How Can Pharma Build Trust In Connected Sensors?

By Ed Miseta, Chief Editor, Clinical Leader

The growth in the use of connected sensors in clinical trials has been nothing short of astounding. An estimated 275 million wearables shipped globally in 2021 and that number is projected to double in the next four to five years.

Connected sensor technologies now have the potential to greatly improve data collection and evidence generation in clinical research. These sensors, which are becoming more sensitive and sophisticated, have the potential to increase the precision and accuracy of clinically relevant signals while also reducing recall bias. They can also improve inclusivity in trials and improve the overall patient experience.

Challenges Prevent Development And Use

Despite the growth and advancements in sensor and wearable technologies, there are still challenges for companies opting to make use of them in trials. HumanFirst, a healthtech company that serves teams looking to deploy connected technologies for clinical research and care, recently convened an Advisory Committee of stakeholders that are driving adoption of digital measurement in clinical trials. The committee included Digital Health leaders from large biopharma companies like Genentech, J&J, Novartis, Biogen, BMS, as well as the Digital Medicine Society (DiMe). Their goal was to discover what must be done to remove key barriers to sensor adoption and trust, which they synthesized into a white paper identifying 5 key challenges and a proposed way forward.  

Evaluating and comparing connected sensors has become more complicated, which is impacting sensor adoption. One concern the committee identified is whether the data generated across sensors is truly comparable and interoperable. For example, sponsors need to know if heart rate variability derived from one EKG sensor is commensurate with heart rate variability derived from another product. The potential of connected sensors and digital technologies in trials is undeniable. For that reason, researchers and clinicians need better evaluation frameworks to assess their benefits and risks. That framework would provide a clear path between makers and users and encourage future development and use.

V1.0 Leads The Way

To overcome some of the existing challenges, V1.0 of a connected technology evaluation framework has been published in Nature npj Digital Medicine. The framework entailed involvement and input from experts across organizations including HumanFirst, DiMe, University of North Carolina, Biohacking Village at DEF CON, and Sage Bionetworks. The framework was based on five considerations.

The first consideration is V3 – Verification, Analytical Validation, and Clinical Validation. This consideration looked at whether the tool measures what it claims to measure and if the measurement is appropriate for the condition of the target population.

The second consideration is security and looks at whether the manufacturer built safety by design, if there is a disclosure policy and a software bill of materials, and whether vulnerability in a single software component of the technology can be exploited and might adversely impact patient care.

The third consideration revolves around data rights, privacy, and governance. It looks at who has access to the data and when, and whether the privacy policy is publicly accessible. Utility and usability are the fourth consideration, which examines how the tool is worn and for how long, how easy it is to deploy and use, and the ease of moving data from the sensor and into a usable analytical environment. It also examines utility and usability, such as how it is worn, battery life, and convenience for the user.

The final consideration is economic feasibility, which examines net benefit versus cost, the cost structure, and how sponsors might use the sensor at the conclusion of a trial.

“Think of this evaluation framework like a drug or nutrition label balancing five components based on your needs,” notes the report. “While this evaluation framework was an important first step, we need to further evolve existing evaluation frameworks to address the challenges experienced by early adopters of wearables in trials, particularly around data formats and standardization.”

Challenges Could Limit Adoption

To further promote the use of sensors in trials, challenges that have the potential to limit adoption will need to be addressed. The Advisory Committee of stakeholders was asked to identify and articulate their learnings from sensor selection and discuss the path forward. Their insights could then be used to create a stronger evaluation framework as a guide for sensor manufacturers.

The Committee was able to identify five key challenges that could limit the adoption of connected sensors in trials. The industry will need to overcome these challenges if it hopes to continue the growth and use of sensors.

The first challenge relates to data collection standards. Without established standards, such as sampling frequency and sensor output units, comparing data streams across different sensors would be a challenge. Without standards, drug sponsors would also be left to decipher preprocessed data outputs and determine how to reconcile them with data streams from other sensors. An additional challenge is the fact that appropriate sampling frequency and wear time depend on the sensor type, body location, and the activity being measured.

Similar to data collection standards, the second challenge exists around longitudinal compatibility. Even with the same connected sensor there are changes in data collection and processing that can occur over time. This can be the result of companies updating operating systems, firmware, and algorithms that run in the background on a connected technology.

For example, a mid-trial firmware update to a smartwatch might include a change in the sampling frequency of an embedded sensor. There is also a high degree of volatility in the manufacturer space which is often due to acquisitions. That means sponsors are faced with a lack of confidence that a given vendor will be around in the future. Both issues can hamper the purchase and use of sensors.

The third challenge relates to commercial-grade products. Wearable devices and sensor technology development must be rooted in patient needs and undergo sufficient user testing. The selected technology must also offer adequate data rights and security when capturing patient data. Unfortunately, for vendors producing consumer-facing technologies for use in clinical trials, it can be difficult to gain approval for the use of their products in pharma companies.

The fourth challenge is the need for near-real-time data access. To use sensor-derived data in clinical trials, researchers need to know in real-time whether patients are using a connected technology in the way required. Those research teams must be alerted as soon as possible if a patient removes their smartwatch or if there is an issue with the data transfer. When selecting a technology for use in a trial, sponsors are often unclear of how much access they will have to participant-level data. Often, sponsors will also not know what access they will have to near-real-time data during the trial. Wear time and compliance data is critical in ensuring that the sensor data is complete, that it conforms to data standards, and will be available for analysis. Sponsors will sometimes find out too late that there is an issue with data quality due to missing or incomplete data, which could jeopardize trial results.

The final challenge relates to geographic availability. Many sponsor companies conduct global trials, and it is essential for them to know if a connected sensor technology is available across many geographic locations.

Is There A Way Forward?

The Advisory Committee believes there is a way forward for connected sensor technologies if the industry can address these challenges. More robust evaluation frameworks and clear data standards by sensor type can have a huge impact on how these technologies are viewed and adopted. The Committee recommends four solutions that will help pharma and biotech companies trust, verify, and adopt sensor technologies for use in trials.

First, set industry standards are required for preprocessed data outputs for each sensor type. An alternative solution would be for all technology manufacturers to report standardized metrics for data collection. Those metrics, such as timestamps, sampling frequency, and output units, are needed to compare data across different sensors.

The second recommendation is the publishing of consensus standards. The Committee believes vendor alliance teams from pharma companies could publish consensus standards across parameters such as certifications, GCP/GXP compliance, and hardware. This will need to be done while recognizing that different use cases will require varying levels of granularity. These standards would streamline and facilitate vendor approval for new organizations.

The third recommendation is the publishing of data access. The committee believes technology manufacturers should publish data access and transfer capabilities for their products alongside API/SDK availability and capabilities. Researchers must understand the granularity with which they will be able to assess patient adherence to specific products and to ensure data quality throughout the study.

The final recommendation again relates to geographic availability. For those sponsors conducting global trials, vendor companies should publish information on geographic availability across the domains relevant to their products.

Click here to view the full white paper.