Artificial Intelligence and Neurology

Artificial intelligence (AI) is now more ubiquitous in healthcare and pharma than ever before, and the magnitude of its role cannot be understated. From areas ranging from data management, to GP consultations, AI has drastically simplified processes that were once laborious.

In the past decade alone, AI has revolutionised how the field of neurology is researched and understood, and has paved the way for numerous critical developments. For example, AI can now diagnose stroke from CT/MRI scans, detect seizures before an attack, and classify neurodegenerative diseases based on gait and handwriting.

Parkinson’s is the second most common degenerative neurological disorder, after Alzheimer’s disease, and is estimated to affect one percent of the population over the age of 60. This can have a significant impact on one’s quality of life, and can present challenges to their loved ones as well. There is an ongoing need to improve treatment and care, implement strategies for prevention, and enhance research into the disease, to keep patients at the fore and ensure they are receiving the care they need.

In Parkinson’s research, artificial intelligence has manifested in the form of genetic sequencing, which has allowed for rapid sequencing of DNA base pairs, and has revolutionised how Parkinson’s is researched and understood. AI has drastically cut the time and costs associated with identifying genes involved with PD, and will continue to play an instrumental role.

Whilst much has been discovered in the past decade, the many complexities and nuances of the disease mean there is still a long way to go in identifying PD-related genes and the cellular processes they support.

Advances in AI have led to revelations about the Parkinson’s pathology, in turn improving treatments for the condition which are tailored to the individual. A significant breakthrough in genetic sequencing has been NeuroX, which was the first DNA chip able to identify genetic variants in a person’s genome to determine any risk for developing a number of late-onset neurodegenerative diseases, including PD. The chip was developed as a result of a 2011 NINDS workshop, in a joint venture between the NIH National Institute of Neurological Disorders and Stroke, and investigators at the NIH National Institute on Aging.1

AI has helped scientists discover that inherited PD has been found to be associated with mutations in a number of genes, including SNCA, LRRK2, PARK2, PARK7, and PINK1. The SNCA gene provides instructions for making the protein alpha-synuclein, which is normally found in the brain as well as other tissues in the body. Studies have shown that alpha-synuclein plays a pivotal role in maintaining an adequate supply of synaptic vesicles in presynaptic terminals. It may also be involved in the movement of structures called microtubules, which help cells maintain their shape.

Many companies are using the benefits of AI to explore Parkinson’s, in areas ranging from diagnosis, to disease progression, to treatment. MIT researchers have developed a sensor which they say can help track Parkinson’s patients’ breathing while they sleep. Tracking is completely contact-free, and the device alerts caregivers to any progression of the patient’s condition. The device emits radio waves, and captures their ref lection to read small changes in its immediate environment.2

This study has explored the link between Parkinson’s and breathing, and the disease is usually diagnosed by a more subjective examination of muscle stiffness, slowness, or tremors. This approach could also be used to help advance the development of new and innovative therapies for Parkinson’s – through making it easier to see when a treatment is working, according to the researchers.

Professor Dina Katabi, Principle Investigator at the University’s AI-focused Jameel Clinic, commented: “Some medical studies have shown that respiratory symptoms manifest years before motor symptoms, meaning that breathing attributes could be promising for risk assessment prior to Parkinson’s diagnosis.”2

Koneksa is another company exploring how AI can predict Parkinson’s disease progression. The company is studying how digitally-collected data points could potentially be used to plot out how an individual’s case of Parkinson’s will advance over time. Researchers aim to identify which data points, collected by a smartwatch or other digital device, can be used to predict and model Parkinson’s progression. As Koneksa founder and CEO, Chris Benko, states: “There are no current diagnostics to detect progression in early PD or in the prodromal (pre-diagnostic) stage, and identifying any predictive digital biomarkers would be a meaningful addition for patients and physicians.”3

As it stands, treatments for Parkinson’s focus on controlling symptoms, and there is currently no cure available. This highlights the importance of using AI in drug discovery, which could accelerate the path to a permanent cure.

BenevolentAI, Europe’s largest private AI company, is collaborating with Parkinson’s UK and The Cure Parkinson’s Trust (CPT), two UK charities, to repurpose at least three existing drugs, and identify two novel drug targets to treat Parkinson’s disease. According to Labiotech: “BenevolentAI’s software uses a computational method known as a ‘five-layer neural network’ to develop models that can predict the bloodbrain barrier penetration and other properties of potential drug candidates. The software’s judgment is continuously updated and improved using machine learning algorithms and feedback from experienced biomedical users.”4

The ongoing development of digital biomarkers is also playing an important role in the diagnosis of Parkinson’s. In the USA, the FDA has approved the use of brain imaging technology to detect dopamine transporters (DaT), an indicator of dopamine neurons, to help evaluate adults with suspected Parkinson’s. The DaT scan uses an iodine-based radioactive chemical, along with single-photon emission computed tomography (SPECT, imaging involving blood flow to tissue), to determine whether there has been a loss of dopamine-producing neurons in a person’s brain. However, as it stands, DaTscan cannot diagnose PD, and cannot accurately discern PD from other disorders that involve a loss of dopamine neurons.

According to researchers in neurology, a highpriority goal is to develop a PET imaging agent which can show alpha-synuclein accumulation in the brain. Alpha-synuclein accumulation in the brain can currently only be confirmed by an autopsy. AI has enhanced all aspects of the PET imaging chain, and is continuing to do so; the ability to detect the protein with an imaging technology in a living person would allow physicians to track the severity of alphasynuclein accumulation over time, as well as help gauge the success or failure of therapies aimed at reducing alpha-synuclein levels. This would be a turning point in accelerating drug development.

When it comes to healthcare, cybersecurity is a prominent issue as patient data can contain very sensitive, personally identifiable information. AI plays a critical role in assisting with breach prevention by proactively searching and identifying previously unkown malware signatures. Patients need to feel secure in providing data to healthcare companies. In order to mitigate the risk of cyber attacks, organisations must ensure they hire sufficient specialised security personnel to properly manage and oversee cybersecurity operations. It is important to use reputable cloud service providers which offer continuous security monitoring and incident response, as they can quickly identify security issues and apply patches.

However, there is the inevitable risk that pharma companies can become over-reliant on AI, which can take away the human aspect of Parkinson’s research. Companies should approach the adoption of AI with caution, and should prioritise successful implementation and management of AI, to make critical, real-time decisions where automation cannot resolve a cybersecurity issue.

Whilst the field of neurology is complex and nuanced, AI has undoubtedly been a gamechanger in simplifying the processes of diagnosis, research, and treatment. For the foreseeable future ‒ until a cure for PD is discovered – researchers hope to improve patients’ quality of life with medications, which can be identified using the capabilities of AI. Although there is still a long way to go, the digital advancements that have been made in the realm of diagnosis and care have completely shifted the way that PD is understood.