What Are Digital Biomarkers? The Complete Guide: Types, Technology, Validation & Use Cases

Digital biomarkers are objective, quantifiable measurements of physiology or behavior gathered by digital devices (like wearables, smartphones or implants) that indicate health or disease processes. In other words, they are the health data collected by your devices that reflect your biological state. Continuous heart‐rate reading from a smartwatch, patterns of typing speed on your phone, or voice changes during calls can all serve as digital biomarkers.

---

Healthcare has always been built on snapshots. A blood test taken once a year. A scan ordered after symptoms appear. A clinical visit that lasts 15 minutes. The human body, however, does not work in snapshots. It works in patterns, trends, and signals that change every hour.

Over the last decade, a quiet shift has been underway: sensors have moved from hospitals to wrists, pockets, and even clothing. Wearable devices and health apps that were widely dismissed as mere lifestyle gadgets, are now being used in clinical trials, disease monitoring, and even drug development. The convergence of increasingly affordable sensors, cloud computing, and machine learning has made it feasible to measure human biology continuously, outside clinical settings. What once required a hospital visit can now be captured passively through a smartwatch, smartphone, or patch.

This shift is transforming how health data is generated, analyzed, and trusted. At the center of this transformation is a new class of health signals known as digital biomarkers.

What is a Digital Biomarker?

A digital biomarker is an objective health indicator captured through digital sensors and algorithms. Formally, it is defined as an objective, quantifiable measure of physiology and/or behavior that is derived from a digital measure and used as an indicator of a biological or pathological process, or a response to an intervention. In simpler terms, it’s any health-related metric (heart rate, step count, voice features, etc.) measured by a digital device.

Unlike a traditional lab test, which is a one‐time ‘snapshot’, digital biomarkers provide continuous health insights, enabling remote monitoring and even early warning of conditions. For example, a smartwatch’s optical sensor (PPG) continuously measures your heartbeats to compute heart rate or blood oxygen – these serve as digital biomarkers of cardiovascular or respiratory status. Other examples include gait patterns measured by a phone’s accelerometer, typing and touchscreen dynamics reflecting cognitive function, or voice tone analysis for mental health. Because they are collected continuously and remotely, digital biomarkers can reveal subtle trends over time (like gradual slowing of reaction speed) that one-off tests might otherwise miss. Essentially, they convert everyday gadget signals into meaningful health data.

The field is expanding rapidly. The global digital biomarker market is already over $1.5–$1.9 billion and growing ~30–35% per year. There are over 1,200 companies offering digital biomarkers with more than 1,400 different digital measures.¹

Biomarker vs Digital Biomarker: The Core Difference

Traditional Biomarkers

Traditional biomarkers are well-established biological measures. They are usually discrete lab or imaging tests, such as blood glucose, cholesterol levels, tumor markers, or MRI scans. In practice, any objective and quantifiable indicator can be a biomarker if it is linked to a biological process. For example, elevated troponin levels indicate a heart attack, a high white blood cell count suggests infection, and the presence of beta-amyloid in the brain is linked to Alzheimer’s disease. 

These are typically captured in vitro (from blood or tissue) or in vivo (imaging) using clinical instruments.² Blood tests (glucose, troponin), genetic markers, imaging findings (tumor size on CT), or biopsy results are all examples of traditional biomarkers. These often require clinic visits and lab analysis. They give clinicians a quick summary of biology at a point in time, and may be invasive or costly.

Digital Biomarkers

Digital biomarkers also reflect biological or health-related states, but they come from digital technology. Instead of a blood draw or lab, they rely on sensors and software. This includes data from wearables and smartphones.

For instance, continuous heart rate or SpO₂ from a smartwatch, sleep duration from a ring, gait speed from a phone’s accelerometer, voice patterns from phone calls, or even smartphone typing speed serve as digital biomarkers. These raw signals, whether it is light pulses in PPG sensors, electrical voltages in ECG electrodes, or accelerometer motion data, are fed into algorithms that extract meaningful features such as heart rate variability, step count, facial expressions and more that correlate with health.

	Traditional Biomarker	Digital Biomarker
Definition	Biological measurements collected using clinical or laboratory methods	Physiological or behavioral measurements collected using digital devices and sensors
Data source	Blood, urine, tissue samples, imaging (MRI, CT), lab assays	Wearables, smartphones, patches, cameras, microphones, connected medical devices
Measuremen frequency	Infrequent, point-in-time (snapshots)	Continuous or high-frequency, longitudinal
Collection setting	Hospitals, clinics, diagnostic labs	Real-world environments (home, work, daily life)
Data collection type	Mostly active (patient visit required)	Passive (background sensing) and active (tasks, surveys)
Examples	Blood glucose (lab test), HbA1c, cholesterol, tumor size on MRI	Heart-rate variability from PPG, cadence from IMU, voice changes from smartphone mic
Temporal resolution	Low (days to months between readings)	High (seconds to minutes, 24/7 monitoring)
Response to change	Often reactive (measured after symptoms appear)	Proactive (detects early trends and deviations)
Cost per data point	High	Low once deployed
Validation model	Long-established clinical validation pathways	Requires analytical + clinical validation of sensors and algorithms

The key difference is ‘how’ and ‘when’ the data are collected. Traditional biomarkers measure biology directly and typically at a single point in time. Digital biomarkers measure proxy signals via technology, often continuously or passively. This continuous monitoring is a crucial differentiator - a strength. It’s like comparing a single photograph (traditional) to a live video (digital). Digital biomarkers leverage modern sensors and apps to fill in details that occasional clinic visits can’t.

Types of Biomarkers: An Easy Taxonomy

Biomarkers are often grouped into categories based on what role they play across the disease lifecycle - from risk identification to treatment response and long-term safety. While the underlying biology or data source may differ, these categories remain consistent across traditional and digital biomarkers. There are primarily 7 types of biomarkers:

1. Susceptibility/ risk biomarker

   
   

   Susceptibility (or risk) biomarkers indicate the likelihood of an individual developing a disease or condition in the future, even before any symptoms appear. They do not confirm disease presence but instead reflect predisposition.

   
   

   These biomarkers are valuable in prevention-focused healthcare, population screening, and early intervention strategies. They help stratify individuals into risk categories, enabling closer monitoring or lifestyle and therapeutic adjustments long before clinical onset. From a research perspective, susceptibility biomarkers are often used to identify high-risk cohorts for longitudinal studies or preventive trials.

2. Predictive biomarker

   
   

   Predictive biomarkers provide insight into how an individual is likely to respond to a specific intervention or therapy. Their role is not to diagnose disease, but to forecast treatment response.

   
   

   These biomarkers support personalized medicine, helping clinicians choose the most effective treatment option while avoiding unnecessary or ineffective interventions. In drug development, predictive biomarkers are crucial for patient stratification and optimizing trial outcomes. Importantly, predictive biomarkers are context-specific, i.e., their value depends on the intervention being considered.

3. Diagnostic biomarker

   
   

   Diagnostic biomarkers are used to confirm the presence or absence of a disease or condition. They answer a binary clinical question: Is the disease present?

   
   

   These biomarkers are central to clinical decision-making and are often required to meet high standards of accuracy and reliability. They are typically used at a single point in time but may be repeated if confirmation is needed. Diagnostic biomarkers form the foundation of disease identification, classification, and treatment initiation.

4. Prognosis biomarker

   
   

   Prognostic biomarkers provide information about how a disease is likely to progress over time, independent of treatment. They help estimate outcomes such as disease severity, rate of progression, or likelihood of complications. These biomarkers are especially important in chronic and progressive conditions, where understanding future trajectories can influence monitoring intensity, care planning, and patient counseling.

5. Pharmacodynamic/ response biomarker

   
   

   Pharmacodynamic (or response) biomarkers measure whether a biological response has occurred after an intervention. They indicate that a therapy is having a measurable effect on the body, even if clinical outcomes are not yet visible.

   
   

   These biomarkers are commonly used in early-stage clinical trials to assess mechanism engagement, dose optimization, and short-term biological impact. Unlike predictive biomarkers (which forecast response), pharmacodynamic biomarkers confirm that a response is happening.

6. Monitoring biomarker

   
   

   Monitoring biomarkers are used to track a condition or physiological state over time, often through repeated or continuous measurements. Their primary value lies in detecting change: whether improvement, deterioration, or stability.

   
   

   They are essential in chronic disease management, long-term follow-up, and real-world evidence generation. Monitoring biomarkers enable clinicians and researchers to move from episodic observation to longitudinal insight. This category is particularly well-suited to digital approaches due to its reliance on frequent data collection.

7. Safety biomarker

   
   

   Safety biomarkers are used to detect or predict adverse effects associated with a treatment or intervention. Their role is to protect patients by identifying early signs of toxicity, intolerance, or unintended physiological changes. They are critical throughout the drug development lifecycle and remain important even after therapies reach the market. In regulated settings, safety biomarkers often carry strict validation and reporting requirements. They support risk management and regulatory compliance.

The table below pairs each biomarker type with a traditional-digital example:

	Traditional	Digital equivalent
Susceptibility/ risk	Family history or genetic tests (e.g. BRCA gene for breast cancer risk).	Smartphone usage patterns or social media behavior that correlate with future depression risk. For instance, reduced mobility or social interaction detected by a phone could flag risk of cognitive decline.
Predictive	Gene variants predicting drug response (e.g. CYP2C19 genotype for anti-platelet therapy).	Baseline digital measures that predict who will benefit from a treatment. For example, high daily step count or physiological stress patterns might predict good response to a heart medication.
Diagnostic	Blood culture confirming infection, biopsy confirming cancer.	Algorithms analyzing gait or speech via smartphone to diagnose Parkinson’s disease; or a smartwatch ECG algorithm that identifies atrial fibrillation (AFib) episodes.
Prognostic	Tumor stage predicting cancer outcome.	Continuous mobility data predicting prognosis. For instance, decreasing walking speed (measured by accelerometer) might predict higher fall risk or faster cognitive decline in dementia.
Pharmacodynamic/ response	Blood drug levels or biomarkers like HbA1c dropping with diabetes therapy.	Wearable tremor sensors showing reduction of Parkinson’s tremor after medication, or a smart inhaler tracking improved breathing patterns after asthma meds.
Monitoring	Periodic MRI scans to monitor tumor size.	Continuous home measurements like a smart ring tracking sleep and vitals to monitor heart failure or COPD.
Safety	Liver enzyme levels rising on a drug indicate toxicity.	Sensors detecting abnormal heart rhythms or drastic blood pressure spikes as a drug side effect. For instance, a sudden drop in activity or oxygen saturation picked up by a wearable might serve as an early warning of drug-induced complications.

Any traditional biomarker can have a digital analogue, sometimes with richer, longitudinal data. Digital biomarkers share the same ‘business goal’ as traditional ones, be it risk detection, diagnosis or monitoring, but achieve it via technology. Traditionally a doctor might check blood pressure once a month. Digitally, a smartwatch could measure blood pressure (or pulse) daily. This continuous data is a digital monitoring biomarker, providing a finer-grained view of the patient’s status.

From Digital Biomarkers to Clinical Insights - The Technology Involved

Digital biomarkers are created through a structured pipeline that transforms raw sensor data into clinically meaningful insights. It begins with data collection, where devices such as wearables, smartphones, or implantables capture physiological and behavioral signals either passively (continuous heart rate, movement, sleep) or actively (prompted tasks like voice recordings or cognitive tests). Passive data offers low user burden and real-world context, while active tasks are useful for targeting specific functions such as reaction time, memory, or speech. These devices capture raw signals, including optical (PPG), electrical (ECG/EEG/EMG), motion (IMU), biochemical, acoustic, visual, or electrodermal; but at this stage, the data is just unprocessed time-series information.

The next stage is signal processing and algorithms, where raw data is cleaned, denoised, and transformed into meaningful features. Algorithms extract metrics such as heart rate variability, gait stability, tremor amplitude, or speech patterns, and machine-learning models may combine multiple features to identify health-relevant patterns. Early outputs may be simple measures like step count or average heart rate, while more advanced models produce composite digital biomarkers, such as disease severity scores or risk probabilities. Calibration and validation at this stage are critical to ensure reliability, repeatability, and alignment with clinical ground truth.

Finally, the processed output becomes a digital biomarker - a value, trend, or predictive indicator that represents a clinical concept - and is linked to a Clinical Outcome Assessment (COA). This association establishes clinical relevance, either by using the digital measure directly as an endpoint or by correlating it with established clinical exams, questionnaires, or diagnostic criteria. In practice, this pipeline is iterative: data collection, algorithm refinement, and clinical interpretation continuously inform each other. Much like lab tests, digital biomarkers turn biological signals into actionable insight, with the advantage of being continuous, scalable, and embedded in everyday life.

Validation and Clinical Utility

To be useful, digital biomarkers must be proven accurate and meaningful. This involves analytical and clinical validation.

• Analytical validation 

  
  

  This ensures the technology reliably measures what it claims. For sensors, that means accuracy, precision, and consistency of data. For algorithms, it means correctly computing the metric (e.g. heart rate error <5%). Basically, analytical validation asks: if we measure the same thing twice, do we get the same result? Is the signal clean enough? It’s akin to calibrating a scale or lab instrument. High-quality digital biomarker development includes rigorous bench testing and comparison to gold standards, like comparing wearable ECG to a clinical ECG to confirm replicability.

  
  

• Clinical validation

  
  

  This demonstrates that the digital biomarker actually relates to the health concept of interest in people. In other words, does the metric predict or reflect clinical outcomes? For instance, if a wearable-derived step-count score is higher, do patients report better mobility? Clinical validation typically involves studies showing the digital measure correlates with disease severity, progression, or response. For example, researchers might show that a smartphone-based memory test score changes in people who later develop Alzheimer’s. Clinical validation often requires population data and statistics.

Digital health measures are only “fit-for-purpose” if both kinds of validation are satisfied. That means a sensor algorithm must produce stable readings (analytical) and those readings must tie to real patient outcomes (clinical). In practice, this might involve testing sensors in lab conditions, running pilot studies comparing the digital biomarker to standard measures and checking for statistical performance: sensitivity, specificity, correlation coefficients.

Because digital biomarkers often inform clinical decisions, they fall under medical regulations. If a smartphone app or wearable is intended for regulated use (like drug trials or diagnosis) it must navigate medical device rules. Developers usually engage with regulatory bodies early, decide if the tool needs clearance or qualification, and build evidence accordingly. The upside is, once a digital endpoint is accepted (as a qualified biomarker/COA), it can be reused broadly in research and trials.

Key Applications and Use Cases

Digital biomarkers are finding an array of applications across medicine. Here are some highlights:

• Clinical Trials: Wearable sensors are increasingly used to enrich clinical trials. By collecting continuous data, trials gain statistical power and patient-centric endpoints. For example, studies use gait and tapping data to measure Parkinson’s progression instead of relying only on infrequent clinic ratings. Cardiology trials use remote ECG or blood-pressure monitors to track outcomes. Overall, digital biomarkers make trials more sensitive (detecting subtle drug effects) and inclusive (enabling participation of those far from trial sites).

• Remote Patient Monitoring: In chronic disease management, digital biomarkers allow doctors to monitor patients at home. For instance, cardiac patients wear patches or use phone ECGs to monitor arrhythmias or heart failure decompensation. Diabetes patients use continuous glucose monitors to keep glucose in range. Respiratory patients use smart inhalers and pulse oximeters to track breathing. Especially during COVID-19 and beyond, clinicians have leveraged wearables to follow blood pressure, oxygen saturation, or sleep metrics remotely, enabling timely interventions.

  
  

• Neurology and Psychiatry: Continuous, subtle changes are crucial in brain diseases. Today, voice analytics (tone, pauses) can flag depression or schizophrenia episodes. Wearables can track tremor and gait in Parkinson’s, enabling more precise symptom scoring. Slowed walking speed or altered gait variability from a wearable may predict dementia risk. Eye-tracking and pupillometry via phone cameras are also being explored as digital biomarkers for concussions or dementia.

• Prevention and Wellness: On the consumer side, many wearables use physiological data to guide healthy behavior. Heart rate, sleep quality and stress metrics give feedback on overall well-being. These are not “medical” per se, but form a kind of preventive digital biomarker: identifying when high stress or poor sleep might warrant medical check-ups. 

• Oncology: Although less mature, oncology is exploring digital biomarkers too. For instance, tracking fatigue or activity levels in cancer patients via wearables can objectively quantify treatment side effects. Smart pill bottles that record medication use can serve as adherence biomarkers.

• Public Health and Epidemiology: Aggregated, anonymized digital biomarker data (like population sleep patterns or mobility trends from smartphones) can inform public health. During pandemics, aggregated heart-rate or temperature data from wearables could help detect outbreaks earlier.

Challenges and Limitations

• Privacy & Data Ownership: Digital biomarkers collect sensitive personal health data (location, behavior, vitals). Protecting this data is critical. Regulations such as the Health Insurance Portability and Accountability Act in the US and General Data Protection Regulation in the EU apply, but often digital health sits in a gray zone. Standardizing consent, anonymization, and ownership rights is still underway. Users too may be reluctant to continuously share data, and ensuring secure storage/transfer is a challenge.

  
  

• Interoperability & Standards: There are countless devices and proprietary data formats. Combining data from different wearables or healthcare systems is hard. Lack of common data standards means researchers must often build custom pipelines for each device. Without standardization, building large, multi-source studies is laborious. Think of it this way: It is akin to trying to merge pictures taken by different camera brands without a common JPEG; it can be done, but requires extra effort.

  
  

• Bias and Generalizability: Machine learning models for digital biomarkers can inherit biases from training data. If a study only includes young, tech-savvy participants, the resulting biomarker may not work well for older or diverse populations. For example, gait features might differ by age or gender, so a “fall risk” model needs a representative sample. Addressing this requires large and diverse datasets, which can be expensive to collect.

  
  

• Validation Burden: Thorough analytic and clinical validation is resource-intensive. Large sample sizes, clinical studies, and sometimes regulatory trials may be needed before a digital biomarker gains acceptance. This can slow down deployment.

  
  

• Data Overload and Interpretation: Continuous monitoring generates huge data volumes. Analyzing it meaningfully requires robust algorithms and computing. Clinicians may also be unprepared for how to interpret streams of digital data. Without clear guidelines, there’s a risk of alert fatigue or misinterpreting fluctuations that are actually normal.

  
  

• Regulatory Uncertainty: The line between biomarker and COA can be blurry, with clear frameworks still in development. This uncertainty can deter investment. Developers may be unsure what evidence is needed for approval.

The Future of Digital Biomarkers

Digital biomarkers are entering a phase of deeper integration and maturity. The future lies in multi-modal biomarkers, where multiple data streams such as physiology, movement, behavior and speech are combined into composite measures that reflect health more holistically. Instead of relying on a single signal, these biomarkers capture patterns across time and context, making them better suited to complex conditions. At the same time, advances in AI and federated learning will allow algorithms to learn from large, decentralized datasets without compromising privacy, improving accuracy while keeping sensitive health data local.

Equally important is the shift toward continuous and ambient monitoring. As wearables become more discreet and new sensor form factors emerge, data collection will fade into the background of daily life. Digital biomarkers will also become increasingly personalized, adapting to individual baselines rather than population averages. This transition, from episodic measurements to always-on, individualized insight marks a fundamental change in how health is understood and tracked.

Together, these trends point toward a healthcare model that is proactive rather than reactive. Digital biomarkers will no longer exist as isolated signals or standalone apps, but as part of integrated digital health ecosystems, working alongside clinical records, genetics, and diagnostics. As regulatory frameworks evolve and real-world evidence grows, digital biomarkers have the potential to move from experimental tools to trusted clinical instruments, enabling earlier intervention, better outcomes, and a more continuous understanding of human health.

---

References:

1. The opportunity of digital biomarkers unlocking next-generation healthcare. (n.d.-c). https://www.capgemini.com/wp-content/uploads/2024/10/The-opportunity-of-Digital-Biomarkers.pdf 

2. Izmailova ES, Demanuele C, McCarthy M. Digital health technology derived measures: Biomarkers or clinical outcome assessments? Clin Transl Sci. 2023 Jul;16(7):1113-1120. doi: 10.1111/cts.13529. Epub 2023 Apr 29. PMID: 37118983; PMCID: PMC10339690.