A few weeks ago, Neuralink’s Brain Chip was implanted to the first human patient ever, coded ‘P1’, standing for Patient 1. His name is Nolan Arbaugh.
Nolan first hears of Neuralink
8 years ago, Nolan was paralyzed from the neck down due to a diving accident. Since then, he had been in a wheelchair, thinking it was going to be that way for the rest of his life. In May of last year, Neuralink announced their campaign for the first human trials. Nolan had heard about the campaign from his friend who was working in the biology industry, who had been keeping up with Neuralink’s fascinating projects. He immediately signed up to be a test subject, and within a swift 6-months, he had underwent the implantation procedure, stayed in the ICU for just a day, and was discharged to go home.
Neuralink
Neuralink was a project, backed by Elon Musk, that aimed to implant ‘Brain Chips’ into human brains. In essence, it is a wireless EEG system that has thousands of channels. Current EEG devices for research purposes are around 32~64 channels, with the channels being non-invasive, measuring brainwaves from contact points with the head.
Neuralink’s ultimate vision is to have these implants be a consumer device, where the average individual can get one implanted into their brains. They envision a world where brainwaves are used to control the world we live in.
However, before they are able to get to that vision of theirs, there are more immediate and justifiable utilizations of the technology, one that is more acceptable to today’s society– leveraging the capabilities of the chip to help patients who are paralyzed.
Machine Learning and decoding brain waves
The chips have the equivalent of >1000 EEG channels, covering more areas of the brain with high precision compared to any other EEGs in market today. With the data collected from the >1000 channels, the brainwaves are then intercepted by a computer, which runs the data through machine learning algorithms to make connections between brain waves and intended actions.
We’ve seen examples of EEG devices being used to control prosthetics, fly drones, play minecraft and other videogames completely hands-free, and much more. And these were with consumer-grade 8-channel, 16-channel devices.
The >1000 channel system created by Neuralink would allow for much finer control and discerning of user intent, and I believe this technology can serve as a big step forward in understanding more about the human brain.
The wave of wearables
Time and time again, I see tech companies leading the wave of innovation for medicine. Apple with their apple watch– they allowed for patients to record their vitals at much more consistent intervals. I say this phrase so much and I love it,
“What used to be a still image of getting vitals once every 3 months (if even), it’s now a continuous video stream of vitals being measured by the minute. 24/7, outside the hospital, in the real-world.”
– By yours sincerely, Daniel Jung
From Apple and their Apple watch to R&D on all-in-one vitals patches for EKG, HR, PulseOx, etc. on their chest, to the promises of smart rings for sleep monitoring,
Tech companies are taking patient care from within the confined spaces of hospitals to the real world.
And they’re smart, too. With the abundance of data they collect and keep to themselves privately, they can leverage the data to train machine learning algorithms to discover new health connections.
To me, it’s as if the amount of data a company have directly correlates to the amount of influence they have over the field. Jensen Huang recently said this about the 4th industrial revolution:
[We’re] not generating electricity as in AC generator … of the last Industrial Revolution, [but] this industrial revolution [is] the generation of intelligence.
Jensen Huang, CEO NVIDIA, 2024 NVIDIA GTC conference
Datasets are the backbone of intelligence. Improper datasets, ones mixed with biases and lack of comprehensiveness will lead to models that are biased, and not comprehensive.
NEJM AI, a new journal that was created in December of 2023, has a datasets and benchmarks section dedicated for recognition of high-quality datasets.
With the trends of wearable devices collecting all sorts of informations from Heart Rate, Pulse Ox, BP, and now brain waves, it’s only a shortwhile until we see deep learning algorithms leverage the technology to make incredible findings about our human bodies.
Excitement for Explainable AI
Current dissatisfaction with healthcare workers in the field is the fact that machine learning algorithms are “Black Boxes”. We have no idea what they’re thinking. However, exciting developments are underway, pushing for explainable AI. Explainable AI is building algorithms that are designed from the ground-up to be human-interpretable. For example, explainable AI would be training a model on decision trees, i.e. Monte Carlo Tree Search. It’s fairly a new topic of exploration for me, but I am very excited to see future works in the explainable AI department in healthcare, as I believe this is exactly the solutions physicians need in order to really leverage AI as an augmenting the workflow.