Eli Lilly revolutionizes clinical trials with intelligent sensor cloud

Digital biomarkers are increasingly playing an important role in improving our understanding of disease and health. Defined as quantifiable and objective behavioral and physiological data collected and measured by digital devices such as implantables, wearables, ingestibles, or portables, digital biomarkers enable pharmaceutical companies to conduct studies remotely without the need for a physical site. This innovative approach is revolutionizing the way pharmaceutical firms conduct research and determine treatment effectiveness.

According to a recent industry report from Research & Markets, the global market for digital biomarkers is set for significant growth at a compound annual growth rate (CAGR) of 36% during the forecast period 2022-2028. But dealing with the data produced by digital biomarkers, let alone acting on it, remains challenging. To address this issue, US pharmaceutical company Eli Lilly decided to turn to the cloud.

“Digital biomarkers offer unique insights into patient health through the continuous and passive collection of data using wearable sensors and remote technology. However, to make the most of these, we needed a sensor cloud to aggregate large volumes of data, perform real-time monitoring of the data, and analyze results in new ways to explore potential innovations. The solution also needed to be compatible with multiple different devices, depending on what was being measured,” says Rich Carter, SVP for the digital office at Eli Lilly and Co.

Lilly’s IT team explored the marketplace for a scalable, near-term solution that aligned with the pharmaceutical’s needs. But it found no high-capacity digital data cloud on the market that went beyond ingesting and storing wearable sensor data to help deliver the necessary insights — “making a case for Lilly to flex its innovation talent and build its own proprietary solution,” Carter says.

“We built a fit-for-purpose ecosystem called MagnolAI, which is a sensor cloud with full-stack capabilities to continuously ingest, visualize, and transform a large amount of real-time wearable sensor signals from Lilly’s connected trials into meaningful digital measures,” says Carter.

The resulting platform, created over the past three years by an agile, cross-functional team bringing together a range of expertise, has earned Eli Lilly a 2023 US CIO 100 Award for innovation and IT leadership.

Turning data into intelligence

MagnolAI ingests raw and processed data from all connected devices leveraged in clinical studies — whether those are off-the-shelf wearable devices to measure heart rate, or a Lilly innovation such as its sensor used to measure defecation for inflammatory bowel disease (IBD). The platform then makes this connected data accessible to Lilly’s data and analytics experts, who in turn create algorithms to better understand the disease journey, help measure the effect of Lilly medicines, and build new products that support successful patient outcomes.

“The team took a device-agnostic approach when designing and implementing MagnolAI’s data capabilities, making it a powerful tool regardless of the device being used. MagnolAI has enough scalability to visualize data from different devices, profile them and generate reports of data quality, including the ability to aggregate and synthesize data from across clinical trials,” Carter says.

But what sets the sensor cloud apart is that, while most solutions focus on data collection, MagnolAI is engineered to turn data into intelligence, he says.

“It empowers users with a unique analysis-driven abilities: to view data at scales and resolutions that fit for analysis purposes, capture exact data points with unprecedented accuracy, and to deliver digital data at full spectrum from cloud to analysis environment anywhere and anytime,” he says. “This level of detail and flexibility, previously unseen in the sector, positions MagnolAI as a game-changer for professionals who demand more from their sensor cloud platforms and their data assets.”

The solution, which is private to Lilly, was built with an important human-in-the-loop design principle, Carter adds, “to allow our researchers to view the data, develop initial hypotheses of data, create algorithms to quantify and verify the hypotheses through iterations and learning cycles.”

Overcoming data challenges

MagnolAI has been used to support around 20 connected trials to date. As its use grows, Carter’s team will need to develop new systems, tools, and pipelines to enable the collection and analysis of new forms and sources of data — no easy task given the volumes of data involved.  

“The ability to capture a tremendous amount of data is exciting, but early on, it was challenging to make sense of this amount of data, especially as we look across different trials,” says Carter. “In some cases, we’re collecting more than 4 million data points in a single day from one patient.”

To navigate this challenge, endpoints from the raw data sets were aggregated and operated on to better understand patterns and associations. “These phases needed to be performed iteratively as researchers developed and validated their hypotheses through multiple rounds. We found that the key to discovery is creating a mechanism that allows researchers to explore digital signals and gain insights through iterations,” he says.

Then there were challenges related to full-spectrum data quality and assurance as defining quality expectations and monitoring for conformance can be difficult in a free-living data environment. “We had to scan the invalid values and noises from wearable sensor signals. Another challenge was aggregating compliance information from signal level to the number of analyzable digital measures from the visit and study levels as it can be very monotonous. Reporting on data quality effectively across the data life cycle is essential and requires concrete expectations and ways of working,” says Carter.

The power of cloud for personalized health

While still in its pilot phase (MagnolAI will be rolled out fully in the next 7-12 months), the project has helped the company in many ways.

Claimed to be the most comprehensive visual computing solution for big data, MagnolAI enables interactive exploration and navigation of large volumes of data at scale. Its real-time data monitoring tool has enabled Eli Lilly to promptly track big data quality and compliance in the connected clinical trials, in-clinic or at-home across the entire patient journey.

One example of how MagnolAI has already been used is studying daytime sleepiness in patients with Parkinson’s. “Until now, the most common way to evaluate success is through patient-reported outcomes, which can be highly subjective. The motivation of the study was to enable the passive collection of patients’ daytime sleepiness behavior through digital devices. From there, teams developed and deployed the algorithm from that data to derive novel digital endpoints to quantify averages for daytime sleepiness. Even though Lilly decided not to move forward with its Parkinson’s drug, the team was able to continue to use this information and data for ongoing studies and different disease states, like obstructed sleep apnea,” says Carter.

“This novel work has allowed various study teams to implement connected devices in trials, as well as the development of digital measures, including supporting and generating digital endpoints in five Lilly’s pain studies, and the development of an algorithm to assess nocturnal scratch with actigraphy in atopic dermatitis patients, winning 2022 Top 100 LRL Innovators for team members,” he says.

The project piloted an exemplary case of Lilly’s homegrown IDS (investigational drug service) technology, “saving nearly $3 million per year from depending on external data platform vendors,” Carter says, adding that the team’s work with MagnolAI has been presented at the IEEE BigData conference, with four manuscripts under review by prestigious journals and one patent filed.

 
The platform also presents the opportunity to partner with organizations outside of Lilly to bring digitally enabled solutions to those who need them, Carter adds. “Lilly is seeking partnerships to leverage MagnolAI and improve this capability together,” he says.

But the main expected benefit of MagnolAI is in improving health outcomes.

“As we utilize the platform in more clinical trials, MagnolAI will help us better understand the disease journey, fuel quicker medicine development, and provide insights that streamline our drug discovery, clinical trial and treatment processes and solutions. We see the long-term value of MagnolAI and are focused on continuing to enhance MagnolAI as we get closer to a full product launch across Lilly,” he says.