It’s not a matter of whether AI will transform health care, but when.
We all recognize that artificial intelligence (AI), and the ability to use predictive analytics to make sense of massive combinations of data, is altering the retail, banking and mobility industries. Today, organizations such as Amazon and Google are able to rapidly mine customer-generated information for insights about individual behaviors and preferences, creating bespoke buying experiences.
AI is changing health care, too, reshaping the business models of life sciences and health care companies as access to data is increasingly democratized. At the BIO International Convention on 20 June 2017, it was my privilege to moderate a wide-ranging discussion about how biopharma and health care companies could use artificial intelligence to solve critical key health delivery challenges and the opportunities and challenges in transforming the biopharma value chain. I was joined by four of today’s top thinkers in the space:
- Dr. Atul Butte, Director of the Institute for Computational Health Science at the University of California — San Francisco
- Jackie Hunter, CEO of Benevolent AI
- Iya Khalil, Chief Commercial Officer and Co-founder of GNS Healthcare
- Nathan Price, Associate Director of the Institute for Systems Biology and co-founder of Arivale
Reflecting on our panel discussion, I am even more convinced than before that life sciences companies must build AI capabilities sooner rather than later. It’s not a matter of whether AI will transform health care, but when. If traditional life sciences players don’t use AI to ask better questions that can ultimately lead to better products and services, nontraditional players will eventually overtake them.
Here are key learnings from our discussion:
1. AI will alter every facet of the biopharma value chain — but especially early-stage research
As we noted in Beyond borders 2017: Staying the course, there’s growing evidence that AI can help identify safer, more efficacious drug targets more quickly and for less money. Jackie Hunter, CEO of Benevolent AI reminded the audience that close to 50% of drugs still fail in Phase II and Phase III drug development due to lack of efficacy or safety signals.
“We tap into so little of the existing evidence,” she said. Using AI, Hunter estimates a biopharma could identify better compounds four times faster, with the potential to reduce the late stage drug failure rate by as much as 20%, a result she calls “transformational” for the industry — and patients.
2. AI will level the playing field
By democratizing data, AI enables biopharma Davids to compete with incumbent Goliaths. A small team enabled by machine learning is just as capable as a big pharma of asking medically relevant questions to develop breakthrough therapies or collect real world evidence demonstrating product value. As Atul Butte noted, “New opportunities for smaller players mean the next Genentech might come from a garage.”
3. New partnerships will be critical
Given the volume and velocity of data generation, no biopharma has access to all the data of interest — or the analytic tools or scientists that can make sense of it. The companies that will ultimately be most successful are the ones that create vertically integrated, diverse teams of clinicians, biologists and data scientists. Data scientists, especially individuals with biomedical training, will continue to be a prized commodity in the coming years. So will access to rich, diverse data sets that combine structured and unstructured data from laboratory tests, electronic medical records, wearables and other consumer-generated activities over extended time spans.
“We need to work with multiple different entities to collect vast amounts of data to elucidate the black box that is medicine,” noted GNS Healthcare’s Iya Khalil.
4. AI won’t replace the expert. It will augment the expert
There was consensus that AI will change how medicine is practiced, resulting in more data-driven decisions that lead to better patient outcomes. Still, don’t expect an algorithm to prescribe your care any time soon. AI isn’t about humans versus machines, ala the sci-fi thriller, Ex Machina. It’s about humans plus machines.
Indeed, AI will enable the experts, the physicians, to do what they already do — treat patients with compassion — but to do it better, sifting through myriad treatment options to optimally select a regimen that is not only personalized to the individual patient but driven by the totality of existing evidence.
Ultimately, how quickly AI transforms life sciences and health care business models will depend on how easy it is for company researchers to combine and analyze data, biopharma’s emerging fuel, in new ways. Interoperability between data sets will be a critical accelerator, especially convenient platforms that encourage patient consumers to want to share their data.
Indeed, as Nathan Price of the Institute for Systems Biology noted, we need to set up systems that create “immediate value” to individuals so that they want to share their data early and often to create dense, data-rich information clouds that can be mined in different ways. Only then will we see the creation of a virtuous circle of data sharing that generates downstream returns, shifting health care from reactive, costly and generalized to proactive, more cost-effective and individualized.