How digital biomarkers are shaping the future of healthcare

What is a digital biomarker?

Traditionally, a biomarker is defined as a biological molecule that can be used to detect the presence of a condition, process or disease found in the bloodstream, tissue or bodily fluids. A biomarker might be something as simple as someone’s blood pressure detected using a pressure meter, through to more complex genetic tests analysed in a lab. What makes a biomarker ‘digital’ is when the measurement is taken through a digital device. Digital biomarkers can vary from heart-rate measurements recorded from a wearable device, to data from smartphone usage, such as a person’s speech.

How are digital biomarkers impacting healthcare?

Digital biomarkers are currently revolutionising medical research, diagnostics and therapeutics, leading to rapid advancements in patient diagnosis and monitoring in the industry, from neurology to cardiology. This technology is impacting healthcare in a variety of ways.

Digital biomarkers for early detection of diseases

Digital biomarkers can help patients by detecting diseases early. Studies have shown that researchers are more likely to detect a disease early by gathering health data from someone’s smartphone or tablet, rather than during a healthcare appointment. Digital health technologies can now derive numerous biomarkers from various devices simultaneously, for example, from wearable heart rate monitors worn by patients and smartphone usage.

Digital biomarkers can be especially useful for the early detection of cognitive impairment, which can be problematic due to many patients being unaware of, or reluctant to discuss this issue with their general practitioners. For example, Parkinson’s disease can be identified by monitoring and measuring any changes in the pace of an individual’s movement, loss of automatic movements and further changes in gait and speech. To show the potential for large-scale monitoring of patients with Parkinson’s, researchers carried out a wearable sensor study involving 954 participants from The Netherlands and North America. Participants used a pebble smartwatch and a smartphone application from the Michael J. Fox foundation to track their activity, sleep, medication, administration and reported symptoms such as tremors. This is a useful example of how data can be gathered remotely from a patient’s real environment of use, providing key insights for more accurate diagnosis and treatment.

Using digital biomarkers in clinical trials

Digital biomarkers have huge potential to support and successfully progress drug delivery and development. The utilisation of this technology in clinical trials, in particular for neurodegenerative disorders such as Alzheimer’s and Parkinson’s, is a key area of interest. This is because the current data available for clinical studies in this area is sparse, with diseases primarily diagnosed through neurocognitive assessments or invasive and costly procedures. The addition of digital biomarkers has already had a significant impact on enabling decentralised clinical trials through the remote monitoring of patients and has the potential to be a huge step towards minimising phase 3 failures.

Preventative benefits

Cardiovascular diseases currently hold the largest market share in the global digital biomarker market, owing to the potential preventative benefits its use can bring. An example of this can be seen in Apple’s recent Food and Drug Administration-cleared algorithm for binary identification of atrial fibrillation, used in the Apple Watch wrist-based electrocardiogram, which can create viewable cardiograms. Similarly, recent cardiac rehabilitation programmes have been coupled up with BioMets (a wearable ECG device) and cardiac digital biomarkers to help patients.

Long-term disease monitoring

For long-term conditions like asthma and diabetes, pharmaceutical companies are increasingly trying to find more efficient approaches to long-term monitoring. One of the key areas being explored is sensing modules in drug delivery devices such as inhalers, which allow pharmacos to measure and monitor the way that devices are being used. The data collected has a wealth of potential uses, such as helping to predict problems in patients’ conditions before they start to arise. It can also be used to identify challenges around patient adherence, highlighting instances where users are not using their inhalers correctly and failing to receive a full dose.

Remote health predictors and screening

Traditionally, psychiatric research has relied on observations that are subjective to the doctor/clinician who is providing them. Psychiatric digital biomarkers can offer an alternative approach to health screening and provide remote health predictors for patients. For example, a smartphone application could be used to detect early signs of treatment failure in chronic schizophrenia patients following hospital discharge, or machine learning-based voice analysis could be used to acquire digital biomarkers of cognitive functioning in individuals suffering from trauma.

Wearables and the data they provide can also be useful for digital phenotyping and the detection of mental disorders such as depression, by examining correlations between daily behaviour, physiology and mental well-being. These biomarkers could be used for a variety of purposes, such as predicting the risk of depression in individuals and upscaling health screening for mental health. An example of this can be seen in a cross-sectional study of 290 healthy working adults wearing Fitbit Charge 2 devices, to help detect subjects at risk of depression. While there were limitations on the ability of the data to detect depression in the whole sample, when contrasting subsamples, the model was able to detect subjects at risk of depression with an accuracy of 80%.

An article by Müller et al. does however suggests that digital phenotyping is more viable for monitoring people who have already been diagnosed with mental conditions on a voluntary basis, meaning this currently has limited use for wider populations. The main concern Müller highlights is that employing random screening on populations would be too great an infringement of privacy. False-positive positives can also arise if individuals are incorrectly selected in a population believed to have, or be at a risk of having, a mental disorder.

Real-time disease monitoring

Recently, biopharmaceutical companies have been utilising data collected from physiological sensors that offer remote monitoring for patients infected with Covid-19. The data extracted from these wearable sensors presents a measure of symptoms and identification of real-time digital biomarkers which are distinguished and reported through analytics. By implementing this plan, contact between patients and the public could be limited to a minimum, which would help prevent the transmission of the Covid-19 virus. Such applications highlight the great potential for real-time disease monitoring through this technology.

What are the current challenges impacting the use of digital biomarkers?

Semantics

Digital biomarkers have a high degree of ambiguity in semantics, primarily because of the merging of several fields (healthcare, engineering and analytics). There is consequently a lack of standardisation in the terminology for novel concepts such as longitudinal data, as well as multiple definitions for the term “digital biomarker” itself in scientific literature. The FDA has addressed some matters regarding these semantics, highlighting that conflating terms can “hamper communication and evidence expectations between medical product developers and regulators.”

Data privacy

The complexity of regulatory control is currently one of the factors hindering innovation around digital biomarkers. Due to deficits in standardisation, as well as their vast applications across drugs, biological and medical devices, there is no consistent regulatory framework that applies to them. The industry is currently faced with minimal guidelines on how to maintain and store the data gathered, which could lead to gross misuse of confidential patient information. There have also been increasing problems around how to correctly analyse and use accurate baselines to identify and interpret pertinent data, due to a lack of research on the topic.

The Health Level Seven (HL7) International organisation suggested the Fast Healthcare Interoperability Resources (FHIR) standard to help make digital data more systematic and interpretable. The procedure of converting data into digital biomarkers has however inflated costs, partially due to the lack of standards and validation protocols in research around their use.

Popular devices like Apple watches and smartphones have also suffered from a cloud of suspicion and disbelief surrounding their clinical validity. Though the ECG app and irregular rhythm notifications for Apple are now CE marked, most health functions are not CE marked or FDA approved. According to the Innovative Medicines Initiative, a partnership between the European Union and the European pharmaceutical industry, as well as a collaboration between international regulatory bodies, both the CE mark and FDA approval are important for the successful adoption of digital endpoints.

The use of health-related and genetic information is another challenge impacting the use of digital biomarkers, as it is subjected to specific rules,falling under “special categories of data” (formally known as “sensitive data”). Under Article 9(1) of the General Data Protection Regulation (GDPR), the processing of health data is prohibited unless it has an exception falling under Article 9(2). It is the responsibility of the data controller, most typically the company providing the digital device that will gather the data, to assess if an exception applies and to ensure relevant safeguards are in place to protect the data subject’s constitutional rights. This can be a complex process, requiring careful consideration from companies seeking to use digital biomarkers to capture personal data.

Proprietary issues can also cause a lack of transparency, owing to the fact that some people will not want to offer up their information and companies are not entitled to it. Applications, such as Apple’s ECG, are almost always proprietary, which means the verification and validation processes of the products, BioMeTs and digital biomarkers are not published to the public market. To help resolve this, studies are looking into developing a standardised open-source data and software platform where digital biomarker information can be shared: a concept that the Digital Biomarker Discovery Pipeline (DBDP) is aiming to achieve.

Clinical validation

Digital biomarker developments follow the V3 (verification, analytical validation and clinical validation) process. The most resource-consuming of the three is the clinical validation process, which usually requires drug development sponsor engagement to facilitate. There is currently a gap in the evidence required for V3 for digital biomarkers, which has also slowed down their progress.

The future of digital biomarkers

The application of digital biomarkers has been a pivotal breakthrough for the medical field and the use of this technology for treatments appears to be growing progressively alongside the use of AI in diagnosing, monitoring and treatment of medical conditions. While the research around their use is still in its early stages and needs further exploration, as pharmacos continue to invest in digital technologies, we can expect the use of digital biomarkers to grow over the next 3-5 years. Once the rules and regulations are established, such as those surrounding the proprietary aspect, businesses and healthcare establishments will have a better understanding of how to utilise and apply this valuable source of data to aid public health.