Dr. Diego Guarin is an Assistant Professor in the Department of Applied Physiology & Kinesiology of the University of Florida’s College of Health and Human Performance. A research and data scientist who, with a team of researchers, developed VisionMD, an open-source video-based analysis tool with the use of a smartphone camera. The software performs an analysis of a person’s motion, speed, and acceleration to diagnose symptoms of movement disorders like Parkinson’s without the need for specialized hardware or cloud-based processing.
Since the debut of VisionMD, the research group led by Dr. Guarin has continued to explore the capabilities of diagnostic software and your phone’s camera. They are discovering new ways to diagnose Parkinson’s using modern mobile processors and crafting technology to develop more affordable and accessible tools for patient care.
Patient Diagnosis
For decades, Parkinson’s disease has been diagnosed in the same way: a doctor watches a patient as they move around and perform tests by tapping their fingers, opening and closing their hands, or holding their arms out in front of their bodies. While the doctor looked for signs of slowness, stiffness, tremor, and smaller movements. It was the use of these observations that was formalized into a rating scale called the Unified Parkinson’s Disease Rating Scale (UPDRS), a cornerstone of what makes up Parkinson’s care today.
Current Standards Work, but Can be Improved.
Parkinson’s disease is not something that appears overnight, and symptoms don’t immediately become obvious after the neurologist diagnoses you with the disease. Parkinson’s progresses over a period of years, developing slowly. A growing body of evidence suggests that long before Parkinson’s can be seen by a tremor in the hand, the disease can already be measured. Unlike many expensive medical devices and new technologies, something old is helping to bring those early bodily changes to light for patients: the use of video.
Parkinson’s Before Diagnosis
Neurologists believe one of the earliest predictors of Parkinson’s and other neurological disorders comes from people with idiopathic REM sleep behavior disorder, or iRBD. These individuals physically act out their dreams during REM sleep, sometimes kicking, punching, or shouting, years before any movement disorder is diagnosed. More than 80% percent of people diagnosed with iRBD will eventually develop Parkinson’s disease, dementia with Lewy bodies, or Multiple System Atrophy. Even though people with iRBD don’t show any Parkinson ’s-related symptoms during a visit to their doctor or a clinic.
To find ways of being able to predict early signs of the disease, researchers turned to video recording and machine learning to help devise methods to help identify changes in the way a person’s body moves for those indicators.
Click to enlarge
In a 2025 study published in npj Parkinson’s Disease, Dr. Guarin and his team detailed their findings. Videos of a simple finger-tapping task performed by healthy adults, people with iRBD, and patients with early Parkinson’s disease were analyzed by computer using machine-learning (ML) algorithms and markerless hand tracking. The analysis included only videos that had already been scored as completely normal, a zero on the finger-tapping item of the UPDRS, with no visible bradykinesia or hesitation.
Using these videos, the system measured movement speed, movement size, and how those movements changed over time.
What was detected by the ML was undetectable by conventional means. While people with Parkinson’s already showed measurable bradykinesia (slower and smaller movements), even though their exams appeared normal. People with iRBD did not show classic bradykinesia, but they demonstrated something else: their movements gradually declined in speed and size as the task went on, A phenomenon known as the sequence effect, in which a patient’s movements gradually become smaller with repetition, growing more abnormal with each movement made. The findings suggest Parkinson’s is affecting the way the body moves long before it’s visually obvious, making the sequence effect a very early diagnostic tool for neurologists to use during examinations expose a longstanding limitation in Parkinson’s care, heavy patient observation.
What these studies indicate is that exams of patients are subjective and at the discretion of the attending neurologists, as is the case when measuring tremor. Neurologists will rate tremors by watching a patient and assigning a score. Depending on the severity of the tremor, two neurologist could rate the tremor differently based on their experience and judgment, and very small tremors could be missed altogether. While wearable sensors are available, for some, the use of the technology can be impractical and generally not used during routine medical exams.
Tremor Diagnosis and Video
In a second published research paper, Dr. Guarin explored whether a tremor could be measured objectively using standard smartphone video alone, without any attached sensors, rulers, or the use of external calibration tools.
They developed a system that uses the human iris as a built-in reference point and combined it by estimating the hand’s position in three-dimensional space and tracking it frame by frame. This allows the software to translate on-screen pixel movement into real-world measurements.
These results made strong comparisons between video-based measurements and manually reviewed data marked hand positions by looking frame by frame through the video to serve as a reference, along with a close correlation to clinical tremor scores.
But the most revealing finding came from patients clinicians had rated as having no tremor at all. The algorithm detected small, rhythmic oscillations that were real but subtle enough that they could have been overlooked during a standard exam. The implications of this data imply that the absence of a visible tremor does not necessarily mean the absence of any tremor. It just means the tremor may simply be below the threshold of human perception.
The inclusion of video for Deep Brain Stimulation
Deep Brain Stimulation (DBS) is one of the most effective therapies for Parkinson’s patients. DBS and aDBS are tools that can be used to modify signals in the brain using implanted electrodes and dramatically reduce tremor, stiffness, and slowness through the adjustments, though trial and error with the physician and the patient taking months to find the right setting.
In a 2026 study published in Brain Stimulation, video was again the subject and tool of researchers to see whether smartphone video could bring objectivity to DBS programming. Patients were recorded performing standardized hand movements with a smartphone during DBS programming sessions. The software identified more than 20 quantitative (measurable movements) features using markerless motion analysis to evaluate speed, rhythm, consistency, and movement decline. They were then combined into a single score that ranked DBS settings based on improved movement. The results were that across every patient studied, a clear optimal setting was achieved, resulting in motor improvements that closely mimicked treatment by levodopa. A faster, more stable adjustment that showed less decline with repeated movements, giving patients and neurologists a more precise way dial in on the best setting for them.
This meant that for the first time, DBS programming could be guided by data rather than a patient’s feedback.
A Different Way of Seeing Parkinson’s
The use of smartphone video has shown its importance in being able to analyze problems facing Parkinson’s Disease, and potentially other movement disorders. As the processing power of smartphone chips continues to increase each year, these devices are becoming capable of handling more sophisticated analysis directly on the device or in real time, providing real benefits for diagnosis and treatment to some of the toughest problems we have yet to resolve, and helping us better understand Parkinson’s disease and its effects on the body.
This also has the potential to move into areas of the world that are underserved because of geography, shortage of neurologists, medical personnel, and government funding. Because the technology relies on ordinary video using smartphones and webcams, it is scalable and accessible, making it compatible with telemedicine and remote monitoring. Giving patients treatment options for those who are unable to travel and for new diagnoses. It does not replace clinical expertise, but it adds a layer of objectivity where the human eye reaches its limits. This technology doesn’t replace doctors, but gives them the tools they need to treat patients in the office, at home, and from a distance.
