’ disease (AD) https://www.ncbi.nlm.nih.gov/pubmed/29977050. In this paper we suggest that, to make progress in all the unknown aspects of AD, we must start by linking all existing AD data sets/sources and only add prospectively to this master linked source. It is not only doable, it is necessary.
A key challenge to this type of progress are the frictions that arise from the competing ideals, competing institutions, technology incompatibilities, the need for global ethics and IRB oversight etc. We can do ‘bigger data’ much better but still must protect patient privacy and to leverage past patient data to provide better care to future patients. We address this problem through offering a suite of software solutions that reduce the friction for sharing and de-siloing data across the industry all while protecting patient privacy.
Q: What are the products and/or services Datavant offers/develops to address this need? What makes Datavant unique?
A: I think it is our ambitions that make Datavant unique. We see a completely unmet set of opportunities in healthcare data that are essentially untouched. One way to visualize the opportunity are as the macro and micro silos of healthcare data. The micro silo problem is reasonably well understood. When people choose to work together and share data via consortia, grant or other data club, there are plenty of ways to do it. Good examples are Project DataSphere (https://projectdatasphere.org/projectdatasphere/html/home) or the extensive use of tranSMART (http://transmartfoundation.org/) by IMI consortia, both efforts we strongly endorse. But what about the macro silos?
The progress on Apple Healthkit is amazing, but my 70-year old mom cannot afford a thousand-dollar iPhone. Where is the Android version? That is a macro silo. More specifically to Datavant, there is the massive gap between drug development and medical practice. Those data exist in different regulatory regimes, in different types of institutions and are fire-walled against competition and re-use. What if every patient visit generated the equivalent of a case report form that could be submitted to the FDA. There has been great progress in real world data and evidence, but the macro silo remains untouched. Lastly, healthcare data is growing rapidly but is still a small percent of data that exists about each of us. There is great energy and promise in social determinants of health. How can the other 99% of data about us that exists outside of our medical records be used to improve and ensure our health?
Today, the core of Datavant’s offering is a on-premise software solution for de-identifying datasets in a manner that preserves the ability to link it to other data at the patient level. This enables our customers to stitch together de-identified datasets that come from a range of sources without compromising patient privacy. Datavant is also rolling out a new service which creates an ‘index’ of anonymous patients from participating data sources, offering a rapid means of identifying overlap (and therefore partnership opportunities) across multiple datasets.
Today, our services target the micro silo problem. As they evolve, our offerings will tackle the macro silo challenge.
Q: What is your role at Datavant and what excites you about your work?
A: As CSO, my role is to think deeply about the problems in healthcare and to help envision solutions that will enable and deliver transformational success for our partners in their data strategies. If I get that right, it automatically ensures innovative products and solutions for Datavant.
Q: When thinking about Datavant and the dom
A: We’re experiencing a moment where a large number of healthcare institutions and healthcare product companies are realizing that they are actually data companies at their core. Datavant offers a path to exploit these untapped aspects of their practices, businesses and research institutions. It is early days but, so far, our greatest accomplishments are the new data linking and privacy preserving capabilities we have brought to these organizations.
For me personally, it is the new and novel research studies and partnerships we have launched that, once complete, will demonstrate the value of thinking about and using data differently. Today, these are proprietary to our partners but they will be revealed in clinical study designs and new publications soon.
Q: What are the short-term challenges that Datavant and its peers are facing?
A: Traditional thinking is always a barrier but also well worth the effort to overcome. This has been a core challenge to my change agent role in every position I have held. I actually love it. Well-intentioned but complex new regulation, such as GDPR has introduced a great deal of uncertainty for any party working with European datasets, though we expect that the path forward on this will become progressively clear in the future.
Q: Is there anything else you would like to share with the PMWC audience?
A: Only that, if any of you have a really complex healthcare data challenge that appears to be a barrier that is preventing transformational progress or change, I would love to talk about it.