Last week I attended a Health 2.0 London event on mobile sensors. The title of the event asked if they were key to remote patient monitoring. I think the answer is an obvious yes; without them it’s not going to be possible. But I’m not sure the event was asking the right question. For me, the question is how we derive actionable knowledge from all the data that we might amass.
There were two accelerometer-based devices on display, Fitbug and Withings; one service that lets you pool all your accelerometer-based activity onto one website, Tictrac; and one mobile heart monitor, AliveCor.
AliveCor immediately impressed me. It provides an accurate ECG trace just by holding a special iPhone case (see here). In the UK it costs £169 and is apparently as accurate as hospital-based machines that cost thousands of dollars. It was easy to see not only how they might be used in clinics but also how patients could be “prescribed” them so that they can record what’s happening in their hearts when they have those “funny turns”. I loved it.
The accelerometer-based devices convince me less. I actually use three accelerometer-based devices every day: Nike’s Fuelband, Fitbit’s One, and the Moves app. I have developed a strange relationship with them whereby on a daily basis I wonder why I am bothering because they don’t seem to be telling me anything interesting and yet at the same time I’m unwilling to stop using them when I see all the fancy charts they’ve drawn of my movements. I seem to be convincing myself that they’ll mean something at some point somehow.
Quite how, I am not sure. It seems to me that the quantified self movement, as it’s called, is busy having nerdgasms about measuring stuff but hasn’t started talking about how it might analyse it to derive something meaningful and compelling enough to change people’s behaviour.
Sooner or later we’re going to have to throw a little academic discipline at how we measure, record and analyse all the observational data collected through mobile sensors. The challenge for academia is that it’s going to have to deal with dirty data – or, more accurately, real world data. What I fear is that academia will revert to type and try to clean the environment. Although I appreciate the need to remove glaring confounders, I think that rather than make the data fit for academic study, academic study needs to become fit for the real world of health-related behaviour.
Later in the week I was at the second NHS Entrepreneurs Day at which we heard about the forthcoming Academic Health Science Networks (AHSNs). There are going to be 15 of them across England and their role seems to be to aid the adoption of new services locally, while helping frontline staff to share their ideas. One of the people who presented at the Entrepreneurs Day was Maneesh Juneja, the leader of Health 2.0 London. While he told the audience about the community he leads I couldn’t help feeling that there is a significant disconnect – if not an overt fissure – between his world and the forthcoming AHSNs.
The Health 2.0 community is experimenting with what’s possible; they’re playing with innovations, usually new technologies, and seeing what value they can create. The community appears to eclectic, with representation from academia, pharma, the NHS, start-ups, designers, doctors, nurses, and students. The only thing they seem to have in common is their willingness to experiment – or learn about experiments that others are doing.
By contrast the AHSNs came across as gatekeepers. “We’re your access to the NHS”, they seemed to be saying, “it’s through us that you’ll be able to trial your products”. Although their tone was welcoming it felt protectionist.
Obviously we cannot do wild experiments on patients. A degree of protection is needed. But at the same time we have to find ways for new ideas and approaches to see the real world in which we can evaluate their effectiveness. The tension between experimentation and protection needs to be understood and slowly resolved.
The one thing that binds these two groups, though, is the need for reliable forms of analysis. The Health 2.0 community is going to have to embrace the analytic methodologies needed to derive reliable, actionable knowledge from all the data they’re collecting. The AHSNs also need these methodologies to assess which products to try on patients.
These days we take for granted that randomised controlled trials should be conducted and assessed in line with the CONSORT statement. But this statement was first drafted 20 years ago and has been constantly revised ever since. CONSORT was started by 30 experts meeting, discussing and agreeing what was needed to assess a trial, a seminal moment in defining how we might trust their findings. Something similar needs to happen with observational data.
Will this happen? Perhaps. In the US, the Robert Wood Johnson Foundation is supporting the Health Data Exploration Project, which is “exploring how to bring individuals and companies who collect…data together with health researchers to uncover insights into personal and population health”. Also, the United Nations has launched Global Pulse, which is “exploring how new, digital data sources and real-time analytics technologies can help policymakers understand human well being and emerging vulnerabilities in real-time”. In England, I understand the Clinical Practice Research Database is considering how it might incorporate data from fitness trackers.
I find all this hugely encouraging. I only hope that the experimentalists and protectionists realise they have the same goal and create something that serves both communities.
Competing interests: I am a member of the Health 2.0 London community and have met Maneesh Juneja but have no commercial ties with him. I paid for my Nike Fuelband but got the Fitbit for free while at TEDMED 2013. I downloaded the Moves app from app store and recall that it was free.
This post was first published on my original blog, Optimising Clinical Knowledge.