Neuroscience: The next AI frontier

 The demand for neurology in the U.S. and across the globe is exceptionally high, with significant discrepancies between the growing need for neurological services and the actual supply of neurologists: According to expert estimates, by 2025 demand for neurology will have grown by approximately 15-20%. But forecasters project that the supply of neurologists will only grow 7% by mid-decade.

While the recruitment and training of more doctors is one solution, embracing the latest technologies and advancements in artificial intelligence offers a more likely and more effective long-term solution.

Fittingly, our understanding of neuroscience has played a pivotal role in advances in AI, and vice versa. From developments in object recognition technologies as seen in self-driving vehicles, to the faster, more accurate detection of breast cancer through Google AI, the study of how the human mind works has proven a rich source of inspiration for many algorithmic approaches and developments in AI and Machine Learning (ML).

So too have advancements in AI helped further our understanding of the human mind, leading to accelerations in neuroscience development. Second only to neoplasms, the nervous system leads in disease types considered in the AI literature. While initially limited mainly to research applications, the implementation of AI into the neurologic clinical healthcare arena has increased steadily with numerous examples of success in neurodiagnostic testing. This has occurred in areas such as neuroimaging, which has empowered clinicians to make earlier and more accurate diagnoses in stroke victims, developing new treatments such as in Autism, and aiding in prognostication with examples in epilepsy.

Clinicians readily acknowledge the mutual benefit of AI in medicine: 89% of healthcare professionals have expressed that AI has enhanced their work and the systems they use, according to research by KPMG. This enhancement of work is the product of AI as a resource rather than an attempt to replace healthcare professionals. Indeed, when working in collaboration with AI, clinicians significantly outperform those who work independently. This is also true when we remove the clinician from the process, as AI is far more effective when it’s guided by the expertise and discernment of a human professional.

Reaping the benefits of AI in neurology

The benefits of partnering physicians and AI tools in healthcare will be far reaching, with the most notable, overarching and long-term benefit being the ability to utilize AI to offload some of the burden of neurological diagnosis and treatment to our colleagues in primary or emergency care by enhancing their abilities. By the end of 2021, the healthcare AI market is expected to exceed $6 billion as more and more companies invest in and develop AI to do just this.

What will this mean for the future of neurology? RapidAI is an exemplary company on the front lines of neurological innovation. RapidAI’s AI and data-driven technology is empowering clinicians to make faster, more accurate diagnostic and treatment decisions for stroke patients. The company has developed a neuroimaging stroke software platform powered by AI to provide faster analysis of data in treating stroke patients. Utilizing this platform combined with telemedicine, I personally have been able to improve my ability to provide pediatric stroke coverage while also enhancing emergent care for pediatric stroke victims who would have otherwise required transportation (typically by airlift) to our children’s hospital.

Advances in neuroscience have also blurred the boundaries between psychiatry and neurology. Neurologists’ ability to refer patients to psychiatrists will also face growing challenges in the years to come, with similar shortages of psychiatrists projected by 2025.

Psychiatric comorbidities, particularly depression, worsen during the course of major neurologic conditions. Population-based studies suggest that one in every three patients who develop stroke, epilepsy, migraines, or Parkinson’s disease will develop depression. Between 30% and 50% of patients with dementia suffer from depression, and between 27% and 54% of patients with multiple sclerosis (MS) have had an episode of major depressive disorder.

Pear Therapeutics offers another promising look at how medical professionals working alongside AI can make meaningful progress in treating major neurological illnesses. The company is conducting a clinical trial in which multiple sclerosis patients suffering depressive symptoms are treated with a digital therapeutic product, developed in collaboration with Novartis.

At Cognoa, we are harnessing AI to equip pediatricians with the tools they need to accurately diagnose the majority of ASD patients through a prescriptive software as a medical device. The AI we employ is able to analyze thousands of different features indicative of ASD as well as draw on troves of data to make a rapid diagnosis and ensure early intervention occurs. We’re part of a rich fabric of companies applying AI in a targeted way to address major challenges in neuroscience. Aural Analytics is another company utilizing AI to propel the neurology field forward, by analyzing speech patterns to detect subtle changes in brain health.

The benefits of AI are wide-reaching across the neurology field. In a recent analysis published in the Journal of Neurology, neurologist Urvish K. Patel and his colleagues found a wide range of benefits to AI for neurological patients. AI and ML can enable efficient diagnosis and treatment of epilepsy, for instance, and can even spot early warning signs of autonomic instability, helping prevent sudden unexpected death in epilepsy (SUDEP). Additionally, AI algorithms can classify and predict the progression of dementia, guiding more intelligent treatment decisions.

Looking toward the future

For all the promise AI has shown as a neurological tool, it is still early days for the development of AI-based medical solutions. Further progress will hinge on the development of additional massive, high-quality datasets on which to train algorithms. Initiatives such as the BRAIN initiative, the Human Brain Project, the Human Connectome Project, and the National Institute of Mental Health’s (NIMH) Research Domain Criteria (RDoC) initiative are helping to address this need. Because human clinicians will be partnering with AI, securing as much input from doctors as possible will be critical to building sustainable, reliable models that maximize accuracy and effectiveness.

The role of regulation and regulators will also be crucial to the speed and depth of AI usage when it comes to aiding the neurology field. As AI started to be incorporated into medical devices, the FDA recognized  that changes would need to be made to the regulatory process regarding the software as a medical device field and are adapting accordingly.

Last April, the FDA proposed a framework that describes the FDA’s foundation for a potential approach to premarket review for AI and ML driven software modifications.  More recently, they held a two-day public workshop on Evolving Role of Artificial Intelligence in Radiological Imaging – all examples of the FDA proactively supporting the possibilities for AI to improve healthcare. The vision of CDRH Chief Medical Officer and Director for Pediatrics and Special Populations Vasum Peiris, M.D., M.P.H. to create a new nationwide network that can support innovation in pediatric device development, dubbed System of Hospitals for Innovation in Pediatrics (SHIP) is another positive development.

This framework combined with new regulations will allow for AI based diagnostics and therapeutics to be utilized across these special populations, including women (non-pregnant and pregnant), pediatrics, and the elderly who have a proportionality higher prevalence of neurological conditions.

As more patients require neurological services in the years to come, AI offers a pathway to addressing the projected shortage in neurologists – acting as a digital colleague to human professionals. And by alerting neurologists to signs of trouble, enabling clinicians to triage cases, and bringing unprecedented precision and accuracy to treatment and diagnostics, AI promises to alleviate clinical bottlenecks and make neurologists more effective. That makes AI-driven neurology a true win-win.