High-frequency Data in AWS

The wearable devices in our trial generated over four gigabytes of data per hour from each participant. In a traditional clinical trial, a site investigator might collect patient data once a week. We spend a lot of time thinking about how to help life sciences companies with these types of trials and the data they generate, but we could never build the infrastructure to hold that amount of information on our own.

If all patients within Medidata’s clinical trials were equipped with monitors that streamed continuous data, we would receive patient information equivalent in volume to the data generated by the Large Hadron Collider. As a company that prides itself on being nimble and being able to move where technology is taking clinical trials, there is no alternative to the cloud.

There has been a nagging belief in various industries that high frequency data isn’t a great fit for the cloud. But this is absolutely incorrect. We wouldn’t have been able to make the advances that we made unless we had cloud architecture, and the same applies for life sciences companies that embrace technological advances to address the problems of clinical trials.

Cloud architecture is in fact the perfect place for high frequency data in the life sciences arena. As an industry we’re going to know a lot more tomorrow than we do today. We made a conscious decision to hold on to data because of the potential to learn new insights from regression testing in the future. Even though a proof-of-concept may be shut down at the moment, algorithms get smarter and smarter. We want to be able to take those algorithms and run them against the old data. It won’t be long before mHealth-generated digital surrogate endpoints will be used to demonstrate clinical efficacy just like traditional clinical endpoints used in drug development.

Stepping further into questions about this data residing in the cloud, data privacy concerns can be addressed simply by the reality of the environment. Through a combination of global scalability and an infrastructure of fully certified and compliant service providers, data privacy concerns are often overblown.

As a company, our entire existence is premised on the fact that we must protect data. We protect data for more than 500 clients, while a pharma company is responsible for protecting the data of just itself. We think we can do it better than a single client because this is all we do.

Manage Insight, Not Data

CIOs are not paid to run data centers; they are paid to generate results for their companies. We need to pivot to more outcome-focused leadership. This will be a slower process if we spend more time thinking about where our data sits than how we extract new insight from our data.

By operating in the cloud AWS, our data science team can be data scientists, not infrastructure managers. They aren’t figuring out how to deal with the volume of data, they are figuring out the best algorithms to generate real insight from the ever-growing flow of digital health data.