Blog ›

The era of brain data - neurotech after the turning point

If Silicon Valley has a favorite economist, it might be Carlota Perez. Her 2002 book “Technological Revolutions and Financial Capital” has been heavily influential at top VC firms as a framework for understanding the relationship between technology and financial capital at different stages of the economic cycle, and neatly predicted the frenzied period in investment and politics of the last decade. More recently, Perez has argued that the economic crisis precipitated by COVID-19 could be the “Turning Point” for the current Information Age economic cycle. If true, this means we are about to see a reorganization of our social and regulatory environment that enables a fuller spectrum of society to capture the benefits of information technology. We should expect the change to be especially profound in areas like education, transportation, government, housing, and healthcare which have seen relatively slow adoption of new technology over the last several decades.

Healthcare, in particular, seems poised for a significant period of innovation post-COVID-19. Healthcare costs have steadily increased as a share of US GDP over the last decade and so the way we regulate, deliver and provide access to, and pay for healthcare was already under increased scrutiny. That this dynamic will be amplified 10x in the post-COVID environment has not been lost on the venture capital community, who sense the potential for startups to make a massive impact (below announcements from last quarter alone):

As ARCH Venture Partner Robert Nelson put to his followers on Twitter:

The coming information technology revolution in healthcare will create mini-revolutions in many individual niches. The niche I think the most about is the “neuro” part of healthcare, since my company Rune Labs provides software for neuromodulation, neurotech, and neuroscience therapeutics. As healthcare evolves over the next decade to fully incorporate information technology— and as society makes the necessary changes to enable this process — we see the demand for neuro healthcare products growing rapidly along with trends in telehealth, increased demand for universal healthcare access, and a renewed focus on mental health. In this article, I’ll outline three theses about how the neuro industry will grow and adapt through this current Turning Point, which I hope can be helpful as a guide for patients, clinicians, researchers, and companies who are invested in development and delivery of neuroscience therapies in the years to come.

Thesis #1 —The need for remote monitoring of brain function will increase as more elements of clinical practice are forced into a telehealth environment

Perhaps the most obvious change in the healthcare environment as a result of COVID-19 is the surge in telemedicine use, brought on by a rapid spike in demand coupled with relaxed regulations. Healthcare investment bank BTIG outlined a few key examples of these forces in a recent update on the digital health market:

- CARES Act allows Medicare to temporarily reimburse telehealth services in any location (not just rural areas) and by any provider (no pre-existing relationship).

- The Federal Communications Commission (FCC) is launching a new telehealth program aimed at using $200 million in new federal funding to improve broadband connectivity for connected health services.

- From Teladoc Press Release (March 13, 2020) — “Patient visit volume spiked 50 percent over the prior week and continues to rise. The company had been handling visit demand consistent with peak flu volumes, but on Wednesday began to see that number accelerate to as much as 15,000 visits requested per day”.

- Janie Jun, Ph.D., associate director of quality and provider strategy at Lyra Health, told FierceHealthcare that demand for video visits has nearly doubled as the pandemic stretches on. Jun said that 85% of Lyra’s mental health visits are now conducted via secure video or telephone calls.

- Providers will increasingly seek solutions to keep patients away from healthcare facilities in the medium term until a vaccine is approved (“healthcare at home”).

Some of the largest public digital health companies help provide effective “healthcare at home” for diabetes and cardiac disease using wearable medical devices, and neurology and psychiatry are not far behind. Over the next few years we will see the first companies using wearable neurotech to manage chronic neurodegenerative and psychiatric conditions get significant market penetration and scale. When data from these devices is used to power diagnosis and decision support it can increase clinician agency closer to the level of a normal clinical visit, with potential implications for reimbursement. Crucially, at-home use of wearables can also enrich the telemedicine experience for the patient, providing extra motivation for patients to keep the same regular schedule for telemedicine as for their normal in-person clinical visits. Neurotech can keep patients engaged in their own care, and give patients confidence that their doctors have all the information necessary to provide them with the best care, even over a video call.

The epilepsy community is likely to be an early adopter of at-home neurotech, with well-accepted links between EEG and symptoms (seizures). Companies including Ceribell and Neuroelectrics have taken the first step of making EEG devices that are user-friendly enough to be prescribed to patients for use in-home. Neuropace was an early mover in this area, allowing patients to record localized brain signals from their RNS implant at-home and share with their clinicians for review. At-home brain signal monitoring will only grow in importance as we improve our understanding of the links between brain signals and psychiatric disorders. Looking forward a few years to new types of data that may be available, there are several high-profile efforts to dramatically increase the quantity and quality of neural data that can be recorded at home, including Bryan Johnson’s Kernel and Mary Lou Jepsen’s Openwater.

As with most big data applications, the key to making at-home nerual data useful will be high-quality data labeling. Fortunately, all of the above technologies and products for at-home monitoring of neural signals can massively benefit from the large install base of data-labeling infrastructure: smartphones and (to a lesser extent) consumer wearables like Apple Watch and Fitbit. Neural data with no additional context can lack the sensitivity and specificity to be useful in diagnosis and decision support, but with the right software and data analytics this context can be provided by the patient’s own mobile devices. In the simplest case these labels can help detect and remove artifacts related to movement or other activity from the neural data. The next level of value is using labels from phone and wearable data to isolate neural signals relating to the pathology from neural signals relating to normal behavior, as has been extensively explored with beta-band dynamics in Parkinson’s disease. At scale, the ability to link neural signals with data collected from every-day mobile devices could provide a new way to detect candidates perfectly suited for a particular therapy or intervention, which has implications ranging from more efficient clinical trial recruitment to larger patient funnels for neuromodulation therapies.

Thesis #2 — Adaptive and closed-loop (AI-powered) neuromodulation therapies will become more attractive as doctors and hospitals are challenged to increase efficiency to meet demands for more universal healthcare access

The combination of a public health crisis and large-scale unemployment caused by COVID-19 seems likely to increase the urgency around a more universal form of healthcare in the US, with concrete changes in this direction already happening at the state and federal level. This increase in demand will not be met with an adequate supply of healthcare workers on the other side *unless* those healthcare workers can be given access to tools which increase their productivity substantially. Fortunately, over the last five years we have seen that AI-based decision support tools can make a real difference here, with initial successes in stroke visualization and diabetic retinopathy diagnosis. The pressure to adopt these tools will grow with increased demand for healthcare from a limited pool of healthcare workers. This sentiment is captured in Bond Capital’s COVID-19 trends report:

“The current crisis is a reminder that our healthcare labor resources were already stretched thin. Automation will continue to make inroads in healthcare to reduce workload and improve the quality of data capture.”

Neuromodulation therapies — Deep Brain Stimulation (DBS), Spinal Cord Stimulation (SCS) and Transcranial Magnetic Stimulation (TMS) — are currently fairly labor intensive for clinicians to administer. Unlike traditional pharma therapies, neuromodulation therapies must be programmed and calibrated on a patient-by-patient basis. This programming accounts for patient-specific anatomy and neural circuitry, possible drug interactions, and, in the case of implantable therapies, the final surgical placement of the device. The labor-intensive nature of neuromodulation manifests itself in several undesirable ways: reluctance on the part of clinicians to prescribe the therapies, limited coverage by insurers who limit reimbursement to “refractory” patients, a high cost to the device makers for employing small armies of clinical field engineers to assist clinicians with programming, and finally sub-optimal patient experience as it can take months to program new devices correctly, and in some cases patients never reach their optimum therapy settings. I’ve written elsewhere about the coming wave of adaptive and closed-loop neuromodulation technology that has been created to address these challenges. The deployment of this wave of AI-enabled neurotech will only accelerate with demands to provide care to more patients for less cost, possibly in a remote setting via telemedicine.

In Deep Brain Stimulation (DBS) the push towards adaptive therapy is being led by companies like Medtronic. In January, Medtronic announced their new Percept system with integrated capability to directly sense brain signals (in addition to providing stimulation from the same device). The Percept is capable of recording local field potentials (LFPs) from deep brain structures from the Parkinsonian neural circuit and process these signals in-situ to control the neuromodulation output. Data recorded from Percept implants will soon create the largest-ever at-home human neurophysiology dataset for that indication, with signals recorded directly from the neural circuits implicated in the pathology. Early evidence from Rune Labs collaborators at UCSF and other teams has shown that these signals might be used to achieve better patient outcomes, both from reaching an optimum steady state more rapidly and from having continuously adaptive therapies which change over time, for example with natural changes in patient symptoms along a circadian cycle. In particular, “beta-band” signals have shown promise as a biomarker for pathological symptoms which may be sensitive and specific enough in a large cross-section of the population to be deployed as a commercial therapy.

In the case of non-invasive Transcranial Magnetic Stimulation (TMS), algorithms which can use patient EEG and fMRI to automatically target stimulation may reduce the length of treatment by as much as 10x. Such is the promise of studies like Nolan Williams’ SANTE trial at Stanford, which through a combination of smart targeting and improved dosing reduced treatment times from 10 weeks to 5 days. We started Rune Labs to build the software and data infrastructure to enable these types of next generation neuromodulation therapies at scale, which in the case of TMS means that clinics could see 10x the number of patients with the same amount of staff and overhead as today. Following that, cost of treatment can come down, and labeling can be expanded to cover a broader range of patients and help more people. TMS could be a great Perez “Golden Age” success story where the full adoption of healthcare IT generally, and AI-enabled neurotech specifically, by TMS clinics dramatically drives down the key therapy cost input (physician labor) and in turn widely distributes the benefits of this important treatment for depression throughout society.

Thesis #3 —There will be a resurgence in demand for neuroscience therapeutics resulting from understanding and awareness of the costs of poor mental health and a post-COVID rethink of the role of nursing homes in our healthcare system

The first two decades of the 20th century have largely disappointed anyone hoping for new treatments for neurodegnerative and psychiatric disease, and have left patients burdened with these conditions with the highest unmet needs in medicine today. Promising candidates for new neuroscience drugs have repeatedly failed Phase III clinical trials, and meanwhile a data revolution coupled with a favorable reimbursement environment in oncology and rare disease has made it more lucrative for pharma companies to spend on these elements of their pipelines instead of on neuroscience. Several pharma companies have shuttered their neuroscience divisions altogether. Lacking adequate treatment — never mind cures — persons with Parkinson’s and Alzheimer's Disease have few choices except high-cost nursing homes and long-term care facilities. People with severe psychiatric illness are similarly lacking for adequate care, with “deaths of despair” becoming such an epidemic that overall life expectancy in the US has actually decreased since the turn of the century.

Fortunately, even before COVID-19 forces were aligning behind a renewed push for new neuroscience therapies. The Brain Initiative has been funding basic and applied research towards new targets, biomarkers, and therapies since 2015, Johnson and Johnson’s approval of esketamine for depression and a loosening regulatory environment for psychedelic medicine has led to experiments with new care delivery models, and some big pharma companies have been redirecting R&D spending toward neuroscience as oncology and rare disease have started to show diminishing returns. Meanwhile, employers have increasingly accepted that productivity loss from poor mental health can have significant negative effects on the bottom line. COVID-19’s perfect storm of social distancing-induced isolation and economic anxiety will sadly add to the already large number of people in the US in need of some form of care for depression and related conditions. The silver lining may be that the universal nature of COVID-19 for our society and widespread acceptance of its impact on our collective mental health may go a long way toward reducing the stigma long associated with seeking treatment, with the result being a large and sustained increase in demand for new therapies for psychiatric disease. COVID-19 is also going to cause the US to reevaluate nursing homes as a viable way caring for our elderly population, many of whom are in nursing homes as a result of a neurodegenerative condition such as Parkinson’s or Alzheimer’s Disease. The current pandemic is exposing the structural and regulatory flaws which have caused nursing homes and long-term care facilities to become “ground zero” for multiple outbreaks. Post-COVID-19 nursing homes are going to be more resilient than before the pandemic, but they will also be even more expensive which will force CMS to reevaluate reimbursement for new therapies that can keep patients well enough to “age in place”.

So, we have promising new science on which to base new potential therapies for neurodegenerative and psychiatric disorders, and society is changing in ways that is going to massively increase demand for these products. Still the question remains of how to actually develop the new therapies, given the challenges and failed clinical trials of the last few decades. This is where neurotech provides hope — leveraging neuroimaging, in particular EEG and fMRI, as intermediate endpoints in clinical trials.

In order for a therapy to come to market it needs to meet its stated goals, and accepted endpoints for neuroscience therapies are notoriously qualitative, typically consisting of scores derived from patient and clinician observations or surveys. Parkinson’s, Alzheimer’s, depression, and chronic pain are all defined as sets of symptoms or behaviors, and thus we evaluate the success or failure of a treatment against its ability to modulate these symptoms or behaviors. Over time certain scales and surveys have become accepted as ways to measure this behavior change. However, tying the success of the therapeutic to a behavioral outcome is inherently problematic, since the therapy — be it a drug or neuromodulation device — may remove an underlying pathology at the neurophysiological level without having an immediate behavioral outcome. In other words, pills don’t teach skills, and a successful therapeutic may require additional steps (for example, cognitive behavioral therapy or social training) to achieve the desired overall behavioral effects on patients.

Autism may be an important test case for this new combination of neurotech and pharma. Researchers collecting EEG data from persons with autism have not only shown differences between persons with autism and healthy controls, but at least two distinctive phenotypes within the autistic population. Notably, neuroimaging-based studies have potentially greater predictive power than genetic screening and classification in autism. In new clinical trials for autism drug therapies participant EEG may be monitored carefully to detect changes in the underlying neural circuit, after which participants might enter a second “teaching” phase of the therapy where they learn new social and communication skills. Additionally EEG may evolve as an inclusion criteria for some trials, as putative drug candidates may be expected to work on one phenotype of autism and not another. Amit Etkin’s group at Stanford has recently shown that EEG may indeed be able to play this role in the development of new antidepressants, with EEG being predictive of patient response to antidepressants vs EEG in a retrospective study.

The COVID-19 crisis has dramatically accelerated changes in behavior, regulation, and expectations around healthcare in ways that have unlocked a new wave of opportunity for neurotech enabled therapies and products. As these come to market, millions of people can look forward to greater relief from seizures and debilitating depression, non-addictive alternative treatments to chronic pain, and years of life back from neurodegenerative disease. Something we get especially excited about at Rune Labs is the positive feedback loop that becomes possible when the high-quality, real world neurophysiological data generated by these new therapies is made available to researchers to create new, even more effective therapies. Many have predicted the next technological revolution to be a biotech-led revolution — with any luck neurotech will be firmly in the mix as well.


Contact us

Our team of engineers and scientists is looking forward to meeting you.