Artificial intelligence provides new ways to track and manage Parkinson's

The 7th World Parkinson Congress in Phoenix, Arizona, highlighted how rapidly artificial intelligence is entering healthcare. Dr William J. Marks, Jr., a neurology professor from Stanford University, explained that Parkinson's is a condition defined by patterns, such as fluctuating symptoms and changing medication responses. Because machine learning excels at spotting complex patterns over time, this technology is a natural fit for improving care. The way AI is used varies completely depending on whether you are a scientist, a doctor, or someone living with the condition. Splitting these uses reveals how the technology assists at every level while keeping human expertise at the centre. How researchers use AI to uncover hidden patterns For scientists, AI serves as an incredibly powerful tool for analysing massive amounts of data to discover trends that the human eye simply cannot see. Spotting invisible signals: In one study, researchers used machine learning to analyse second-by-second data from wrist-worn sensors. The AI discovered that periods of non-movement or non-walking increase over time in people with Parkinson's compared to healthy controls. This type of signal is extracted from vast datasets and would remain completely invisible during traditional clinical observations. Improving diagnostic accuracy: Researchers are training AI models on standard MRI scans paired with expert clinical diagnoses. While AI cannot definitively diagnose an individual yet, these algorithms are learning to predict with reasonable confidence whether a person has typical Parkinson's or an atypical Parkinsonism syndrome, such as multiple system atrophy or progressive supranuclear palsy. Developing explainable models: Scientists are actively working to solve the black box problem, which is the current inability to explain exactly why an AI model produces a specific output. Researchers are also focused on generalizability, ensuring that AI models trained on one specific group of people work just as well for individuals with different demographics or milder symptoms. How clinicians and doctors use AI to enhance care For neurologists and general practitioners, AI acts as a digital assistant that enhances clinical judgment rather than replacing it. Analysing life between appointments: Doctors usually only see a snapshot of a person during a brief clinic visit, often when medication is working well. AI platforms can continuously process data from wearable sensors, app-based motor tasks, and patient-logged medication timings. This alerts doctors to hidden issues like medication wearing-off between appointments. Guiding advanced therapies: By surfacing long-term trends that standard electronic health records often miss, AI helps specialists see exactly what a typical day looks like. This objective data helps doctors decide if medication adjustments are needed, or if advanced options like deep brain stimulation, focused ultrasound, or pump-based therapies are warranted. Supporting general practitioners: There is a global shortage of movement disorder specialists. AI tools help diffuse expert clinical knowledge to primary care physicians and non-specialist clinicians, allowing them to access reliable, organized information to better support their patients. The clinician in the loop: Regulatory bodies like the FDA approve AI tools strictly as adjuncts to care, such as making kinematic measurements to aid a doctor. AI is not a standalone diagnostic tool, and clinical expertise must always remain in the loop to guide medical decisions. How people with Parkinson's use AI for daily management People living with Parkinson's and their care partners are already using both general and specialised AI to take control of their daily health. Appointment preparation and education: Preparing for a neurology visit can be overwhelming, and key questions are easily forgotten. People use general-purpose AI platforms to generate organised lists of questions, explain what a specific symptom means, and outline relevant lifestyle changes. Conversational data querying: Specialised, fit-for-purpose Parkinson's platforms are being designed with patient input. These platforms use large language models so people can ask questions in plain English, such as asking why they feel unwell in the morning, and receive an analysis based on their own wearable data. Objective symptom tracking: Built-in tools allow passive monitoring of overall activity levels alongside the active detection of tremors and involuntary movements throughout the day. This helps individuals understand how their bodies respond to medication timing. A note of caution on trustworthiness While there is great optimism, healthy caution is necessary. A striking case study published in Nature showed how a researcher in Sweden fabricated a non-existent eye condition called bixonimania, creating fake papers and AI images. Within weeks, major AI systems picked up the fake condition and fed it back to users as fact. This highlights that general AI models can surface inaccurate information. It is vital to triangulate any AI-generated facts with authoritative medical sources and always discuss insights directly with your healthcare team. The true value of AI lies in enhancing the vital conversation between you and your doctor, linking digital data directly to your lived experience.