The Role of AI in Neurotechnology
Neuroscience and Artificial Intelligence (AI) have both thrived over the past decade. Progress within these two fields has become increasingly interdependent on one another. When devising AI algorithms, developers look to the neural networks of the brain to inform the creation of machine intelligence. Meanwhile, Neuroscience has come to employ more and more intelligent algorithms to better identify and unravel the neurological signals that underlie human emotion, identity, memory, cognition and much more. This interplay between AI and neuroscience has also enabled rapid development within the field of neurotechnology. To provide further understanding of the relationship between AI and neurotechnology, this post will break down the different roles that AI plays in improving neurotechnological functioning.
We’ll focus this article on one of the most common types of neurotechnology: the brain-computer-interface (BCI). For those that are new to the topic, a BCI is a technology that connects the brain with an external device or application, allowing the subject to interact with that external application through brain activity alone. These devices can come in all shapes and sizes. BCI’s always include some sort of electrode or touchpoint through which it can monitor the subject’s brain activity. From here, the BCI needs to identify the relevant neurosignals for whatever task the BCI is being used to perform. The neurosignals then may be converted into a format that is sharable with an external application. Artificial Intelligence can help optimize all of these processes.
Detecting and Processing Neural Signals
A significant role that AI plays in neurotechnology is signal detection. At any given moment, there are a multitude of different neural signals that occur in the brain. AI can be trained to sort through these many neurosignals to find the key activity patterns that are relevant for the task the BCI is meant to do. For example, there are some AI algorithms that have been developed to detect certain neural signal biomarkers that can predict neurological abnormalities like epilepsy or schizophrenia.
AI can also be very effective when used as a signal processing tool. The techniques used to measure brain activity typically record raw electrical signals or the patterns of blood flow associated with those electrical signals. AI algorithms can convert these raw signals into a format that is compatible with external applications. This translation process allows users to control a computer cursor or directly interface with a software program just by using their brain activity.
AI and Bi-Directional BCIs
Finally, AI can interpret neural signals and use that information to modulate the activity of an external application. To accomplish this, BCIs need to work “bidirectionally”. When using a BCI, information can flow in two different directions. It can flow from the brain to the external application via a BCI--(e.g. when brain signals are collected and interpreted) or a command can flow from the external device to the brain via the BCI (e.g. when an external application functions through the BCI to alter brain activity, usually through low grade electrical stimulation). When a BCI functions bidirectionally, that means the device has the ability to operate in both of these directions.
To grasp this, let's consider the example of a deep brain stimulator (DBS), which is an FDA approved, surgically implanted device that sends small electrical signals through the brain to treat Parkinson’s, OCD and severe depression. Typically, these devices are set by clinicians to perform treatments throughout time. Patients must check in with their clinicians often to monitor and adjust treatment frequencies. In recent years however, adaptive DBS has become increasingly popular. Rather than a DBS device working in one direction (e.g. the BCI acts upon the patient to alter brain function) the device operates in both directions using AI to collect and interpret brain signals. These signals provide insight into the patient’s neurological conditions. That information can then be used by the BCI to adapt the patient’s DBS treatments automatically without the need of a clinician.
The complementary expansion of neuroscience and artificial intelligence has advanced the sophistication and capabilities of neurotechnology. Artificial intelligence has the potential to optimize many processes within BCI functioning. The resulting impacts on the health and medical fields, as well as a host of other sphere’s like advertising, gaming, and education could be profound. It is all the more important that we grasp how AI and neurotechnology function and interact together as these technologies become an increasing part of our everyday lives.