I have been a big proponent of Adaptive Deep Brain Stimulation (aDBS), often championing it as the next frontier in neuromodulation. However, intellectual honesty demands an accounting of both its advantages and disadvantages. The technology operates on a closed-loop system designed to adjust stimulation in real time, relying on reading a single signal—specifically the beta frequency band—in the local field potentials from the area around the electrode to dictate the amount of current delivered. But the system has a critical vulnerability: when it fails to pick up a strong enough beta signal, the algorithm has a tendency to rail low.*
This constraint has become my reality. For the last few weeks, my system has been frequently dropping my stimulation to the lowest possible amplitude within the predefined parameter space set at my last clinical visit. That baseline setting is appropriate for sleep, when my physiological demands are minimal and I do not require as much intervention. But navigating the physical demands of the day requires significantly more juice. Being artificially under-stimulated during my active hours renders the adaptive feature a liability rather than a benefit.
Far from optimizing my performance, the adaptive setting railing low left me more bradykinetic than I had been in a while. My range of motion was constricted, my speech was noticeably more slurred, I shuffled my gait more, had more freezing episodes and I had a few falls over the last few weeks.
Another issue unique to aDBS is the lack of patient control. The current iteration of aDBS does not allow us to make minor adjustments the way the older system, often referred to as continuous DBS (cDBS), did. This ties us to our clinics for every adjustment, no matter how minor. For this technology to be more viable, patients should be granted some ability to adjust the parameter space themselves.
This dependency creates an unsustainable burden for both patients and clinicians. For me, it means enduring physical strain while also fretting over the logistical friction of coordinating an appointment just to tweak a few parameters. For my doctors, it translates to schedules bogged down by routine troubleshooting that should be manageable by patients. An advanced technology that at times needs manual tweaking by highly specialized neurologists isn’t scalable.
There is, however, one significant advantage to having an adaptive DBS system. To borrow from comedian Mitch Hedberg’s joke about broken escalators simply becoming stairs—”sorry for the convenience”—an adaptive DBS device can fairly easily and conveniently be switched to cDBS when needed. However, I do feel a pang of guilt on the occasions when I have done this due to the wasted hours my clinical team spent tweaking my adaptive algorithm.
Nevertheless, I know they would agree that the goal is for me to feel better. Thus because of the unpredictability I’ve been experiencing, I am saying goodbye to aDBS for now and have switched back to cDBS for more control. While it lacks the theoretical elegance of a closed-loop system, continuous stimulation provides the reliable, manual control I need to function without the risk of the algorithm bottoming out my amplitude when the signal drops.
The broader issue this raises is the field’s over-reliance on the beta signal as the sole proxy for disease.** In the real world, the beta signal is volatile—it can be washed out by voluntary physical movement, obscured by medication cycles, or simply drop out despite worsening motor symptoms. When an algorithm chases this single, unstable brainwave, it loses sight of the actual patient.
To move forward I believe the field needs to find a way to make better use of wearable data. That kind of continuous telemetry offers the opportunity to capture a much more accurate representation of how patients are doing and feeling, moment to moment, in the real world. Wearables capture some of the physical reality of the disease—tremor, gait, balance, and dyskinesia—with much more fidelity than a marker trapped deep in our brains or any clinical scale currently can.
Until adaptive systems can synthesize real-world behavioral data alongside neurological signals, the technology will remain compromised. Device manufacturers must recognize that a “smart” implant that ignores crucial signals in favor of a single, flawed metric is, given enough time, likely to be manually overridden. Leaving these breakthrough technologies rendered obsolete by the very people they were designed to help.
* “Railing low” occurs when the system’s algorithm automatically drops your stimulation amplitude to its minimum baseline setting because it fails to detect a strong enough neurological signal to justify delivering more power.
** Note from Dr. Alfonso Fasano: “aDBS is able to do more than just detect Beta frequencies and we’re learning new tricks daily, now that it’s commercialized.”