OR WAIT null SECS
In Part 1 of this article, I described a collaboration between Brock Heinz from Spaulding Clinical and Joe Dustin from Medidata, which resulted in a very interesting proof-of-concept.
Editor's Note: This post originally appeared on Medidata's blog Geeks Talk Clinical and is re-printed with permission.
In Part 1 of this article, I described a collaboration between Brock Heinz from Spaulding Clinical and Joe Dustin from Medidata, which resulted in a very interesting proof-of-concept. Heinz created an app that serves as a bridge between Medidata Rave and popular quantified self devices like the Fitbit or Nike Fuelband.
So what might this proof-of-concept mean for the future of clinical trials?
Personally, I find the most exciting potential use case for this proof-of-concept to be in virtual clinical trials.
As Heinz pointed out during our discussion, Phase I trials are currently the most “quantified” of clinical trial phases. By pulling quantified self data into Medidata, we could bring this same level of quantification (or better) to other phases as well. And we could do it at far greater convenience for patients.
I see three overarching benefits to a quantified self-enabled virtual trial, which I’ll discuss in more depth below.
Each benefit described is an extension of, and enabled by, the previous one.
Perhaps the most straightforward and obvious benefit of a quantified self-enabled virtual trial is access to rich high-quality data.
Rather than being dependent upon site visits for data, we’d have access to a continual, or at least regular, feed of it. The collection process would require little, if any, human intervention, eliminating the possibility for data entry error. Furthermore, the data collected would be more “real,” since collection would occur in the patient’s natural surroundings, rather than an artificial site environment.
Perhaps most interestingly, we’d have access to new kinds of data.
This article has focused on quantified self data because that’s the focus of Heinz’s proof-of-concept. But this proof-of-concept could be expanded to serve as a bridge between Medidata Rave and other types of data as well.
The big advantage of application programming interfaces (APIs) like those provided by Medidata is the relative ease of doing exactly what I’ve described. For example, you could incorporate genomic or microbiome data into the patient’s virtual trial record. Incidentally, Genentech recently teamed up with 23andMe, which has an API similar to that of Medidata’s, to advance genomic testing in clinical trials.
In fact, I’d imagine you should incorporate multiple data sources into a virtual clinical trial, which brings me to the second benefit.
Why stop at EHR data? Why stop at quantified self data? The virtual clinical trial of the future could incorporate multiple disparate data sources into one patient record.
This virtual clinical trial would be more automated, collaborative, modular, and personalized. Given the need for pharma to shift from blockbuster to personalized drug development, the personalization aspect should be of particular interest.
Extensive discussion of such a clinical trial model is beyond the scope of this article. But to illustrate the possibilities, here’s one example of a powerful capability enabled by these improvements. Using a big data approach, the system could be programmed to perform real-time risk analysis, potentially alerting us to serious health issues before they occur.
Weight fluctuation, for instance, can signal a serious cardiac condition, as Heinz noted during our discussion. If patients regularly monitored their weight at home using a “connected” scale, that data could be continuously fed into and analyzed by a virtual trial system. Should that system detect weight fluctuation indicative of a cardiac condition, it could alert us to the potential issue.
This example is just one of many opportunities we have to create a more agile and efficient clinical trial model.
Last but certainly not least, the virtual trial of the future could provide a better patient experience in three major ways.
First is greater convenience. This model would reduce or eliminate the need for site visits. Furthermore, the seamless data collection enabled by such a trial would eliminate the more onerous tasks that we require of patients.
Second is quality. I don’t think it’s a stretch to suggest that a more automated, collaborative, modular, and personalized clinical trial has the potential to provide higher quality for patients.
Third is connection. As discussed at the beginning of this article, many patients view clinical trials as being disconnected from their broader health (and human) interests. Once we have the ability to connect the pieces of a formerly disparate patient puzzle, we can better align clinical trials with these interests.
Our goal should be to align clinical trials and patient interests so well that the distinction between each becomes almost invisible. With the convergence of clinical research and clinical practice, the patient can truly be at the center.
This final benefit also creates a desirable byproduct.
If we provide a superior clinical trial experience for patients, that stubborn patient recruitment bottleneck will certainly open up.
The primary purpose of this article was to describe Heinz’s proof-of-concept and what it might suggest for clinical trials of the future. I’ve taken an admittedly ambitious and optimistic bias in this description, largely because the opposite bias can easily be found elsewhere.
But by no means am I ignorant to the difficulties of implementing the clinical trial model I’ve described. Substantial technological, regulatory, institutional, cultural, and operational barriers exist.
However, the difficulty of overcoming these barriers shouldn’t prevent us from trying.
We can begin by experimenting with different elements of our existing clinical trial model. For example, quantified self devices aren’t currently advanced enough to collect all of the data needed for clinical trials. However, activity trackers like the Fitbit, which work with Heinz’s proof-of-concept, still have potential value.
For one, activity data could be used to verify a common protocol requirement. Though patients are often asked not to change their exercise routine during a clinical trial, we have no objective measure of compliance. Activity trackers could change that.
Alternatively, activity data could provide a useful endpoint for some trials. For studies related to pain or mental health, an increase in activity might be suggestive of meaningful quality of life improvement. Interestingly, Mayo Clinic recently published a study indicating that Fitbit data could be used to predict surgery recovery time.
Experimentation with our existing clinical trial model won’t be without challenges or failures. But in my experience, very little worth doing is.
What I’ve described is just one person’s thoughts on a future clinical trial model, and I have no doubt these thoughts will change over time. So I’d like to hear your vision. What might the clinical trial of the future be like?