Latest News

Guiding Principles For Academic-Industry Partnerships

By Deborah Borfitz

September 3, 2019 | Dwindling government funds for research and development (R&D) has universities doing everything possible to maximize returns on what taxpayer-supported grants bring in. Ironically, they often ignore the deeper pockets of industry that now provide 67% of R&D funding—three times more than federal agencies, according to Michael Becich, M.D., Ph.D., associate vice-chancellor for informatics in the health sciences, and chairman and distinguished university professor, biomedical informatics, at the University of Pittsburgh School of Medicine.

In his keynote address at the recent Next-Generation Dx Summit, Becich went over some of the rules of engagement for academic-industry partnerships and lessons for digital pathology collaborations. His personal goal, he says, is to put Pitt on the "top 10 list" for such partnerships.

Attracting licensing revenue from pharmaceutical companies and device manufacturers can be lucrative for academic medical centers, he says, pointing to well-known examples such at the University of Florida (Gatorade), University of Wisconsin (Warfarin) and Stanford (Google). It also adds "real-ness" to the work of innovation-minded faculty, potentially aiding in their recruitment and retention. Since 1980, he notes, universities have been getting all rights to intellectual property (IP) generated from federal grants.

To forge partnerships with industry to commercialize IP and develop products requires "an ecosystem to teach academics how industry works," says Becich, co-director of Pitt's Center for Commercial Applications of Healthcare Data (CCA) that specializes in translating university innovations into healthcare-critical products. Becich is also a serial entrepreneur who most recently co-founded SpIntellx, a computational and systems pathology company providing "explainable" artificial intelligence tools for pathologists and biopharma companies.

The Pittsburgh Health Data Alliance (PHDA)—a collaboration combining the University of Pittsburgh's medical research expertise with Carnegie Mellon University's computer science and machine learning capabilities and UPMC's deep data and track record of commercialization—is helping to shepherd ideas originating in academia to clinical settings, Becich says.

Universities receiving the most industry funding, including Northwestern, Columbia and New York University, are getting it thanks largely to "one-drug wonders" that are patent-protected for 10 to 20 years, says Becich. His success metrics for Pitt are instead based on startups, licensing and IP development.

Becich shared eight guiding principles for a successful academic-industry partnership: clearly define goals and scope, develop a data-sharing plan, set a timeline, clarify each participant's responsibilities, develop a budget, establish IP ownership, clarify manuscript authorship and determine if geographic considerations are important. Many partnerships are short-lived based on shortcomings in one or more of those arenas, he says.

Pulling from the playbook of Donald Taylor, his counterpart at the CCA, Becich says it's important for partners to know their strengths and differences and align their incentives. "Define the appropriate solution scope to meet academic and industry goals, identify necessary resources and engage project management professionals from both sides."

The workflow between partners can be complex, he continues, citing whole-slide imaging as an example. Conflict of interest management, which is required when more than $10,000 is swapping hands, is a "veritable minefield." The university and health system may have conflicting rules; disclosures are required with every talk, published paper and press release; and policies extend to family members.

At Pitt, Becich says, faculty can spend up to 20% of their time on outside consulting time and accept payment but it must be disclosed. The policy defines activities that do and don't constitute "consulting" and forbid faculty from running a study on patients from their practice, accepting industry sponsorship of research in their lab, having their own graduate students work in their company and serving as director of a company from which they are receiving consulting payments.

A new model is under development by the PHDA and CCA, Becich adds, including templates and predefined terms. Meanwhile, the university is in the process of building a Pitt Data Commons bringing together many of the computational and information entities across Pitt to serve teaching, research, and library needs.

In early August, Amazon Web Services also provided PHDA scientists with financial support for eight product commercialization projects involving machine learning. As announced on its website, these include initiatives that aim to create an individual risk score for cancer patients, enable doctors to better predict the course of a person's disease and treatment response, use a patient's verbal and visual cues to diagnose and treat mental health symptoms, and reduce medical diagnostic errors by mining all the data in a patient's medical record.

"Keep an eye on us," says Becich. "We have a model that's working."