After breathlessly ascending the peak of inflated expectations, and careening down through the trough of despair, digital health seems poised to re-emerge, battered but not beaten, in the form of digital medicine: digital health plus evidence plus humility.
That was my key take-away from a fascinating and constructive meeting, the Harvard Digital Medicine Symposium, I was privileged to participate in earlier this week at Harvard Business School, organized by Business School Professor Ariel Stern, Brigham & Women’s physician Will Gordon, and Elektra Labs CEO Andy Coravos.
As Coravos discussed on her recent Tech Tonics podcast (here), and explains (along with co-authors) in more detail in a fantastic primer, recently published here, digital medicine:
“describes a field, concerned with the use of technologies as tools for measurement, and intervention in the service of human health…. As a discipline, digital medicine encapsulates both broad professional expertise and responsibilities concerning the use of these digital tools. Digital medicine focuses on evidence generation to support the use of these technologies.”
In his welcoming remarks, Gordon observed that in our daily lives, digital has arrived; he described, tongue in cheek, the “smartphone sign” utilized by some doctors when deciding if a patient is healthy enough to be discharged from the hospital: if they’re using their phone, they’re healthy enough to leave. (There’s even a publication on this – here.) The challenge addressed by the symposium was how to bring the digital and data fluency we see in the rest of our lives, along with integration so deep we don’t even think about it, to healthcare – not as precious innovation projects but as a foundational enabling technology that advances the goals of patients, physicians and other caregivers, and developers of therapeutics.
Rather than offer a comprehensive recap of the conference, I thought I might instead touch on several key themes that stuck with me (and for which I seem to have vaguely legible notes – apologies in advance for the worthy content not adequately captured).
Context, Ethics, and Evidence
In a thoughtful talk of considerable breadth, Eric Perakslis of Duke University (his Tech Tonics episode here) pointed out that most of the data about patients isn’t in their health records, but rather, it’s in the digital exhaust of their daily lives – the patterns around where they live, eat, and travel, and shop.
The importance of context was also emphasized by neurosurgeon Jonathan Pomeraniec, who referenced Harvey Cushing’s famous quote, “A physician is obligated to consider more than a diseased organ, more than even the whole man – he must view the man in his world.” (As an aside, a Cushing quote is required in all neurosurgery talks, much as a Fuller Albright quote is necessary in endocrinology lectures, and a Francis Peabody [if in Boston] or William Osler [everywhere else] quote is needed in any discussion of the practice of medicine.)
Perakslis also emphasized the importance of balancing benefit and risk from the use of digital technologies, asking “how do we keep the positive energy [associated with digital] from leading to harm?” He especially emphasized the need for aspiring health tech entrepreneurs in Silicon Valley and elsewhere to familiarize themselves with “the culture of bioethics and evidence” in which physicians and scientists are steeped, a fluency he seemed to regard as table stakes for engagement with healthcare stakeholders. Not only is this the right thing to do, he argued, but it’s in the best interest of the entrepreneurs, who need to realize, Perakslis said, “you can’t get to monetization without evidence.”
Another key theme was that adoption of digital technology requires relevance (to avoid the pervasive “your solution is not my problem” situation).
Dr. Bill Wood, of the University of North Carolina, is a cancer researcher who is interested in characterizing the performance status of patients. He emphasized that the key for digitalization is “relevance to clinical problems,” and highlighted the need to “construct an evaluative framework” for digital technologies that includes measurement verification, consideration of cybersecurity and data governance, consideration of usability from both patient and clinical trial research site perspective, and validation within the intended context of use.
Lara Mangravite, President of the non-profit Sage Bionetworks(disclosure/reminder: I was a founding advisor a decade ago), emphasized the importance of relevance to patient-participants. She pointed out that one of Sage’s key learnings has been that digital studies have astonishingly steep retention curves – i.e. many participants are interested initially, but then there’s a staggering drop-out rate. Initially, Mangravite says, the Sage team thought this was an anomaly related to the unusual publicity their first study received, when it was promoted by Apple when ResearchKit was unveiled, but then they recognized it was a prominent feature of subsequent studies as well. “Participating in these research studies is a daily decision,” Mangravite observed, adding that if you want to increase retention, then participants must get value out of it as well – ideally information that impacts their daily lives.
Not surprisingly, relevance is also a high priority for regulators; Francesca Cerreta of the European Medicines Agency (Europe’s FDA), for instance, emphasized that “data must be linked to meaningful clinical benefit.”
Need For Change
Several participants spoke passionately of the need for change, highlighting ways in which the existing system is far from ideal, and how new approaches could help.
Noted Harvard cardiologist Calum MacRae, for example (his recent Tech Tonics here) offered a critical “travesty of a mockery of a sham” perspective of contemporary medicine that would do Fielding Mellish proud. MacRae noted medicine is “personal, local and non-scalable by design,” and suggested, when thinking about the design of new studies incorporating emerging technology, that “existing medicine performs so poorly it’s almost unethical to compare yourself to it.”
Dr. Brad Hirsch, CEO of clinical trials startup SignalPath (his Tech Tonics here) emphasized the “need for clinical studies to get away from PDFs, Excel, and post-it notes,” and argued researchers must instead adopt a “digital schematic for trials.”
There’s an urgent to leverage the expertise of a new set of experts, argued Dr. Ken Gersing of the National Center for Advancing Translational Science (NCATS). “The A team [in the digital and data space] are not M.D.’s,” Gersing said. “They’re at Verily. There at Facebook.” He added, “With the new [digital and data] approaches, fishing is actually a science – it used to be a derogatory term.”
There be Dragons
New technologies come with risks that were are just beginning to recognize and acknowledge, many speakers argued.
Liza Hoffman, a clinical social worker and clinical director of the digital psychiatry division at Harvard’s Beth Israel Deaconess Medical Center, discussed the challenge presented by consumer-oriented health apps; her research found that patients tend to select apps based on cost (preferably less than a dollar, and ideally free) and reviews; she described her efforts coaching patients to consider additional qualities such as credibility of source and data protection afforded.
There were several talks focused on cybersecurity (tl;dr it’s something that health tech companies ought to worry a lot more about) and on the need to remain cognizant of evolving legal framework for these technologies (tl;dr it’s complicated and likely to remain so). (I would also recommend several pieces on these topics by Perakslis, including this on cybersecurity, and this on ethics, with a focus on settings characterized by humanitarian crises.)
Capturing a prevalent theme of the day, the FDA’s Bakul Patel emphasized that while he often hears the FDA praised for being “pro-innovation,” this is explicitly not the Agency’s goal. “We don’t want to drive innovation, we want to drive responsible innovation,” he said.
Many speakers offered talks that seemed deeply rooted in pragmatism, versus more aspirational (some might say magical, some might say visionary) thinking.
Forbes colleague Greg Licholai (also a lecturer at Yale School of Business, the co-director of the Yale Center for Digital Health, and the CMIO of the contract research organization PRA) offered a “recipe” for digital medicine: trust, reliability, and quality, and emphasized both the promises of digital medicine (access, convenience, efficacy, personalization, reduce stigmatization, and evidence) as well as what will be required to deliver this (safety standards, quality standards, interdisciplinary collaboration, education, business model and pricing, responsibility). He particularly emphasized the importance of quality, noting “Regulatory is sexy, quality is not,” a quote he attributed to David Amor of Pear Therapeutics. (The sentiment seemed to catch everyone by surprise, none more so than the regulators in attendance, who seemed distinctly unprepared for their Right Said Fred moment.)
The importance of a grounded understanding of the underlying technology was emphasized by several speakers, including Duke’s Jessilyn Dunn, a biomedical engineer who emphasized the need to think about what’s going on under the hood of wearables, and said “methods are needed to tune both hardware and software.”
In a similar vein, Ariel Dowling, a Stanford-trained mechanical engineer (who, disclosure, I met for the first time at this conference and learned she had recently joined Takeda), aimed to demystify machine learning (ML) for the audience, explaining ML is basically using stats to identify patterns in data, and creating a model to make predictions. “ML synthesizes a complex set of data into a singular outcome,” she said.
A key pragmatic challenge, faced by many innovators developing new technologies, is how to fit these novel approaches into existing workflows. Jeff Abraham, VP of Market Access at Akili, presented a captivating overview of the challenges is company has faced integrating digital therapeutics into an infrastructure built around pills. He suggested that what the current system seem to want is basically “to put the algorithm into a pill bottle and ship it” – which I thought was a particularly compelling and accurate image. Abraham said he’s been guided by the mantra, “disrupt, don’t interrupt.”
Yet another challenge is what might be called the last mile problem – or perhaps more accurately, the first mile problem: the challenge of connecting technology with real patients. For example, in his Tech Tonicspodcast, MacRae described the pervasive and underappreciated challenge so many patients face in pairing digital blood pressure monitors with their smartphones. Similarly, at this week’s symposium, physician and safety expert David Bates shared data from a (small) study suggesting that only 43% of patients studied were able, on their own, to figure out how to enter basic information, like blood glucose, into a smartphone app.
Still Not Clear That Tech “Gets It”
The day ended with a fireside chat I conducted with former FDA Commissioner Robert Califf, a cardiologist who now splits his time between Duke and Verily. Califf offered a characteristically grounded perspective, saying he thought a lot of the health tech products are unlikely to be adopted because they’re not actually very useful, especially to those who might need them the most. He emphasized his distain for the term “provider,” which he thought debased the unique role of clinicians, and he highlighted the need to consider patients outside the trendy urban areas where a lot of tech seems to originate, citing rural communities as particularly underserved.
More generally, my sense is that he’s come away from his tech immersion as underwhelmed and perhaps somewhat disappointed; he is clearly appreciative of the power and potential of tech, but seemed to feel – echoing a sentiment Perakslis voiced earlier – that while the engineers drawn to this space may grok technology, most still struggle to deeply, empathetically understand both the actual problems that need solving and the people in the trenches – the patients, physicians and other caregivers, as well as the therapeutics developers — who the technologies are ostensibly designed to help.
A New Hope
While this core problem seems in many ways unchanged from what I originally described in 2011 (here), I (naturally) see several reasons for optimism. First, the champions of change – the leaders represented at this symposium – have a grounded view of what’s required, including an appreciation that this will be a heavy lift, as well as the recognition that evidence, not merely enthusiasm, will be required for progress.
Second, the “digital transformation” wave that’s washed over the rest of business is increasingly forcing the question, “what about healthcare?” Yes, healthcare is different in some ways, but at another level – c’mon, there must be something we do to raise our game. At the very least, senior executives seeking to drive digital transformation in their healthcare organizations are forcing managers to become increasingly creative in parrying these efforts, and to look more diligently to find at least a few digital wins to which they can point.
Finally, and most encouragingly, I see a new generation of researchers who grew up immersed in data and digital, beginning to populate incumbent healthcare organizations and pharmaceutical companies, bringing with them a new set of expectations and capabilities – folks such as Dowling, MD/PhD student (and my former intern) Erik Reinertsen, Apple physician-scientist Andrew Trister, and Google physician-scientist Alvin Rajkomar (magnificently profiled in the NYT by Katie Hafner) spring immediately to mind. With time, effort, and bit of luck, I suspect they will be the change we all wish to see.