Understanding AI: A Powerful Tool in the Fight Against Parkinson’s Disease
Artificial Intelligence (AI) is all around us and becoming more and more ubiquitous in many aspects of our lives, including Parkinson’s disease (PD) research and clinical care. But what exactly is AI, and how is it accelerating PD research and improving how doctors care for patients?
AI refers to computer systems capable of performing tasks that typically require human intelligence. This includes functions like learning, problem-solving, pattern recognition, and decision-making. AI aims to simulate human intelligence by using algorithms, data, and computational power. Machine learning (ML) refers to a specific subset of AI that focuses on enabling systems to learn from massive amounts of data.
Five Ways AI is Impacting Parkinson’s Research:
Artificial intelligence is significantly impacting research by opening up new possibilities in the diagnosis, monitoring, and treatment of PD. The use of AI does not come without caution and concern, and it must be used wisely and responsibly; however, the current and potential benefits are quite significant and worth noting.
Early Diagnosis and Biomarker Discovery
AI could aid in early detection of PD by identifying patterns in patient data that might be otherwise overlooked. Any quantifiable difference between people with PD and people without PD could be used as a biomarker, a measurable characteristic in the body that indicates that disease is present. Biomarkers for PD are essential not only for diagnosis but also for disease monitoring, as well as for ensuring that the most appropriate people are enrolled in clinical trials. AI can process more data faster and more thoroughly, discovering minute discrepancies between groups more accurately than human analysis alone.
Machine learning techniques could be used on a variety of types of datasets to mine them for potential biomarkers. These include MRI data, genetic data, and digital signatures from movement and speech data.
Tracking Disease Progression and Severity Assessment
By analyzing video recordings of patients’ movements, AI can objectively assess the severity of Parkinson’s symptoms, providing a more accurate and consistent way to track disease progression. This can be done with video-based gait analysis and pose estimation, a computer vision technique that uses AI to identify the movement of key points on a person’s body (like joints) to determine their posture or pose. These techniques lead to more consistent, objective severity scoring and symptom tracking over time. Wearable devices can capture this data and monitor changes in movement, balance, and other symptoms to assess the disease and its progression. Unlike periodic clinical visits, wearable devices generate continuous data streams which could provide a more complete and accurate picture of a patient’s daily experience, enabling more responsive treatment adjustments.
Personalized Treatment Plans
AI can analyze a plethora of individual patient data to develop personalized treatment plans, tailoring medication dosages or therapies based on their specific needs. By utilizing patient-specific data such as age, genetics, symptoms, and medical history, AI can help clinicians personalize treatments that are more likely to work. Rather than generic standard of care treatments, AI could allow clinicians to optimize dosage, combine therapies more effectively, and minimize side effects based on individual risk factors and disease characteristics.
AI-Aided Drug Discovery
AI can be used to analyze vast amounts of molecular data to identify potential drug targets for PD. The molecular structure of a relevant drug target can be very complex. It requires tedious work to delineate which compounds might be most effective in interacting with Parkinson’s-related proteins. AI can dramatically shorten the drug discovery timeline and reduce the cost and failure rate of experimental treatments. In essence, the drug development timeline would also be expedited with the use of AI to get promising drug candidates to the clinic faster as well.
Deep Brain Stimulation Optimization
Deep brain stimulation (DBS) is a surgical treatment method in which electrodes are implanted in specific areas of the brain to deliver electrical impulses that help regulate abnormal brain activity. This treatment can reduce symptoms like tremor, rigidity, and bradykinesia (slowness of movement). While this treatment option is very effective for some patients, the success of DBS depends heavily on fine-tuning the stimulation parameters. AI can significantly enhance DBS by analyzing patient data and automatically suggesting optimal settings based on symptom patterns and brain activity. This new approach to DBS is referred to as adaptive DBS (aDBS) and is currently available. Machine learning algorithms can even adapt in real-time, adjusting stimulation to match the patient’s needs throughout the day, improving both symptom control and quality of life.
Key Ways AI is Impacting Parkinson’s at the Doctor’s Office:
AI is also beginning to improve technology and patient care within the clinic. These technologies are streamlining clinical workflows, improving the accuracy of care, and offering Parkinson’s patients a more personalized and supportive experience at the doctor’s office. The oversight and input from healthcare professionals is absolutely critical; however, the progress and efficiency AI is providing is a game-changer.
Generating Medical Notes
AI-powered documentation tools can listen to conversations between doctors and patients and then generate structured clinical notes in real time. This not only reduces the administrative burden on clinicians but also improves the accuracy and completeness of records. This can lead to more thorough discussions in subsequent visits along with enhanced patient care.
Answering Routine Questions and Calls
AI-based virtual assistants are increasingly being integrated into clinics and patient portals to handle common tasks and routine calls. This technology can help update appointments, manage medical reminders, and answer simple questions. These systems free up staff time while ensuring that patients receive timely and consistent support, even outside business hours.
Remote Monitoring
AI-powered wearable devices can monitor movement patterns and other symptoms in real-time, allowing for remote patient monitoring and timely adjustment to treatment plans. These devices can be paired with mobile apps capable of continuously tracking symptoms like tremors, bradykinesia, or sleep disturbances. The health care team can remotely review this data and adjust medications or therapies without requiring continuous in-person visits. This not only enhances patient convenience but also allows for quicker intervention when symptoms begin to shift.
Risk Alerts
AI can analyze data from wearable devices, electronic health records, and symptom notes to identify patterns that may signal a decline in a patient’s condition. Automatically flagging high-risk patients can help clinicians prioritize care and act immediately when needed to prevent further complications.
Customized Health Education
Various AI algorithms can tailor educational content like videos, articles, and interactive tools to match a patient’s specific diagnosis and literacy level. This personalization helps patients be more informed about their conditions and take an active role in managing their health when possible.
Overall, AI holds great promise for advancing research and clinical care in PD by providing more accurate diagnoses, better disease monitoring, and potential for developing new treatment strategies, ultimately improving the quality of life for patients.
How APDA is supporting AI in PD Research and the Clinic:
The American Parkinson Disease Association (APDA) is supporting cutting-edge AI research across several institutions through the APDA-funded projects below:
- Dr. Mata from the Cleveland Clinic Foundation uses machine learning to predict who will develop dyskinesias. Learn more about his research here: https://www.apdaparkinson.org/research/investigators/ignacio-mata-phd
- Dr. Ning from The Regents of the University of Minnesota, Twin Cities, uses machine learning to analyze brain networks in PD and understand why disruption of movement and cognition occurs in PD. Learn more about her research here: https://www.apdaparkinson.org/research/investigators/yuxiao-ning-phd
- Dr. Roemmich, from the Hugo W. Moser Research Institute at Kennedy Krieger, Baltimore, Maryland, uses pose estimation, which is an AI tool that models where joints are in space. The goal is to create a method of monitoring motor symptoms of PD via video visits. Learn more about his research here: https://www.apdaparkinson.org/research/investigators/ryan-roemmich-phd
Article published from this work: https://pmc.ncbi.nlm.nih.gov/articles/PMC10965062/pdf/pdig.0000467.pdf
- Dr. Guan from the University of Michigan, Ann Arbor, Michigan, uses machine learning to predict PD based on voice recordings. Learn more about her research here: https://www.apdaparkinson.org/research/investigators/yuanfang-guan
Article published from this work: https://pmc.ncbi.nlm.nih.gov/articles/PMC8763910/pdf/42003_2022_Article_3002.pdf
We extend our thanks to Clark Jones, PhD, and Lianna Badran for their contributions to this blog.
Tips & Takeaways
- Artificial intelligence holds great promise in improving PD research and clinical care
- AI can potentially impact and accelerate biomarker discovery and treatment options
- AI is already present in the clinic, improving remote monitoring of PD symptoms and DBS adjustments
- APDA has funded numerous research projects that utilize the amazing potential of AI. To learn more about the research we fund, visit the Research section of our website.