Exploring the use of Telehealth for Parkinson Disease and Dementia Monitoring: Kinan Muhammed, MD, PhD

Telehealth is a key part of neurologic care, expanding rapidly during the COVID-19 pandemic and remaining widely used for conditions like Parkinson disease (PD), epilepsy, and stroke follow-up.¹ It can help patients access specialists and allow more frequent monitoring, which is especially valuable for patients with chronic or mobility-limiting conditions.² However, challenges persist, including limited ability to perform full neurologic exams remotely, disparities in digital access, and ongoing reimbursement and regulatory uncertainty.¹ As remote monitoring tools and digital assessments continue to evolve, telehealth is helping shift neurology toward a more hybrid, data-driven care model.²

Emerging technologies and the use of telehealth were two of the main points of emphasis at the 2026 American Academy of Neurology (AAN) Annual Meeting, held April 18 to 22 in Chicago, Illinois. During the meeting, Kinan Muhammed, MD, PhD, consultant neurologist and co-founder and Chief Medical Officer for Kneu health, gave a talk on a new smartphone application aimed at monitoring the symptoms of patients with PD and dementia. Expanding on these themes beyond the meeting, Dr. Muhammed spoke with NeurologyLive^® about the real-world impact of this technology.

In this conversation, he outlines the platform’s decade-long research foundation and its use of active motor and cognitive tests to generate objective, longitudinal data that integrate directly into clinical workflows. Dr. Muhammed highlights the technology’s ability to close the six-month “snapshot” gap between in-person visits, support medication optimization and advanced therapy timing, and enable new models of proactive, remote care across health systems. Furthermore, he detailed real-world outcomes that can help clinicians stratify risk, personalize interventions, and ultimately transform the future delivery of neurodegenerative care.

NeurologyLive: Can you provide a general overview of the smartphone application and how it works?

Kinan Muhammed, MD, PhD: The application is based on a decade's worth of academic research. We've probably built one of the world's largest longitudinal cohorts of patients with PD who have been studied using the digital platform. It's made up of three main components. A patient facing software as a medical device that uses the sensors on the patient’s smartphone to monitor a range of objective measures of motor and cognitive function, as well as other measures including blood pressure, sleep and medication adherence.

The sensor data then goes in real time to an analytics platform that we developed, and we convert those outputs into clinical scores that are based on algorithms and machine learning models we developed at Oxford University. We can then provide clinically meaningful metrics that are presented on a clinician dashboard, which is integrated into the electronic health records. Clinicians can use this in real time to start managing patients in more objective ways. Essentially, it allows remote and objective monitoring of patients with PD.

As for the measures themselves, they are formed from seven active motor tests. The first is a sustained phonation test, so the patient says “ahh” into the smartphone, and we can extract various metrics of voice. Next is a balance test where the phone is put in the patient's pocket, they stand upright, and the accelerometer and gyroscope objectively measure the degree of balance and postural sway.

Then there's a walking test, where the phone goes in the pocket again, the person walks forward, turns around and walks back and we can measure parameters of their gait. There's an alternating finger tapping test, where they use their index and middle finger to alternate tapping on the screen of the phone. We can look at the rate, rhythm and breaks in taps and compare the left side to the right side.

We then have a reaction time test with a press and a release component, and finally a rest and a postural tremor test, where they hold the phone in the hand, and the accelerometer will measure the degree of tremor both when the hand is at rest and when it's outstretched in front of them. Again, we can compare the left side and the right side. In addition to the motor tests, there are also a battery of cognitive tests that look at different cognitive domains. We monitor processing speed, immediate and delayed recall and visuospatial memory. We've also got executive function measurements that can be captured using the phone objectively.

In summary, all of those elements plus questionnaires, patient recorded outcome measures, postural blood pressure recordings, and sleep diaries get processed and are displayed to the clinicians, who then use them in the real world to make decisions.

What gaps in the current PD diagnosis monitoring does this technology aim to address?

There's multiple challenges in the PD and the neurodegenerative space, and it's a growing problem. The issues get more and more complex as the disease progresses and conditions like PD are incredibly heterogeneous, no two patients are exactly the same.

Access is a big problem because of the growing prevalence of the condition. There aren’t enough movement disorders specialists in the US to manage a million PD patients, so access is a very big issue. Additionally, medication management is highly subjective, it's very much trial and error. When you see a patient, clinicians will take a history and examine them and recommend treatment. They then review again in a few months to see if it had the intended effect, which isn't the optimal way of managing patients.

In terms of the gaps that this platform addresses from a monitoring perspective, the current status quo is snapshots every four to six months. When clinicians see a patient in the clinic, they have a short amount of time with them. They may make some changes, and then see them again six months later, so they don't know exactly what's going on in between. Many things can happen, including deterioration and worsening of signs or symptoms, which could mean patients end up in hospital or the emergency department.

This is where the longitudinal monitoring of the platform comes into play. It can allow clinicians to intervene early and proactively to stop these problems from happening. Some of the outcomes that I recently presented at AAN 2026 were showing how those interventions made a difference. Using these digital tools, clinicians can objectively see the degree of response to various drugs. We've got data also on treatments including deep brain stimulation, so you can really start to objectively monitor how a therapy is working and whether you should change it, and even how you should change it. That gives clinicians more confidence in management around medication.

There's also gaps in identifying patients at the right time for certain therapies. Identifying people in the right therapeutic window for treatments like deep brain stimulation, foslevodopa–foscarbidopa subcutaneous infusions and other device-assisted therapies can be challenging. We found from the work that we've been doing that we can identify people who would be suitable for those kinds of interventions from the combination of digitally captured measures.

Finally, during consultations, a lot of the time is spent trying to gather relevant information from the history, examination and family members. The platform basically captures all of this in advance, and then when you do see the patient in the clinic, you can be much more focused and spend your time on those clinical meaningful decisions that you need to take in partnership with the patient, rather than spending time trying to gather subjective information. So again, it helps facilitate that too.

What were the key findings of the study, and how might they impact care for patients with PD?

The findings I presented at AAN were based on a real-world evaluation. The platform is FDA cleared and actually live and running in clinical settings, so we’re evaluating its real-world impact in a clinical environment.

There are a few key points. First, we wanted to make sure we validated the tests themselves. We looked at more than 500 patients and linked their Movement Disorder Society - Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) scores with the digital tests. The MDS-UPDRS is essentially the gold standard for measuring the degree of impairment in a patient. We took those scores and found that the individual motor component scores correlated significantly with the individual digital tests that I described earlier, so we knew we were mapping to the correct clinical features.

We then deployed the platform in a real-world setting across 11 hospitals in the UK and 2 providers in the US, on the East and West coasts. Patients used the platform anywhere from 6 to 24 months, depending on when each site came on board. Over that time, more than 1 million digital measures were captured, across more than 900 patients, which really shows how scalable the platform is.

In terms of outcomes, starting with the patient perspective, the average age was about 70 years, and 93% of patients reported that it was easy to use. From a usability standpoint, that was very encouraging. We also found month-on-month increases in patient empowerment, confidence, and knowledge. Some of the information is fed back directly to patients in the app, which allows them to be more proactive in managing their condition themselves, and that was a really satisfying finding.

From a clinical impact perspective, we found that the platform could facilitate new models of care. In the US, instead of relying solely on snapshots every 6 months, nurse practitioners and physician assistants were able to call patients on a monthly basis and check in using the data, facilitating more frequent remote care. Because of that, more than half of patients had their medications optimized based on the data.

In addition, two-thirds of clinicians reported that they could understand disease progression or treatment response faster than the current standard of care, because the data was readily available to them. There were also efficiency gains. We saw about a 20% improvement in clinic efficiency because everything was prepopulated and accessible to clinicians.

In the UK, we conducted a sub-study at one site looking at emergency department admissions. We found a reduction in ED visits among patients using the platform, down by 1.2%, compared with a national increase across the UK. That suggests that more proactive management was leading to better outcomes.

The final piece I presented at AAN was the ability to digitally phenotype patients. What that means is that we can take the objective motor data collected by the platform and stratify patients into different groups or categories. We identified three main groups: a motor-preserved group, where patients were generally doing well across measures; a tremor-dominant group, where rest and postural tremor were the main issues; and a coordination-impaired group, where finger dexterity, reaction time, and gait were most affected.

We then looked at outcomes associated with these groups. Those in the coordination-impaired group were more likely to fall, more likely to have hospital admissions, had worse quality of life, and worse cognitive outcomes overall. This knowledge allows clinicians to use the data proactively. For example, intervening earlier with physiotherapy, bone protection medication, or medication adjustments for patients at higher risk of falls. Overall, it enables a much more proactive approach to care.

What are some of the limitations of using smartphone-based technology for disease tracking?

One of the original concerns we had was whether patients, in an older population who are more likely to have these neurodegenerative conditions, would have access to smartphones and the internet required to use the platform. What we found was that fewer than 5% of patients didn’t have access to the technology. We chose smartphones because they’re already widely available and in most people’s pockets. As time goes on, this becomes even less of a limitation, because people are increasingly familiar with the technology as they grow older. While we initially thought this might be a barrier, it really hasn’t been.

The other concern was that these are active tests, meaning patients need to engage with the app to generate data, rather than passive monitoring. We chose active tests for a few reasons. First, they allow us to replicate a more controlled, clinic-like environment. With passive monitoring, you can’t always distinguish between real physiologic signals and external factors, like being on a train or a bus. Active tests are more controlled and allow for more precise measurements. They also allow us to compare different sides of the body and capture more granular data across domains like tremor, voice, and gait. They’re more sensitive, but they do require engagement.

To address that, we spent a lot of time on user testing to optimize the experience. Ultimately, we found a 60% weekly active usage rate, which shows strong engagement. I think that’s driven by the fact that patients receive useful feedback from the app, and they also know their clinical team is using the data to inform care decisions. While these were initial concerns, they haven’t proven to be major limitations in practice.

With the app also expanding into dementia symptom monitoring, how might it be adapted for those populations, and what would clinical use look like?

In the UK, we’ve already started expanding the platform into dementia, particularly in patients with mild cognitive impairment. The platform includes cognitive assessments that monitor different domains over time. In primary care, when there’s concern about cognitive decline and a patient is referred to secondary care, they can use the platform during that waiting period. This allows for triaging and stratification, helping identify which patients may need to be seen sooner versus those with more subjective concerns.

In the US, this could be especially relevant for newer therapies, such as amyloid-targeting treatments. Early identification of appropriate patients is important, because delays in treatment can limit benefit. The platform could help identify candidates earlier and also support monitoring once treatment begins, including tracking side effects and response.

From a practical standpoint, we also have a caregiver app that complements the patient app. It allows caregivers to provide additional information, prompt medication adherence, and communicate concerns with clinical teams. It supports both longitudinal tracking and day-to-day management for patients and caregivers.

What could implementation of this type of technology mean for the future of clinical care in neurology?

I think it could really change the model of care. With the increasing prevalence of these conditions, there simply aren’t enough specialists to meet demand, and access is going to become more challenging. Technology like this allows for closer monitoring and more proactive care. It enables more personalized management and improves access for a growing patient population.

One of the limitations of traditional telehealth is that you can’t physically examine the patient. These types of tools provide objective measures that can act as a substitute for parts of the neurological exam. When combined with clinical history, that allows clinicians to assess patients more meaningfully and at scale.

It also allows clinicians to prioritize care based on need, using objective data. That has implications for efficiency, resource allocation, and overall healthcare costs. Ultimately, I think this kind of technology has the potential to transform how we monitor and manage neurological diseases, allowing us to care for more patients in a more effective and proactive way.

For more PD coverage, click here.

REFERENCES
1. Wechsler LR, Demaerschalk BM, Schwamm LH, et al. Telemedicine quality and outcomes in stroke: a scientific statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2017;48(1):e3-e25. doi:10.1161/STR.0000000000000114
2. Dorsey ER, Topol EJ. State of telehealth. N Engl J Med. 2020;383(2):154-161. doi:10.1056/NEJMra2003539