Reviewed by the Help Dementia Editorial Team — our editors review every article for accuracy against guidance from the National Institute on Aging, the Alzheimer’s Association, and peer-reviewed sources.
Digital biomarkers are measurable biological signals collected through wearables, phones, and home sensors that researchers can use to detect Alzheimer’s disease earlier and track its progression more objectively than traditional clinical visits. Unlike the PET scans and lumbar punctures that require expensive clinic appointments, digital biomarkers capture real behavior patterns—how someone walks, sleeps, types, or moves through their day—continuously and from home, offering researchers a window into cognitive decline as it actually unfolds rather than in snapshots during office visits.
These tools matter because Alzheimer’s pathology begins silently in the brain up to 20 years before symptoms appear. A person might feel fine while amyloid and tau proteins accumulate, but subtle changes in gait speed, sleep fragmentation, or keystroke dynamics can hint at that underlying damage long before memory problems become noticeable. This shift from waiting for symptoms to catching the disease in its silent stages has fundamentally changed how researchers approach prevention trials and early detection.
Table of Contents
- Which Physical and Behavioral Changes Can Researchers Actually Measure?
- Why Are Digital Biomarkers More Reliable Than Doctor’s Office Assessments?
- How Are Clinical Trials Using This Data?
- What Are the Practical Barriers to Adoption?
- Can Digital Biomarkers Predict Who Will Actually Develop Symptoms?
- What Does Continuous Monitoring Look Like in Practice?
- Where Digital Biomarkers Fall Short in Real-World Dementia Care
- Frequently Asked Questions
Which Physical and Behavioral Changes Can Researchers Actually Measure?
digital biomarkers capture changes that happen long before someone forgets their appointment. Gait slowing—the way walking speed and stride length decrease—shows up in people with preclinical Alzheimer’s, often years before cognitive tests find anything wrong. A person using a smartphone-based app might show slight delays in reaction time or accuracy on memory tasks, but more telling is their keyboard dynamics: the time between keystrokes becomes less consistent, finger taps become lighter, and the rhythm of typing changes in ways the person never consciously notices. Sleep disruption is another early signal. People in the preclinical stage often fragment their nighttime sleep or shift more toward daytime napping, patterns that a smartwatch tracks automatically.
Some research has found changes in voice patterns—a subtle slowing of speech rate or changes in how high or low someone’s voice sits—detected through smartphone microphones during daily conversations. The advantage here is scale: researchers can enroll hundreds or thousands of people into studies without requiring them to travel for appointments, and the data streams continuously rather than once a year. One real limitation is that these changes are subtle. Researchers need to collect months of baseline data before abnormal patterns emerge clearly enough to be confident in what they’re seeing. A single keystroke that’s slower doesn’t mean anything; detecting a meaningful trend requires statistical power that only comes from thousands of observations.
Why Are Digital Biomarkers More Reliable Than Doctor’s Office Assessments?
Traditional cognitive testing relies on how a patient performs on a single day when they’re in a clinic—awake, focused, often anxious, possibly on no breakfast. A person with mild cognitive impairment might actually perform fine that day because they’ve compensated well, or they might perform worse because of sleep deprivation the night before. A doctor watching someone walk from the waiting room to the exam table has seconds of data; a sensor collecting continuous gait parameters has weeks or months. The repeatability problem haunts standard testing. If a neuropsychologist gives the same cognitive test twice in a month, a person’s score can fluctuate by 5 to 10 points just from practice effect, distraction, or variation in how awake they are.
Digital biomarkers reduce this noise because they’re collecting thousands of measurements under real-world conditions—walking up stairs at home, not following a line in an exam room; typing emails, not transcribing a test sentence. But there’s a critical catch: digital biomarkers require validation against actual brain pathology before they can replace clinical judgment. Researchers cannot assume that a person whose gait has slowed slightly over a year has Alzheimer’s disease without confirming it through PET imaging or cerebrospinal fluid testing. The biomarker becomes useful only once studies have shown the relationship between the signal and the actual disease. Early data is promising, but large validation studies are still underway.
How Are Clinical Trials Using This Data?
One of the largest applications right now is in anti-amyloid drug trials, where researchers want to know if their treatment actually slows cognitive decline. Traditional trials measure cognitive change once a year or twice a year, which requires participants to perform well on tests. With digital biomarkers embedded in a study—say, a smartphone app that tracks reaction time and memory task performance twice a week—researchers get far more granular data on how quickly decline is happening. The Anti-Amyloid Treatment in Asymptomatic Alzheimer’s (A4) trial, for example, is following cognitively normal people with evidence of brain amyloid to see if removing amyloid prevents decline.
Researchers added digital cognitive testing so they could measure changes week by week rather than waiting a year for the annual clinic visit. This allows trials to be more efficient: if a drug is truly slowing decline, digital biomarkers may detect it before traditional tests do, potentially shortening the time needed to show benefit. Wearable sensors are also being used to track secondary outcomes. Studies of sleep-promoting interventions can now measure whether a drug actually improves sleep architecture—not just asking someone if they slept better, but measuring how fragmented their sleep was before and after treatment. This level of objectivity prevents placebo effect from obscuring whether a treatment actually works.
What Are the Practical Barriers to Adoption?
For digital biomarkers to move from research studies into routine clinical care, they need to work reliably in the actual population of people with cognitive decline. A smartphone app works well in a research study with educated participants who are motivated, but the average person with mild cognitive impairment may struggle with updating software, forgetting to charge their phone, or getting frustrated with the interface and abandoning it. Data privacy is another real constraint. Continuous measurement of keystroke dynamics, voice patterns, or location data (inferred from phone GPS) means collecting very sensitive information.
Patients understandably worry about what happens to data that reveals their movements and behavior patterns. Regulatory frameworks around data security for these biomarkers are still developing, and many older adults—the population most at risk for Alzheimer’s—are uncomfortable with the level of monitoring required. The cost question is practical too. A smartwatch that a research participant receives for free as part of a clinical trial is one thing; asking patients to buy their own wearables and ensure they’re compatible with research-grade software is another. Insurance rarely covers the cost of wearables used purely for research or early detection, so adoption would depend on patients financing it themselves.
Can Digital Biomarkers Predict Who Will Actually Develop Symptoms?
This is where the hype often outpaces reality. Researchers have shown that gait slowing, cognitive slowing, and sleep fragmentation correlate with amyloid accumulation in the brain and predict faster cognitive decline. But correlation is not destiny. Some people with these biomarker changes will never develop dementia in their lifetime, especially if they’re 75 and have other serious health conditions that will likely claim them first. A 55-year-old with a family history of Alzheimer’s, amyloid in their brain, and measurable gait slowing is in a different risk category than an 85-year-old with the same findings.
The same biomarker pattern means something different depending on age, education, cognitive reserve, and competing health risks. Using digital biomarkers to identify truly high-risk individuals requires not just measuring the biomarker but understanding it in context—a complexity that simple algorithmic screening approaches often miss. There’s also the question of what to do with the information. If digital biomarkers predict that someone will decline cognitively over the next 5 to 10 years, the proven intervention options are limited. Cognitive training, exercise, Mediterranean diet, and sleep optimization all show modest benefit in observational studies, but no treatment has definitively prevented decline once someone has preclinical Alzheimer’s disease. So telling someone their keystroke dynamics suggest early decline, but then offering only lifestyle modification, can feel more anxiety-inducing than helpful.
What Does Continuous Monitoring Look Like in Practice?
A person enrolled in a digital biomarker study might download an app that sends them 3 to 4 brief cognitive tests each week—matching symbols, remembering lists, simple reaction time tasks. Meanwhile, if they’re wearing a smartwatch, it silently records their heart rate variability, sleep patterns, and physical activity. Their smartphone’s microphone might record brief voice samples during the tests, capturing speech rate and voice quality. None of this requires a clinic visit; it all happens at home on the person’s schedule.
The research team aggregates this data weekly, looking for meaningful deviations from each person’s own baseline. A person whose keystroke timing has historically been steady might show increasing variability, or someone whose sleep has been fragmented might show even more fragmentation. Statistical algorithms flag these changes, but a human researcher still reviews them before concluding anything is wrong. The feedback loop is real-time enough that if the data shows genuine decline, the research team can reach out to the person and potentially refer them for clinical evaluation.
Where Digital Biomarkers Fall Short in Real-World Dementia Care
Even with all the advantages, digital biomarkers still cannot replace clinical judgment and cognitive testing for making a diagnosis. A person with a 3-year history of forgetting appointments, getting lost in familiar places, and struggling to manage finances has dementia regardless of what their smartphone app says. Conversely, someone whose digital biomarkers look normal but whose family reports clear decline in function needs evaluation, not reassurance from an algorithm.
In practice, digital biomarkers are becoming a screening and monitoring tool within a clinical framework, not a replacement for one. A primary care doctor can use gait data from a smartwatch during a routine visit to flag someone who might warrant neuropsychological testing. A neurologist can use week-to-week changes in response time from a cognitive app to gauge whether a new medication is helping or hurting. The biomarker becomes another piece of evidence, valuable precisely because it’s objective and continuous, but only useful when integrated into a conversation with a clinician who knows the person’s full history, values, and clinical context.
- —
Frequently Asked Questions
Do I need a smartwatch or special device to participate in digital biomarker research?
Most digital biomarker studies use smartphones only, since most people already have them. Some studies provide wearables like smartwatches or wristbands as part of the research protocol, but you won’t be required to purchase equipment.
If digital biomarkers show I’m declining, does that mean I’ll definitely develop dementia?
No. A biomarker change means your risk is higher than someone without that change, but many people with early biomarker changes never develop cognitive symptoms, especially if they’re older or have other health conditions. Your doctor would use this information alongside clinical symptoms and your personal situation to assess actual risk.
Are digital biomarker studies only for people with a family history of Alzheimer’s?
No. Some studies focus on people with family history or genetic risk, but others recruit broadly from the general population to understand how these biomarkers perform in real-world groups. Talk to your doctor about which studies might be appropriate for you.
What happens to the data from my phone or wearable?
Research teams are required to protect your data under strict privacy regulations (HIPAA in the US), but the level of data security varies by study. You can and should ask each research team about their data privacy practices, how long they keep data, and whether it will be shared with other researchers before enrolling.
Will my insurance cover digital biomarker testing?
Not currently. Digital biomarker testing for early Alzheimer’s detection is still largely part of research studies, not standard clinical practice, so insurance does not cover it. In clinical trials, the research team provides devices and testing at no cost to participants.
Can my doctor order a digital biomarker test to check my cognitive health?
Probably not yet. Digital biomarkers are not approved by the FDA as diagnostic tools, so most doctors cannot order them outside of a research study context. As validation studies expand and FDA guidance develops, this may change, but clinical digital biomarker testing is not standard of care today. —
