The Panagora Pharma 7th Annual Digital Biomarkers in Clinical Trials Summit recently featured a compelling presentation by Rana Zia Ur Rehman from Johnson & Johnson Innovative Medicine Group. Rehman, a specialist in data science and digital health, delved into the intricate world of physiology for inflammation monitoring, focusing on the relationship between physiological changes and cytokine responses. His presentation provided valuable insights into the potential of digital biomarkers for long-term clinical trial monitoring and the challenges associated with their implementation.
The Role of Heart Rate and Heart Rate Variability
Rehman emphasized the importance of heart rate and variability (HRV) as physiological markers in predicting inflammation and flare-ups, often associated with various immune-mediated diseases. Traditionally, these markers are assessed using ECG-based devices, considered the gold standard due to their accuracy in measuring electrical signals and heart timing. However, while ECG devices are highly accurate and reliable for short-term clinical monitoring, their high cost, complexity, and need for professional setup limit their use in long-term applications. This limitation has driven the exploration of alternative devices that are more practical and cost-effective for extended use.
PPG-Based Devices: A Viable Alternative?
Rehman highlighted the growing need for cost-effective, long-term monitoring solutions, introducing Photoplethysmography (PPG)-based devices as a practical alternative, especially for chronic conditions requiring continuous assessment. PPG devices measure changes in blood volume using light sensors, making them less invasive and easier to use than ECG devices. Commonly found in wearable technology such as smartwatches and fitness trackers, PPG devices offer affordability, ease of use, and the ability to continuously monitor over long periods, particularly useful for remote patient monitoring.
Despite their advantages, PPG devices can be unreliable due to motion artifacts, which occur when movement disrupts sensor accuracy. Rehman explained that the underlying mechanisms of ECG and PPG devices differ: ECG measures electrical signals, while PPG assesses changes in blood volume. This difference makes PPG devices more susceptible to external factors such as motion and skin tone.
Validation and Error Rates
Rehman and his team conducted a comprehensive clinical trial involving 25 healthy volunteers to validate the effectiveness of PPG devices. The study aimed to compare the accuracy of PPG devices with that of ECG-based devices, which served as the gold standard reference. Participants wore two PPG devices on their dominant and non-dominant hands for two weeks, with a vital patch (an ECG-based device) used as a reference.
The study found that PPG devices showed a mean absolute error rate of 9 to 18 milliseconds for the R-to-R interval. Regarding heart rate accuracy, the error was less than one beat per minute, which is considered good. However, the accuracy varied depending on the time of day and activity level. Nighttime data, when participants were mostly stationary, was more reliable due to minimal motion artifacts, while daytime data was more prone to inaccuracies due to increased movement and activity. These findings emphasize the importance of accounting for external factors when using PPG devices for long-term monitoring.
Impact of External Factors
The study also examined the impact of various external factors on device accuracy, including motion, skin tone, and the amount of data within an epoch. Regarding motion, Rehman noted that stationary periods and lying positions yielded the most accurate readings, while motion and reduced data coverage led to higher error rates. Due to the increased motion artifacts, the PPG devices struggled to provide accurate measurements during activities such as walking or exercising.
Regarding skin tone, Rehman acknowledged that skin tone could impact the accuracy of PPG devices. Different skin tones can affect the absorption and reflection of light used by PPG sensors, potentially leading to variations in data accuracy. While the current study cohort was predominantly white, Rehman mentioned that device settings could be adjusted for different skin tones, and future research should investigate this factor further to ensure reliable assessments across diverse populations.
Inflammation Response and Digital Vital Signs
Rehman presented findings from two studies that explored the relationship between digital vital signs and inflammation response. The first study used a vital patch on ten healthy volunteers for 24 hours, while the second employed a PPG-based device for 14 days. Both studies involved administering lipopolysaccharides (LPS) to induce a cytokine response.
In the first study, the vital patch continuously monitored physiological markers such as respiratory rate, skin temperature, heart rate, and HRV. The observations included increased respiratory rate, skin temperature, heart rate, and decreased HRV, reflecting the body’s response to inflammation.
The second study, involving a longer monitoring period of 14 days, used a PPG-based device. LPS was administered on the sixth day, and the device continuously tracked physiological markers. Key observations included increased heart rate in both the active and control arms following LPS administration. The heart rate exhibited a sinusoidal pattern, rising during the day and falling at night. Rehman emphasized that this pattern needed to be accounted for in the analysis to distinguish between natural circadian rhythms and responses to LPS.
Circadian Patterns and Data Analysis
Rehman highlighted the importance of accounting for circadian patterns when analyzing heart rate data. He compared two approaches for analyzing heart rate data: discrete and continuous modeling. Discrete modeling involves setting a baseline and comparing data points to this baseline, but determining an appropriate baseline can be challenging. In contrast, continuous modeling uses a cosinor model to fit a sinusoidal curve to the data, allowing for the extraction of circadian properties such as amplitude and peak timing.
Rehman presented a visualization of heart rate data over 24 hours, showing the sinusoidal pattern of heart rate changes. He highlighted the importance of correcting for motion artifacts and using appropriate modeling techniques to analyze the data accurately. These insights are crucial for effectively implementing digital biomarkers in clinical trials and long-term monitoring.
Summary
Rehman reiterated the potential of PPG-based devices for long-term physiological assessment despite their susceptibility to motion artifacts. He stressed the need for careful data analysis and validation to ensure reliable results. The studies demonstrated that clinical trial digital vital signs could effectively track inflammation responses, provided that circadian patterns and other external factors are accounted for. The presentation concluded with a Q&A session, where Rehman addressed questions about device synchronization and the impact of skin tone on PPG device accuracy. While acknowledging the study cohort’s lack of diversity, Rehman emphasized the need for future research to explore these factors further. The insights gained from these studies pave the way for more effective and practical long-term monitoring solutions in digital health in clinical trials.
If you want to access the complete Digital Biomarker Summit event recordings, click here.
Moe Alsumidaie is Chief Editor of The Clinical Trial Vanguard. Moe holds decades of experience in the clinical trials industry. Moe also serves as Head of Research at CliniBiz and Chief Data Scientist at Annex Clinical Corporation.
