Digital Biomarkers for Neurodegenerative Disease: Early Detection, Monitoring & Clinical Trial Uses

Digital biomarkers are transforming how we detect and manage neurodegenerative diseases. For conditions like Parkinson’s and Alzheimer’s, which affect millions globally, symptoms appear only after years of brain damage. Wearable and mobile sensors can capture subtle signs such as changes in gait, speech, sleep, cognition, etc. long before clinical diagnosis. This early-warning capability enables prompt intervention, more accurate disease monitoring, and powerful endpoints for trials of new treatments.

---

Neurodegenerative diseases are notoriously silent in their earliest stages. Clinical Parkinson’s symptoms, for instance, emerge only after roughly 60% of dopaminergic neurons are lost¹, and Alzheimer’s pathology can accumulate decades before memory problems appear.²

Photo by Milad Fakurian on Unsplash

Digital biomarkers offer a “second eye” to catch these hidden changes. Consumer-grade technologies such as smartphones, wearables and tablets can continuously track speech, sleep, motor activity, eye movements and other behaviors. This passive and active monitoring can yield real-time digital signals like changes in gait, voice, or typing patterns, often before clinical symptoms appear.

Early detection matters because it can dramatically expand the window for intervention. With neurodegeneration happening silently, catching the disease “upstream” could allow neuroprotective treatments much earlier. At the same time, these biomarkers are highly sensitive: daily home-based measurements can reveal gradual decline or treatment effects that clinic visits often miss. In short, digital biomarkers promise to revolutionize how we screen for, track, and treat Parkinson’s and Alzheimer’s.

Key Digital Biomarkers for Neurodegenerative Disease

Neurodegenerative changes manifest across multiple domains, and modern devices can capture them. Common digital signals include:

• Gait and balance (IMU sensors): Accelerometers/gyroscopes in shoes, wearables or phones measure walking speed, stride length, postural sway and variability. Subtle gait slowing and asymmetry can appear years before diagnosis. Research indicates that gait speed has been seen to decline about 12 years before cognitive impairment.³

  
  

• Tremor and fine motor: Wrist or finger-worn sensors can measure how often (frequency) and how strongly (amplitude) tremors occur, which is important in Parkinson’s. Simple phone or tablet tapping tests also track slowed movement. A meta-analysis reveals that typing patterns can detect Parkinson’s with about 86% sensitivity and 83% specificity, as well as early cognitive decline with good accuracy.⁴ This ensures that those with disease are not missed and those free of it are not wrongly labelled.

  
  

• Voice and speech: Acoustic analysis of speech can reveal monotone, slowed speech, or articulation deficits. Early patients of Parkinson’s disease show detectable changes in pitch, loudness and articulation within just a year.⁵ Today, machine learning models using voice data can predict future dementia with ~0.81 AUC accuracy.⁶

  
  

• Cognitive activity: Apps or web games can administer memory and attention tasks repeatedly. Passive metrics like smartphone usage patterns, typing speed, and GPS mobility have also been linked to cognition. For example, reduced smartphone typing latency and more erratic phone use correlate with early cognitive decline.

  
  

• Sleep and circadian patterns (actigraphy, bed sensors): Sleep disturbances often precede neurodegeneration. Studies show that poorer sleep quality increases the risk of developing Alzheimer’s. Wearable trackers or under-mattress sensors measure sleep quality and fragmentation. Trial cohorts like RADAR-AD and IDEA-FAST even use home EEG and bed sensors to monitor sleep and fatigue.⁷

  
  

• Multimodal combinations: Most groups now favor combining signals. A single smartphone can log steps, voice, and touchscreen behavior at once. In practice, a multi-device setup (wrist accelerometers, insoles, camera, smartphone) can simultaneously monitor gait, balance, tremor, typing and more. By fusing these diverse streams, AI models can form robust digital biomarkers for subtle neurologic change.

AI generated image

Disease-Specific Digital Biomarkers

Digital Biomarkers for Parkinson’s Disease

Parkinson’s disease (PD) produces distinctive motor changes that can be easily tracked using digital tools. These include:

• Tremor and Bradykinesia: Wrist-worn accelerometers quantify the intensity and frequency of resting/postural tremors. Finger-tapping tests measure bradykinesia (slowness), and micro-switch-based keyboard tasks capture fine motor speed. In a phase 1 Parkinson’s trial, participants did daily phone-based tests for voice, tremor, tapping, balance, and walking. These sensor measures clearly distinguished people with Parkinson’s from healthy controls, and closely matched clinical rating scores. Even when neurologists couldn’t see a tremor, the phone sensors picked up subtle movement changes, showing that digital tapping and tremor tests can sensitively quantify symptom severity.⁸

  
  

• Gait and Posture: Accelerometers and gyros on ankles, shoes or hips measure walking patterns. Parkinson’s patients often exhibit shuffling gait, shorter stride, reduced arm swing and increased step variability. Both gait speed and stride length tend to decline with disease progression, while step-to-step variability worsens. Gait measures serve as valuable pharmacodynamic biomarkers of drug response.

  
  

• Speech and Voice: Voice recordings (through phone apps or smart speakers) capture Parkinson’s related speech deficits, and can help track DBS effects or early PD progression. Parkinson’s disease often causes hypophonia (quiet speech), monotone, and articulation problems. 

  
  

• Sleep: REM sleep behavior disorder and other sleep disruptions are early signs. Wearable sleep trackers or pressure mats can quantify REM interruptions and overall sleep quality. Likewise, heart rate variability (via wearable ECG) can indicate autonomic dysfunction in Parkison’s. These non-motor signals, while secondary, enrich monitoring.

Digital Biomarkers for Alzheimer’s Disease

Alzheimer’s and related dementias primarily degrade cognition, but leave imprints in behavior and everyday activities. Emerging digital biomarkers for Alzheimer’s include:

• Passive Cognitive Markers: Changes in routine phone or computer use can signal mild impairment. Overall keystroke speed and error rate tend to decline as cognitive function worsens. Likewise, reduced smartphone app usage, erratic navigation patterns (GPS showing confusion), or decreased communication might indicate growing memory problems.

  
  

• Speech and Language: Voice can flag dementia. Early Alzheimer’s often brings semantic and lexical decline. Speakers use simpler syntax and fewer unique words. Acoustic features like increased jitter have also been linked to future dementia risk in healthy adults. Routine voice monitoring could be a low-cost way to screen for Alzheimer’s disease.

  
  

• Sleep and Circadian Rhythms: Fragmented sleep and circadian disruptions often precede clinical Alzheimer’s. Passive actigraphy can detect increasing wakefulness at night or napping trends. In fact, Alzheimer’s research apps ask daily about sleep and fatigue, and some studies use at-home EEG or bed pressure sensors to quantify sleep architecture. 

  
  

• Navigation and Mobility: Spatial disorientation is an early cognitive sign. Smartphone GPS tracking can quantify how far and in what patterns a person travels. A shrinking “life space” or getting lost outside familiar areas could be an early Alzheimer’s digital marker. Similarly, subtle gait change, although not specific to Alzheimer’s, when combined with cognitive signals can improve accuracy.

  
  

• Other Behaviors: Browser history and social media usage have been explored. Early Alzheimer’s patients may spend less time on cognitively demanding tasks or have irregular online routines. These are active research areas.

Clinical Utility & Trial of Digital Biomarkers

Digital biomarkers have clear applications across the patient and research journey.

1. Early Detection and Screening

   In primary care or community settings, digital tests could flag at-risk individuals. For example, an older adult’s smartphone might quietly administer brief memory quizzes or analyze their speech during a call. Deviations from normative patterns can trigger a referral. Since motor symptoms or cognitive tests now only confirm what has been degrading for years, these home-based biomarkers act as early warning systems. Studies show that simple camera-based eye-tracking, combined with AI models, can distinguish people with Parkinson’s from healthy controls.

2. Progression Monitoring

   Once diagnosed, patients could wear devices or use apps that continuously log data. Unlike sporadic clinic visits, daily or weekly digital measures capture the trajectory of decline. For instance, subtle increases in gait variability or slowing in tapping speed over months can quantify progression. This granular tracking means clinicians might detect deterioration earlier and adjust care. It’s analogous to having a wearable “fitbit for neurology” that records every step or voice snippet.

3. Disease-Modifying Therapy (DMT) Trials

   New Alzheimer’s and Parkinson’s disease drugs aiming to slow degeneration require sensitive endpoints. Digital biomarkers can serve as surrogate endpoints if validated. For instance, studies can measure change in home-based gait speed or finger-tap latency as an outcome. Because sensors gather large data volumes, they can strengthen statistical results.

4. Remote Patient Follow-up 

   Telemedicine has grown, but digital biomarkers can take it further. After an intervention (surgery, medication change), patients can stay connected through apps. For example, an app might prompt a weekly 5-min motor test, or the phone’s inertial sensors may monitor walking distance. This continuous remote monitoring helps detect complications or non-compliance early, without burdensome clinic travel. 

5. Regulatory and Pathway Development 

   Agencies such as the FDA and EMA are beginning to recognize digital endpoints. For qualification, robust evidence linking a digital measure to clinical benefit is needed. Current efforts include validating smartphone cognitive tests against amyloid PET and papertests. As these validations accumulate, expect to see digital biomarker-based endpoints in pivotal neurodegenerative disease trials.

Validation and Practical Deployment

Introducing digital biomarkers into practice requires careful validation and attention to real-world issues. Each digital measure must be anchored to an accepted standard. Many studies therefore correlate sensor outputs with clinical scales. A digital biomarker must reliably track changes over months/years. In general, a biomarker is stronger if it can predict future disease or track changes as the disease progresses.

Clinicians will ask: what exactly is the device measuring? Understanding which features drive an AI score is important. For instance, if a voice analysis tool signals “high dementia risk,” a doctor should know whether that’s due to slowed speech rate, vocabulary changes, or acoustic tremor. Algorithm transparency and physiologically interpretable metrics are needed for clinical adoption.

Another challenge is data quality and noise. Home environments are messy. Data from at-home sensors will inevitably contain noise: dropped signals, multitasking behaviors, or housemates triggering motion sensors. User’s engagement with the technology is also critical. Fortunately, evidence suggests good feasibility when solutions are user-friendly. Introducing devices early, co-designing with users, and minimizing required interactions boost long-term use.

Additionally, non-neurological issues can affect signals. For example, arthritis might slow tapping or shuffling gait independent of PD. Eye tracking could be hindered by glasses or dry eyes. Study designs must account for these: collect baseline health data, include age-matched controls, and possibly adjust algorithms. In practice, combining multiple independent signals helps: if one channel is obscured by co-morbidity, others may still work.

Like with all digital technology, continuous monitoring raises privacy concerns. Systems must secure data and obtain informed consent, especially with sensitive signals like GPS location. Thankfully, many platforms now anonymize and aggregate data for AI. Patient groups, ethicists and other stakeholders are also actively developing guidelines.

In summary, robust digital biomarker deployment relies on thorough validation against clinical gold standards and across diverse patients, as well as practical design that addresses noise, adherence, and ethics. When done right, we gain an “always-on” clinical sensor network that significantly augments traditional testing.

The Promise of Digital Biomarkers for Neurodegenerative Diseases

Digital biomarkers are ushering in a new era for neurology. By capturing subtle, everyday signals: a spoken sentence, a step taken on an app, or a sleep pattern overnight,  we can detect and measure Alzheimer’s and Parkinson’s disease in ways never possible before. These methods complement traditional exams and can speed up diagnosis, personalize monitoring, and accelerate clinical trials. Clinical trials benefit too, replacing sporadic clinic scales with rich sensor endpoints can improve efficiency and sensitivity. While challenges like data noise and user adherence remain, these tools can work well when they are designed around users. 

These novel endpoints will likely become standard tools in neurology. Integrating validated digital biomarkers into practice and research could change the trajectory of neurodegenerative diseases for millions worldwide, catching decline when intervention is still possible and making trials more powerful.

---

References:

1. Sun YM, Wang ZY, Liang YY, Hao CW, Shi CH. Digital biomarkers for precision diagnosis and monitoring in Parkinson's disease. NPJ Digit Med. 2024 Aug 21;7(1):218. doi: 10.1038/s41746-024-01217-2. PMID: 39169258; PMCID: PMC11339454.

2. Chudzik A, Śledzianowski A, Przybyszewski AW. Machine Learning and Digital Biomarkers Can Detect Early Stages of Neurodegenerative Diseases. Sensors (Basel). 2024 Feb 29;24(5):1572. doi: 10.3390/s24051572. PMID: 38475108; PMCID: PMC10934426.

3. Kourtis LC, Regele OB, Wright JM, Jones GB. Digital biomarkers for Alzheimer's disease: the mobile/ wearable devices opportunity. NPJ Digit Med. 2019;2:9. doi: 10.1038/s41746-019-0084-2. Epub 2019 Feb 21. PMID: 31119198; PMCID: PMC6526279.

4. Alfalahi, H., Khandoker, A.H., Chowdhury, N. et al. Diagnostic accuracy of keystroke dynamics as digital biomarkers for fine motor decline in neuropsychiatric disorders: a systematic review and meta-analysis. Sci Rep 12, 7690 (2022). https://doi.org/10.1038/s41598-022-11865-7

5. Rusz J, Krack P, Tripoliti E. From prodromal stages to clinical trials: The promise of digital speech biomarkers in Parkinson's disease. Neurosci Biobehav Rev. 2024 Dec;167:105922. doi: 10.1016/j.neubiorev.2024.105922. Epub 2024 Oct 18. PMID: 39424108.

6. Lin H, Karjadi C, Ang TFA, Prajakta J, McManus C, Alhanai TW, Glass J, Au R. Identification of digital voice biomarkers for cognitive health. Explor Med. 2020;1:406-417. doi: 10.37349/emed.2020.00028. Epub 2020 Dec 31. PMID: 33665648; PMCID: PMC7929495.

7. Brem AK, Kuruppu S, de Boer C, Muurling M, Diaz-Ponce A, Gove D, Curcic J, Pilotto A, Ng WF, Cummins N, Malzbender K, Nies VJM, Erdemli G, Graeber J, Narayan VA, Rochester L, Maetzler W, Aarsland D. Digital endpoints in clinical trials of Alzheimer's disease and other neurodegenerative diseases: challenges and opportunities. Front Neurol. 2023 Jun 15;14:1210974. doi: 10.3389/fneur.2023.1210974. PMID: 37435159; PMCID: PMC10332162.

8. Lipsmeier F, Taylor KI, Kilchenmann T, Wolf D, Scotland A, Schjodt-Eriksen J, Cheng WY, Fernandez-Garcia I, Siebourg-Polster J, Jin L, Soto J, Verselis L, Boess F, Koller M, Grundman M, Monsch AU, Postuma RB, Ghosh A, Kremer T, Czech C, Gossens C, Lindemann M. Evaluation of smartphone-based testing to generate exploratory outcome measures in a phase 1 Parkinson's disease clinical trial. Mov Disord. 2018 Aug;33(8):1287-1297. doi: 10.1002/mds.27376. Epub 2018 Apr 27. PMID: 29701258; PMCID: PMC6175318.