Prof Michael Okun: My Take
Can we trust AI chatbots for Parkinson’s disease information? Generative AI chatbots use artificial intelligence to create conversational answers to questions, but in health care a convincing answer is not necessarily a correct or safe answer. Lange and colleagues describe in a new paper in The Lancet Regional Health Europe what happened when a Parkinson’s-specific AI chatbot called jAImes was evaluated during real-world use. Key points: - The chatbot handled 2,035 conversations and 6,146 messages over 129 days, and automated screening classified 88.6% of conversations as good. - Human review uncovered clinically important errors that automated monitoring missed, including incorrect medication information and a failure to appropriately escalate suicidal ideation. - The authors propose CARE-LLM, a monitoring framework combining automated screening, expert human review, random sampling and a feedback loop to identify and correct failures. My take: AI is coming to Parkinson’s care, whether we are ready or not. The exciting part is its potential to put useful information into the hands of folks and families 24/7. The sobering part is that an AI answer can sound terrific and still be dangerously wrong. This study reinforces for me that we should not simply ask whether an AI performs well. We should ask what happens when it fails, whether we can detect the failure and how quickly a human can get into the loop. Here are 5 points that resonated w/ me: 1- A Parkinson’s-specific AI grounded in curated information can perform well across thousands of real-world conversations, but well is not the same as perfectly safe. 2- Automated AI monitoring can itself miss important AI mistakes, which is why human expert review remains critical. 3- Medication advice and mental health emergencies are areas where getting an answer almost right may not be good enough. 4- AI systems used in health care need surveillance after deployment, not simply testing before launch. 5- The future should not be AI versus health care providers. It should be carefully designed AI w/ health care providers, rigorous monitoring and the person w/ disease remaining at the center. https://www.thelancet.com/
Team Fox Detroit gala celebrated 10 years last night and an honor to unbox and give away the first 300 Parkinson's Plate books. Team Fox is creating impact and has raised 1.6 million dollars for Parkinson's research! Donna and Mike are knocking Parkinson's out of the park. https://us.amazon.com/Parkinsons-Plate-Health-Manage-Journey/dp/077880741X
The Team Fox Gala in Detroit celebrated 10 years of impact and 1.6 million dollars of fundraising for the Michael J Fox foundation last night. The event also featured the first 300+ Parkinson's Plate books freshly printed, unboxed and delivered from the shipyard. https://us.amazon.com/Parkinsons-Plate-Health-Manage-Journey/dp/077880741X
What resonated with me from the final session of DBS Think Tank XIV on digital health-guided neuromodulation was how rapidly digital tools are beginning to connect the entire therapeutic journey, from predicting and planning to programming and long-term monitoring. Alireza Gharabaghi showed a compelling vision for Digital DBS, using data and machine learning to predict outcomes, refine surgical targets and trajectories, and enable physiology-informed and remote programming. Diego Guarin showed how something as accessible as a smartphone video can be transformed through computer vision into quantitative measures of bradykinesia and other motor features, potentially making DBS programming more objective, scalable and available beyond expert centers. Vibhor Krishna extended this concept with automated motion-sensor functional mapping, which can reduce the number of settings clinicians need to test, while his broader work in advanced imaging, tractography and focused ultrasound reminds us that digital health is not simply about wearables or apps; imaging and data can help us choose the right patient, identify the right circuit, target it more precisely and objectively measure what happens afterward. It was a fitting way to close the Think Tank: the future of neuromodulation may be increasingly digital, objective, personalized and accessible, but its success will still be measured by whether we use these tools to improve the lives of the people we serve.
What resonated with me from this morning’s DBS Think Tank session on advanced imaging in neuromodulation was how imaging is rapidly evolving from a picture of anatomy into a roadmap of the circuits we want to understand and modulate. Nico Dosenbach took us into the somato-cognitive action network (SCAN) and the provocative idea that Parkinson’s may involve abnormal SCAN-subcortical connectivity, with successful therapies including DBS appearing to normalize aspects of this network. Erik Middlebrooks showed how sequences such as FGATIR can reveal structures and boundaries that conventional MRI may miss, moving us toward direct visualization and increasingly individualized DBS targeting. Mallory Hacker then tackled one of our field’s most difficult questions: as we explore DBS earlier in Parkinson’s disease, can sophisticated imaging and connectivity analyses help us separate symptomatic neuromodulation from true disease modification, and perhaps identify the fiber pathways associated with progression? What struck me is that the next generation of imaging may help us move beyond asking where is the electrode? Should we be asking what network did we engage, how did we change it, and did we alter the trajectory of disease?
One of my favorite moments from this morning’s DBS Think Tank came from Nico Dosenbach, who reminded us of the famous story of Charles Steinmetz and Henry Ford. When Ford’s machinery failed, Steinmetz studied the problem, made a chalk “X” on exactly the right spot and told the engineers where to make the repair. His famously large bill was essentially: a small amount for making the chalk mark and the rest for knowing where to put the X. What a perfect metaphor for modern neuromodulation. Implanting a DBS lead is one thing; understanding the circuits, networks and connectivity well enough to know exactly where to put the X is where the real magic lives.
What resonated with me from our DBS Think Tank session on the Ethics of Neuromodulation was the reminder that advancing technology also means advancing our responsibility to the people receiving it. Gabriel Lázaro-Muñoz from MGH challenged us to think about human fidelity as neurotechnology moves from carefully controlled studies into the real world, including what happens to patients over the lifetime of an implanted device. Joe Fins from Weill Cornell reminded us of the ethical problem of abandonment and of the power of listening to patients through qualitative interviews to understand benefits and risks that our traditional outcome scales may miss. Importantly, empirical work is also helping us move beyond the often-repeated fear that DBS fundamentally changes who a person is; patient studies have generally not supported a narrative of substantial adverse personality change and, in some cases, patients describe feeling closer to themselves again. Cynthia Kubu from Case and formerly Cleveland Clinic brought this home through years of empirical neuroethics research: we need to measure what matters to patients, follow their goals as they evolve, and examine autonomy, control, personality and quality of life rather than focusing only on motor scores. Finally, perhaps our greatest ethical challenge is that too many people who could benefit still never reach DBS. Maybe part of changing that begins with changing our language. “Deep brain stimulation” can sound frightening; “brain pacemaker” is simpler, more familiar and may better communicate what we are actually trying to do: use an implanted device to help restore function and improve lives. Thanks to Jen Purks from UF and Amanda Merner from MGH for sharing their preliminary data on how we should be approaching persons w/ disease and teaching us that "words matter."
What resonated with me from today’s DBS Think Tank session on electrophysiology-guided neuromodulation was how rapidly we are moving from simply recording brain signals to understanding which physiology actually matters and then using it to guide therapy. Todd Herrington from MGH showed how human neurophysiology can help us understand the circuits underlying both motor and nonmotor symptoms and ultimately sharpen how we deliver DBS. Ashwini Oswal from Oxford took us inside the beta burst, reminding us that “beta” is not one monolithic signal: timing, duration, cortical-STN interactions and pathological versus potentially physiological bursts matter, and dissecting these features may help us build smarter adaptive DBS. Esther Florin brought a network-level perspective from Germany, using sophisticated electrophysiology and MEG to understand how DBS reshapes brain networks and to identify physiological markers that could help optimize stimulation. The message for me was simple: the future of DBS will not just be about where we stimulate, but about listening carefully to the brain and letting its physiology help tell us when, where and how to stimulate.
What resonated with me from today’s DBS Think Tank XIV Industry Blitz was the remarkable pace of innovation and, even more importantly, the growing power of academic-industry partnership to move the field forward. Andrew Haddock from Boston Scientific, Ben Isaacson from Medtronic, David Greene from NeuroPace and Lyndahl Himes/Yagna Pathak from Abbott each provided a glimpse into where neuromodulation is headed, from smarter devices and sensing to more personalized and adaptive approaches. Industry continues to blaze new paths, but the magic happens when engineers and innovators work shoulder-to-shoulder with academicians and clinicians to ask the difficult questions, perform the rigorous studies and generate the evidence necessary to push the horizon. The technology is exciting, but technology alone is not the destination. The destination is impact: better therapies, better outcomes and better lives for the people we serve.
What resonated with me most from this morning’s DBS Think Tank session on sEEG-guided neuromodulation was the sense that we are moving beyond simply asking where to stimulate and toward understanding the circuits, signals and physiology that can tell us how to stimulate better. Sameer Sheth from Baylor showed how sEEG can become a powerful tool for mapping human brain networks and using what we learn to build better and more personalized DBS approaches. Jonathan Parker from Mayo Clinic highlighted the remarkable information embedded in evoked potentials and how probing circuit responses may help us understand connectivity and ultimately guide neuromodulation. Prasad Shirvalkar from UCSF brought this beautifully into the world of chronic pain, demonstrating how intracranial recordings can reveal individualized pain biomarkers and potentially open the door to closed-loop therapies. Three different perspectives, but one powerful message: the future of neuromodulation may depend less on treating a spot in the brain and more on listening to, understanding and intelligently modulating the network.
What resonated with me from today’s DBS Think Tank session on neuromodulation of mood and cognition was how quickly we are moving from simply observing these remarkably complex human circuits to recording, understanding and potentially modulating them in the real world. Casey Halpern challenged us to think differently about appetition and the circuits driving behavior, Nanthia Suthana showed the promise of responsive amygdala neuromodulation for PTSD, and Cory Inman took us into the fascinating world of capturing and potentially enhancing episodic memories as they are actually formed. What an outstanding session from Casey, Nanthia and Cory, and an important reminder that as we push neuromodulation into mood, memory and cognition, the promise is enormous, but so too is our responsibility to understand exactly what we are changing and why.
What resonated with me from our first session of the DBS Think Tank was the remarkable promise, and also the complexity, of neuromodulating the cerebellum. We are learning that the cerebellum is far more than a motor structure; its circuits reach into movement, cognition, emotion and behavior, opening exciting possibilities for treating diseases we once thought were beyond the reach of neuromodulation. But the challenges are equally real: Where do we stimulate? Which pathways matter? What biomarkers should guide us? And how do we avoid disrupting healthy cerebellar function while restoring abnormal networks? The opportunity is enormous, but getting the circuit right will be everything. Great talks by Dr. San Luciano, Kalia and de Hemptinne.