Network science, the science examining the structure of relationships between social entities, is a complex but burgeoning field in healthcare. Based on research from pioneers like Nicholas Christakis, data from network studies might inform important questions like “What communities are likely to adhere to a particular medication?” and “How does one person quitting smoking affect others’ decisions to quit in his/her community?” Institutions ranging from big pharma to hospitals are using network science to learn about important trends relevant to their industry. However, it is difficult to interpret such information, and there is still a relative dearth of data out there.
Activate Networks, based out of Newton, MA, is one company attempting to translate network analytics into information that companies can use. We had a chance to see one of their presentations by Activate Networks Director of Analytics Luke Matthews at the Harvard Business School Health Care Conference last month. Activate Networks was co-founded by Dr. Christakis and others, and the company is one of the leaders in providing actionable insights from big network data. We spoke with Matthews this past week:
Ravi Parikh, Medgadget: Can you give us a bit of your background? How did you get involved in network analytics?
Luke Matthews, Activate Networks: I first used network analytics during my doctoral dissertation research on white-fronted capuchin monkeys in the Ecuadorian Amazon. I’m a PhD anthropologist by training, and at the time I was studying whether monkeys who were close together in their troop’s social network learned the same foraging behaviors. Social network theory and statistical analyses were a powerful way to test whether the monkeys learned foraging techniques from those they had more opportunities to observe, which in fact, they do.
Then I expanded my work on social networks to study how human social networks influenced the diffusion and inheritance of cultural information like Iranian tribal rug designs and religious beliefs of Christians. I became convinced by this work that the predictive abilities afforded by social network analysis also could have powerful applications in industry and government, and I was quite keen to be a part of developing such applications. So, when the opportunity with Activate Networks came about it was a great fit for my interests and skills.
Medgadget: How did you help early on with Activate Networks? How far has the company come since then?
Matthews: I started working at Activate Networks almost two years ago, about 8 months after the company was formally launched. When I started our only product was in the life sciences sector mainly in the context of drug launches. We’ve been growing rapidly since and each year we have moved into new sectors including the healthcare provider space and our Organizational Network Analysis division that drives results through organization social networks. We are continuing to expand our product offering in our established business verticals and are exploring new opportunities.
Medgadget: I have read about the work of Dr. Christakis in this area. How is Activate Networks leveraging that information for businesses?
Matthews: Dr. Christakis is a cofounder of Activate Networks and continues to be active as a board member and advisor. Activate Networks has an exclusive license to commercialize some of the scientific insights developed in the Christakis lab. Dr. Christakis’ published work provided the basis for Activate Networks’ initial technology. In particular, Dr. Christakis showed that many network characteristics cluster in a network out to 1, 2, and 3 degrees of separation. So, that means you have similar characteristics to your friends, but also to your friends’ friends and even your friends’ friends’ friends. We continue to use this insight to map network influence through the network more broadly than most other software that examine only an individual’s immediate social relationships.
Medgadget: Can you give us examples of specific populations for which businesses have used network analytics successfully?
Matthews: One specific case from our life sciences vertical is a client who was interested in ‘generic conversion’, or getting doctors to prescribe an equally suitable generic drug rather than the brand name version. Activate Networks applied its proprietary process to identify the most influential physicians in the market who, if converted, would have the greatest overall impact on use of generics throughout the physician to physician network. The client then applied our Influence Index™ in one city while applying their existing targeting process in a comparable nearby city in the same state. The existing process is largely the same across the pharmaceutical industry, and it involves sending detailing reps to the highest volume prescribers within a drug class. The two alternative lists (Influence Index™ vs. Volume) were used by the same drug rep to determine which doctors s/he visited. The rep also used the same detailing message in the two locations. So, the only difference was which prioritization scheme was used.
The results showed 50% greater generic conversion in the city that used ANI Influence Index™ as compared to the city that used a more standard volume-based decile approach. That’s a substantial difference that amounts to real revenue for the client. We have demonstrated similar levels of improvement in other verticals, such as improved efficiency from our Organizational Network Analysis and increased Facebook connect invitations for an online wellness community whose network we mapped.
Medgadget: Your company is organized into 3 divisions: Life Sciences, Health Plans and Hospitals, and Organizational Network Analytics. Why did you divide the company that way, and which division has been most active?
Matthews: We organized the company in this manner because each of these divisions have customers with very different use cases and business concerns. So, for the Life Sciences space we map physician to physician relationships with an eye toward leveraging peer to peer influence about pharmaceutical and medical device choices. Our product for Health Plans and Hospitals relies on similar networks of physicians, but in this case coordination of patient care, referrals, and place of service are all of primary importance. The Organizational Network Analytics division is based on the pioneering work of Rob Cross at the University of Virginia. In this division, we map the networks of employees at large corporations with an eye toward increasing workplace efficiency and collaboration. Historically this work has been conducted with surveys filled out by the employees, but increasingly we are doing this work from passively collected email log data. We don’t examine any of the text fields (subject line, email body, etc.) which protects employee privacy and the whole process saves everyone a lot of time as compared to doing another survey.
Medgadget: Who are some partners that you have successfully worked with so far to implement your analyses?
Matthews: We have worked with Healthways, the leading wellness company, on developing wellness programs that leverage the power of patient social networks, and with Navvis, its subsidiary, on developing products for hospitals that leverage the power of physician social networks.
We are working with the Premier healthcare alliance to help health systems better align physicians and coordinate care.
We have worked with numerous Fortune 500 companies through our Organizational Network Analysis division to help them address issues around innovation, collaboration, leadership development, and talent management by understanding how work really gets done in their organization.
Medgadget: What do you see as the potential of network analysis in the future of health care?
Matthews: Networks analysis is going to play a critical role as the healthcare industry transforms into one that involves more pay for performance, less pay for service, and hopefully more efficient pricing structures in general. Peer to peer networks of both physicians and patients have an incredible influence on factors that directly affect people’s health, such as how care is coordinated among physicians, hospitals, and payers as well as the drugs and devices that are actually deployed. The businesses that navigate the upcoming changes best are going to be those that understand the real world networks they operate in and are able to leverage that understanding into actionable insights. Activate Networks provides both pieces of this puzzle, robust understandings of the network and, perhaps even more critically, simple and actionable recommendations based on those understandings.
Medgadget: Is there any other information that would be helpful for us to know?
Matthews: I think one of the biggest insights from network science is that if we want to learn about people, we should look at them not necessarily as individuals or as members of formal groups only, but also as members of social network made up of people with strong relational connections to each other. Particularly if we want to understand how they learn new information or change behavior, or if we want to encourage them to adopt a program, we should look at them in their networks in social space, and not only at demographic information or formal group affiliations.
Website: Activate Networks…