While most "Apple fans" were eagerly anticipating the Apple Watch announcement (up to $22,000 for a WATCH? Really?), many in the medical and research world were particularly interested about Apple's vision to create a mobile medical research platform -- "ResearchKit."
Last year, Apple introduced HealthKit, a platform that allows health and fitness app developers to share data with each other and the Apple "Health" app. It still causes quite a stir as the validity of the data is questionable, and many clinicians are unclear about how to interpret this data when patients ask.
Over the last year, more and more physicians and researchers have become involved in mobile health -- or simply "mHealth" technologies -- seeing possible benefits for their patients. The fact that by this year over 90 per cent of individuals will own a smartphone, means that apps could lead to real, measurable health impacts.
In turn, big players in Silicon Valley -- Apple and Google for instance -- realize they need to create sustainable partnerships with the medical community should they wish to make a significant step into the health arena. These partnerships are already being seen -- just three weeks ago, the Dean of Harvard Med School wrote about his experience at Google and the benefits of mHealth.
The announcement of ResearchKit earlier this week began with Apple executives explaining limitations that many medical researchers encounter. These are what the start-up world calls "pain points," and includes recruitment difficulties (which limits sample size), issues with subjective data collection, issues related to frequency of data collection, and communications flow between the subjects and researchers.
For ResearchKit, Apple partnered with several leading hospitals and Universities -- Stanford, Massachusetts General Hospital (affiliated with Harvard), University of Pennsylvania, University of Rochester, Mount Sinai (NYC), and the University of Oxford (UK). Each University is affiliated with a particular app -- for instance the Parkinson's Disease app, "mPower" was developed at the University of Rochester. The other four introductory ResearchKit apps described yesterday involve the study of heart disease, asthma, diabetes and breast cancer.
Apple isn't the first company to tackle the current limitations in clinical research. Other devices have tried to tackle some of these issues through "crowdsourcing" data. However, Apple is positioned to be the first company to do this well, primarily due to sheer market reach, easy of use, and now with clear strategic partnerships with academia. As a trained epidemiologist that loves clinical research, I see several exciting possibilities with ResearchKit, but have several concerns as well.
The idea of having more people engaged in research is exciting, particularly the possibility reaching more people than a traditional study. This might allow for a more diverse sample population, which could feasibly allow results to be more generalizable. Enrolling large numbers of people increases sample size, which in statistics, reduces the chance of "Type 2 error" -- something researchers strive for when drawing conclusions for their study.
I can also see the benefits of ResearchKit for rare diseases (once these apps are rolled out) -- some diseases are so rare that they are difficult to study on a large scale, and it can take a very long time (i.e. years) and hundreds of thousands of dollars to enroll enough people. Through mobile technology, it may be possible to involve individuals around the world to study any given rare disease.
By announcing that ResearchKit will be an "open source and open platform," Apple is following the noble lead of various leading medical journals (e.g. Lancet, PLoS) to allow free access in order to advance research for various diseases. This truly globalizes research, and could potentially lead to collaboration between developers and institutions across the globe, which is pretty incredible.
ResearchKit offers an intersection between the medical research community (which typically progresses at a slow pace) and the technology space (which, using Apple as an example, moves rapidly) which could lead to entirely novel ways of conducting research, new methodologies, changes in some antiquated regulations (which could be beneficial in terms of bringing new treatments to the market), and innovation in clinical research in general.
Clearly, ResearchKit is an exciting step forward for clinical research. That said, there are seven (7) things that are somewhat unsettling, and perhaps some of these hesitations might be clarified in the coming months.
1. Not all Research is Created Equal: It's important to recognize that ResearchKit facilitates "observational/survey" research for now. There are benefits to this type of research, but the level of evidence created is different from, for instance, a randomized controlled trial (RCT). That said, depending on the data shared, some users could be chosen to participate in an RCT at a nearby centre, and reaching them through the App would make it easier.
2. Role of HealthKit: ResearchKit can pull data from the user's HealthKit apps -- this relationship between both platforms needs to be clarified to protect the consumer. For instance, a consumer may use HealthKit to track their physical activity and nutrition, but may only want a specific portion of the data shared. It's unclear how ResearchKit is able to distinguish between various preferences.
3. Informed Consent: ResearchKit is currently an "opt-in" service -- so users have to specify if they want to share their data through a given ResearchKit app. There are obvious concerns around informed consent -- a key element of clinical research. How can ResearchKit ensure that the participant is fully informed of the risks and benefits of participation, and that their consent is valid?
4. Screening and Diagnostics: Traditional research studies clarify the role of the study in screening and diagnostics. Little is known about how each ResearchKit app can function as a "screen" in itself. For instance, mPower uses subtle detections in "tapping" of the screen to detect tremor -- a symptom of Parkinsons Disease(PD). Yet, how might it distinguish between other causes of tremor (e.g. anxiety, thyroid issues, caffeine) and how impactful could a false suggestion (or progression) of PD be on the user of this app? Moreover, whose responsibility is it to reassure the participant or provide next steps for diagnosis?
5.Ethics: Related to the point above, there might be some foreseeable ethical implications of ResearchKit. The lines between the duty of the researcher (through ResearchKit) vs. a clinician's duty (according to the Hippocratic Oath) are blurred. Further, when does the "consumer" (of an iPhone or an App) become a "research subject" and/or "patient" and what are the implications of these shifting roles? In traditional studies, these roles are carefully delineated, but the relative anonymity on both sides for ResearchKit makes things less clear.
6.Generalizability: Though sample size could be improved through ResearchKit, merely by having access to a larger pool of consumers, there is likely to be some form of selection bias that will always be limiting. For instance, it is unlikely that 100 per cent of consumers with a given disease will participate and enroll in a study through an App. What distinguishes the ones that do from the ones that don't, how to we capture characteristics of those that "opt out" and how does this impact how we can apply results to the general population.
Apple's announcements about ResearchKit were certainly exciting, and filled with potential for the future of clinical research. However, a cautiously optimistic view might be the most prudent at this time. There are many questions, and even while researching this article, it is clear that participating institutions (e.g. Stanford) might be best positioned to provide clarification. Hopefully clinical researchers will have more information over the coming weeks so that ResearchKit can make the advances it so earnestly hopes to make.
MORE ON HUFFPOST: