One of the most interesting features of digital biomarker technology is the composability of the operating systems, sensors, signal processing, and app-based components.
Four years ago, we published one of the first peer-reviewed papers on defining the components of a digital biomarker in Nature npj Digital Medicine. One of our key charts outlines five completely different ways you could construct a seemingly ‘simple’ measure of atrial fibrillation (AFib):
Over the past few years, investment in digital biomarker technologies has proliferated with 600+ companies developing technologies across 13+ therapeutic areas. The developers of these measures have a set of processes as they design and deploy an evidence-backed, fit-for-purpose digital health technology suitable for your patient population.
Today we’re releasing the second of three posts simplifying the digital health technology (DHT) decision making process, focusing this time on hardware (e.g., connected sensor modalities) versus software (e.g., signal processing, algorithms, apps) technology vendors.
With the dramatic development of both hardware and software-based components for digital biomarkers over the last decade, researchers looking into digital endpoints may be overwhelmed by the sheer number of options:
Historically, the majority of tools used to collect digital measures have been physical sensors — e.g. wrist-worn accelerometers, ECG patches, and digital scales. These sensors tended to have a 1:1 hardware:software relationship — an accelerometer simply runs accelerometer software.
Excitingly, the maturation of the digital measurement space has also led to the development of many software-based measurement tools that can be lightly or completely decoupled from underlying hardware. These software components, some of which are FDA-cleared under the category Software as a Medical Device (SaMD), expand the possibilities far beyond that traditional 1:1 model.
Software meets hardware, in practice
In HumanFirst's Atlas platform, we include not only SaMDs, but any software with the capacity to measure a digital endpoint. Because not all software is designed with the same goals, we categorize software-based components in four ways:
Independent software components are critical because they can be repurposed (i.e., “composable”) and customized to fit the needs of the population, hardware, or even measurement of interest by a digital measure developer.
For instance, imagine you are a research scientist developing a study in patients with heart failure. These are some questions you might consider:
- You know atrial fibrillation will be a digital measure, but which tools are analytically and clinically validated?
- Which digital health tools are available and acceptable from a utility and usability perspective?
- What if you want to include additional digital measures such as daily physical activity and sleep — will you need to add a second DHT or is there a single tool for the job?
- How do the data come back to your study team, and who might be the right data science and analytics partner to integrate with and analyze your data?
You will need to be able to answer these questions to choose which hardware and software components can work together for your digital measurement needs.
Making a decision
In the example above, let’s assume you’ve narrowed your list down to two digital health tools:
- Wrist-worn Apple Watch 8 with FDA-regulated software for the measurement of atrial fibrillation (see in Atlas)
- Chest-worn Cardea SOLO patch, a class II medical device that captures heart rhythm from a single-lead ECG (see in Atlas)
Comparing features can help build trade off scenarios as you make a nuanced decision:
For today’s landscape drop, we cut the data by component type represented in the Atlas platform, highlighting the companies developing hardware and software technologies to collect digital measures. Logos are ordered by count of evidence from each company within each component type.
Physical Sensor Components
Software-Based Components
We’re on a mission to enable the seamless incorporation of evidence-based digital health technologies & digital biomarkers into clinical research, resulting in better, differentiated treatments that improve health outcomes for all humans. Explore more open-access DHT data on GoHumanFirst.com.
