Technology’s role in precision medicine is impacting the way researchers and clinicians are able to explain and predict health-related outcomes. Increasingly, through the use of innovative devices such as portables, wearables, interactives, implantables and ingestibles, life science companies are looking to use ‘digital biomarkers’ to measure physiological, cognitive or behavioural characteristics.
Digital biomarkers provide access to new types of insights and offer unique advantages by allowing data to be generated frequently and as part of the patient’s daily life. Moreover, these low-cost, high-efficiency tools can be integrated into existing biomarker strategies. With an increase in well-validated digital health tools and continued investment in this technology, we can expect to see digital biomarkers play a major role in diagnosis and health outcomes.
How can we use digital biomarkers?
Digital biomarkers can be integrated into an existing biomarker strategy as a complementary tool or as a stand-alone measure. As with traditional biomarkers, well-designed and validated digital biomarkers can generate significant advantages in study design; they can help to recruit the optimal study population, predict study outcomes or yield efficiencies in study length or size. Regulators and payers are also increasingly interested in the insights yielded by digital biomarkers. From their perspective, they can generate meaningful evidence of safety and efficacy, support marketing claims and inform payer review of clinical utility and value. Examples of digital biomarker usage cases include:
Using an interactive app to pre-screen potential study participants; for instance, a game to assess cognitive ability as a predictor of beta-amyloid burden and mild cognitive impairment in potential candidates for an Alzheimer’s study
Using behavioural data, measured objectively and at home on the days prior to and after collection of a traditional blood-based biomarker, to yield contextual information, such as activity monitoring in a diabetic or pre-diabetic population
Using digitally acquired voice recordings at home via a mobile app to track respiratory function over time, as a complement to less frequent in-clinic measurements. Such monitoring could also allow for early prediction of treatment success, eg in a study to evaluate asthma treatment outcomes
Collecting real-world evidence in the patient home setting to yield new claims. For example, monitoring heart rate and rhythm data in patients receiving post-treatment for myocardial infarction (MI) to evaluate long-term outcomes.
Developing the right digital biomarker
Integrating digital tools into a therapeutic development programme requires careful consideration of the unique features of the digital technology, its specific ability to contribute to the clinical condition or therapy under evaluation and the data backing its scientific and regulatory utility. The starting point for a digital biomarker is often the inclusion of a digital tool to generate exploratory data sets in an early-phase clinical study. This can be done with a clear hypothesis in mind – for example, inclusion of activity monitors in patient populations suffering from mobility disorders. Alternatively, digital apps and sensors can be included more broadly at study enrolment to generate a breadth of physiological and behavioural data. These data sets can then be analysed to identify a variety of potential biomarkers for future validation.
In many ways, this model parallels traditional biomarker development where, for example, blood samples can be collected early in clinical research to then support genome-wide association studies to detect genetic biomarkers. In fact, all of the traditional biomarker usage cases apply to digital biomarkers (see Figure 1). These usage cases are further enhanced by the unique advantages presented by the digital nature of these biomarkers.
However, digital biomarkers can also present unique challenges versus traditional biomarkers. Digital technology shifts at a different speed from medicine – and exists in an entirely different, less-controlled ecosystem from traditional laboratory medicine. The increasing variety of available hardware and software yields many permutations of possible digital data and tools to deliver added value for a precision therapy programme. This variety provides tremendous advantages in terms of flexibility, diversity and adaptability of tools available to generate a digital biomarker, but it also requires a clear strategy to translate the initial signals into a reliable, validated digital biomarker than can be used to yield meaningful scientific evidence. Defining a clear digital biomarker strategy upfront is key whether it covers early exploratory use, is designed for study cost-saving efficiencies or is intended to support regulatory approval of a marketing claim.
Validating digital biomarkers
A successful digital biomarker strategy integrates the perspective of the therapeutic innovator (usually a traditional pharmaceutical company) with the biomarker developer (usually a tech company or start-up). It takes into consideration the clinical objectives of the therapeutic, the nature of the hardware and software driving the digital biomarker, and the interpretability of the data generated. Specifically, a successful digital biomarker strategy requires careful consideration of these elements:
Concept of interest: What exactly is being measured by the digital biomarker?
Scientific validity: What scientific rationale links the applicability of the digital biomarker to the clinical condition?
Usage case: Which biomarker use case maximises the value of the biomarker to the development programme?
Context of use: How will the unique features of the patient population and use setting (eg home, clinic) affect the acquisition of the data?
Data: What raw data and meta-data are generated, and how will they be handled and reported?
Analysis: Which analytical approach will be used to translate the data into a digital biomarker? How reliable is the analytical approach for the usage case and context of use?
Technology: Which technological features are closed and can be controlled by the sponsors? Which are open and depend on the user or a third-party developer? How can the digital biomarker be ‘locked’ or made sufficiently robust to take these factors into account?
Analytical validity: To what extent can we rely on the digital biomarker data being generated? How accurate and precise is the data? How repeatable and reproducible are the measurements and outputs?
An understanding of these factors can inform a digital biomarker development and validation strategy to optimise the research, regulatory, market access and stakeholder value of the biomarker. Such validation strategies may entail a variety of approaches, some of which are uniquely suited to digital biomarkers, such as the use of ‘virtual’ studies completed remotely, without clinical sites, or through the use of advanced statistical approaches that pool data from the same digital tool across multiple studies. The validation strategy may then evolve over time as increasing evidence supports the utility of the digital biomarker and as the context of use expands to new patient populations or indications. Properly validating a digital biomarker can thus yield long-term value, as its utility increases over time with regulators and payers.
As life science and technology companies continue to invest in digital health, it is clear that digital biomarkers are increasingly poised to deliver significant value to precision medicine programmes. They present a low-cost, high-efficiency opportunity to yield rich, meaningful data sets to guide study enrolment, inform therapy selection, track outcomes and generate real-world evidence. Applying careful strategic thinking at the start of a digital biomarker programme can guide programme development to ensure it remains aligned with the goals of the precision therapy programme and maximises its ability to deliver value throughout the development life cycle.