As a clinician, this news is extraordinarily exciting. Here's why.
A brain computer interface is something that records neural activity in a person, and then decodes that neural activity to allow them to control objects in their environment. When we think about sources of neural information, it's useful to divide them into non-invasive (EEG, fMRI, MEG, NIRS) technology, and to invasive technology (intracortical electrodes, SEEG, ECoG) systems.
Fundamentally, the DARPA grant is about acquiring new sources of neural information. As the article points out, the latter category is happening almost exclusively as people are hanging out in epilepsy monitoring units with nothing better to do, or as part of early clinical trials looking at safety profiles of implanted devices.
Simply speaking, the divide between invasive and non-invasive systems is about the signal-to-noise ratio of neural information. The closer you get to single neurons, the closer you get to the source of neural activity, the higher the signal quality you get to decode with. Consider that EEG systems (the most commonly used non-invasive methods) average neural activity over the range of several 10^-3 to 10^-2 meters, through the attenuation of spinal fluid, dura mater, skull, five scalp layers, and the electrode interface. Electrodes used in modern intracortical BCI systems are 10^-5 meters in diameter.
To my mind, the DARPA announcement is extraordinary -- these technologies have the possibility of upping the SNR for non-invasive methods, and acquiring fundamentally different sources of neural activity. Importantly, the methods seem to have the ability to work at the bedside, without needing very large and very expensive devices (e.g. fMRI, MEG).
Here are some reasons why this would be important:
1) Someone has been admitted to the intensive care unit, and needs a breathing tube on a ventilator for life saving purposes. Some people in this situation are awake and alert. Literature suggests that being intubated and awake is incredibly scary, since you're aware but can't communicate. Having a non-invasive beside method that could decode your thoughts would be a dramatic increase to the person's quality of life, given their temporary stay. Note that eye-trackers may not be feasible, given the amount of stuff in the ICU room, the fact that they're lying down, etc.
2) If you ask people with locked in syndrome them about their priorities for improving their quality of life, the biggest need is improving their ability to communicate (as opposed to spelling boards, or looking up/down to communicate). Note these people cannot use eye-trackers, due to their neurologic condition.
3) Ongoing clinical trials have had volunteers undergo implantation of electrodes into their brain. The devices implanted have been percutaneous (i.e. had a portion go through their skin). These devices have allowed some participants to feed themselves, either using muscle stimulation or through robotic arm control. Not only would having a non-invasive method obviate the need for a surgery, but having a method for "writing into" the brain would provide closed-loop control of the arm/robotic limb they were controlling, putatively increasing the quality of control.
For those interested, I have posted a relevant screening-exam paper on the main thread that asks the question, "what do we find if we scan the brains of 2000 people?":
I was inspired to spin-this out from a different post about full-body MRI scans.
This paper is asking the question, "If I scan the brains of 2000 people, what can I expect to find?" The findings more or less recapitulate the natural incidence of diseases. They found some brain changes related to aging, a few aneurysms, some benign brain tumors, and even a malignant brain tumor.
I'd like to focus purely on the emotional and economic implications, moving away from the case-by-case medical management side of things. Here are some of the implications of this study:
1) Congratulations -- your full body scan found a brain aneurysm. Being diagnosed with a brain aneurysm is scary and anxiety producing. Thankfully, there is pretty good literature to suggest when one should and shouldn't have their aneurysm treated by their surgeon (or neuroradiologist). After an expensive consult with a subspecialist, your surgeon would like you to have yearly checkups and repeat scans, and possibly even some invasive dye tests. Your new aneurysm may affect your ability to get insurance. It may result in you changing your lifestyle (not going on climbing trips, diving, buying more expensive disability insurance etc.). If may scare you into having the rest of your family getting screening exams for brain aneurysms. None of these outcomes are benign.
2) The kind of benign brain tumor reported from this study are meningiomas. On autopsy series, 1-2% of the population have meningiomas, and they are blissfully unaware of their presence. Finding one means an unnecessary trip to your local friendly neurosurgeon, committing to serial radiographic investigations, and also having insurance implications.
3) The study did discover a malignant brain tumor. I suppose one could make the argument that doing a full body scan is worth it to find a malignant tumor; yet brain tumors affect 1:10000 people, whereas the aforementioned (largely, but not always) benign conditions affect 1-2:100 people.
Of all of the points listed, their eusocial nature is (to me) the weirdest. A quick Google search [0] shows that eusocial organization actually manifests not only with behavior, but with changes in physical characteristics. The 'queen' is bigger (presumably to facilitate having babies) and the guards develop longer teeth. Presumably, they're still diploid.
So, to anyone who speaks evolutionary genetics: what gives? My understanding was that it's evolutionarily advantageous for eusocial insects to cooperate at least in part because they're haploid.
Simulators are effective only as their verisimilitude. Certain neurosurgical procedures are well-reproduced using simulators (most notably the insertion of external ventricular drains[0] for hydrocephalus). Surgical simulation have also shown to be be effective in systems where the principle goal is to get the trainee used to getting visual and limited tactile feedback in specific settings, such as laparoscopic procedures. One thing that has been extensively explored is providing trainees biofeedback about their roughness of manipulating tissues. This in itself is useful; but, to use a musical metaphor, it's more like practicing scales than practicing a performance.
It turns out that recreating tissue in a VR environment that looks and feels like real tissue is a really hard problem. This is a growing industry, and I have no doubt that developers / biophysicists / clinicians will continue to produce more and more realistic systems. Frankly, we're just not there yet. Right now, there is no substitute for the real thing.
Put another way: imagine using a flight simulator where one couldn't reproduce the physics of a real plane. This would be somewhat useful, but have limited transference to actually flying a real plane.
While these salaries may be dramatically different than what the doctors make when they finish residency, you'll see they're decent.
That being said: as a first year resident I once did the math and found that based on the number of hours I was working per month, I was making less than minimum wage.
Could we pay residents better? Absolutely. But, my suspicion is that people would still want to be doctors even if you paid residents half of what they currently made. I don't think medical school admissions would sky-rocket if you paid residents twice as much as they were making right now.
I'm told by my anesthesia friends that it take about 6 weeks to train someone to be technically proficient at putting someone to sleep and waking them back up again.
The five years of training are needed to understand why you're doing it, and to see enough disasters to know how to get yourself out of the disaster situations.
While I don't fly planes, I would imagine flying a plane would be similar.
I also drive a car. While I was technically proficient while I got my license, it took years to get to the point of predicting the behavior of other drivers on the road, and to not panic when unexpected events happened (hydroplaning, odd pedestrian behavior, etc.).
Hyper-specialization has been an approach taken by the medical establishment. For instance, within the last decade, the "vascular surgery" specialty opened up as a specialty straight out of medical school. Vascular surgery procedures used to be a subset of procedures largely (but not exclusively) performed by general surgeons (neurosurgeons still routinely do carotid endarterectomy procedures, for example). There are murmurs of "Spine surgeon" becoming a separate specialty as well, which is currently performed either by neurosurgeons or orthopedic surgeons (often with dramatically different specialty-biases in their surgical technical and reasoning).
Sub-specialists are needed and are important for a variety of reasons, including realistic demands on training time. If my loved one had a single medical issue, I would certainly wanted them operated on by the person who does nothing but that operation.
Woe is the person with two complex medical issues in a hyper-specialized hospital institution. Consider the situation where someone has two medical problems in direct conflict with each other. For instance: some people can develop chronic subdural hematomas (translated: blood clots on the surface of the brain) which causes neurologic dysfunction. The general advice here is to stop any blot-clotting medication someone is on, to prevent further bleeding or expansion of the cSDH. However, what if that person has an artificial heart valve and needs to be on blood thinners to prevent strokes?
How do we proceed? Ask the neurosurgeon, and the answer is "stop." Ask the cardiologist, and the answer is "don't stop."
Sub-specialization is happening, and it's important. But it turns physicians into finer-grained hammers: so every problem they can hit becomes a nail.
In my opinion, one will always need generalists to make the tough calls and to identify which specialist is required. The only way to train generalists is to know what the outcomes are for both stopping and continuing the blot clotting medication.
North American Neurosurgeon here. My opinions are based on my personal experience and my understanding of the literature. I am not an expert in residency well-being or work-hour restrictions.
First: the literature is unequivocal about the effects of chronic sleep deprivation. This cellular aging article fits within what is known already. Sleep deprived humans are dumber, more stressed, and now age faster that non-sleep deprived humans.
The tacit assumption for work-hour restrictions is that sleep deprived physicians result in inferior medical care for patients. It turns out that the literature does not support the tacit assumption. When researchers looked at complications/mortality pre- and post- work-hour restriction, there was no difference. This result is somewhat counter-intuitive, and has been reproduced in various contexts. The one mortality benefit that has been for physicians themselves: less physicians get into car accidents after their call shifts. It's worth noting that there is literature to suggest that less ICU errors are made by sleepy residents, however, this also didn't translate into mortality differences.
So what's happening? There are a few interpretations of these results. The first is that resident physicians really aren't important for patient care, and that the majority of care provided comes from attending physicians. For anyone who has been admitted to an academic institution, this interpretation is silly and is unsatisfying.
The second is that there are sufficient checks and balances. When a sleepy physician makes a silly mistake, nurses can catch errors "this patient has an infection; are you sure you don't want to start antibiotics?" and pharmacists can catch errors "You ordered a hundred times the lethal dose. I'll correct that for you." When the day-team comes by and hears about the patient, they can quickly fix the errors made. Having made some silly errors myself, this certainly plays at least part of what happens.
The third is that most of medicine is rote. Frankly, after a few years of doing the same thing over and over again, you don't really need that much brain power to do a lot of medicine. I have admitted hundreds of people with brain tumors over the years, and the immediate workup and management for the majority for patients is identical. I can do this in a sleep-deprived state. I am not saying that some cases are more complicated than others. As a HN analogy: consider how much sleep you need to print "Hello World!" in your favorite language.
The fourth is that the ability to capture complications / mortality is increasing at the same rate as residency work hour restrictions are resulting in better patient care. In other words, we're searching and finding problems we wouldn't have caught otherwise. It's hard to counter this argument, but I don't know of any data for or against this position.
There are certainly more reasons, but I think the point is made. It's not obvious, it's multifactorial, but it's definitely robust.
OK, so the obvious question is this: Why do resident physicians need to work so hard?
First, let's look at the statistics. Suppose you're providing neurosurgical care for a catchment area of a million people. Brain tumors, as a category, affect 1:10,000 people. This means you'll be admitting 100 new brain tumors a year. However, not all brain tumors are the same (in fact, depending on how you want to slice the pie and what you consider a tumor, there are on the order of 200 different tumor diagnoses). Some tumors are common (e.g. metastatic lung tumors) and some are literally a one in a million diagnosis. In order to see the gamut of tumors in your training, you simply need the time in hospital. In my estimation, brain tumor surgery has a mean of 3.5 hours, with 95% of the cases being between 90 minutes and 24 hours. This means that the average academic center is doing at most two tumors per day. The argument here is that you need the hours to see the cases. There's no doubt that sleepier doctors learn worse than non-sleepy doctors. But I can tell you that I have participated in very rare operations in sleep deprived states, and I remember the approaches much better than common operations I've seen in non-sleep deprived states.
Suppose we are interested in maximizing physician well-being, regardless of the literature supporting patient outcomes. The simple solution is to simply restrict work hours. So, who runs the hospitals? One approach is to hire more residents. Well, residents eventually need jobs. Suffice it to say, swamping the medical profession with more doctors who have less training is not an intuitively satisfying solution to the problem. Moreover, more doctors means more handovers. Increasing handovers has been shown to result in order more unnecessary tests for patients, and to result in inferior patient care.
Restricting work hours means that surgeons may have to train longer. Well, at minimum in the US, becoming a neurosurgeon is 4 + 4 + 7 years of post-secondary education (undergraduate, medical school, residency -- residency used to be 6 years). Many residents choose to become subspecialists, adding another year or two (or possibly 3 or 4). Further work hour restrictions could turn the 15-year training process into a 16 or 17 year process. Fair enough, this could solve the problem. But really? From a purely economic perspective, this is silly. Think of the opportunity cost of educating yourself for 20 years (i.e. 20 years of debt) to work for 25 more before being asked to retire.
Suffice it to say, it's a tough problem. In recognition of this last point, many specialties are moving towards competency based training. That is, graduating surgeons when they meet a competency rather than having spent sufficient time in hospital. This is a no brainer in other fields (e.g. flying planes, building bridges). While this is in fact reflecting the times (greater emphasis on patient safety than in the past), there's no doubt in my mind that this is being implemented because those in power feel that they are graduating surgeons less competent than the generation before them.
This last point, is in my opinion, the important question to ask. If anyone knows of any literature about this last point, I'd be grateful.
This is absolutely true. Another way of saying this is: at a population based level, a very cheap and effective way to decrease the incidence of spina bifida is by providing folate supplementation. However, some mothers are genetically pre-disposed to having children with NTDs, and certain medications interfere with metabolism and can increase the risk of NTDs.
When we talk about clinical trials, we are often interested in answering (1) a primary outcome, which is articulated as a specific yes/no questions based on a measurable clinical outcome, and then (2) a series secondary outcomes, which are answers to auxiliary answers. Clinical trials are usually designed to answer a specific question for a specific population.
Many (actually, most) children born with myelomeningoceles will develop hydrocephalus, which is a mismatch between spinal fluid production and resorption. This is usually treated using a ventriculoperitoneal shunt -- a shunt that diverts fluid from the brain to the abdominal cavity. Shunts can be life saving, but most eventually stop working and need to be surgically revised.
The MOMS trial compared the need for VP shunt and mortality between pre-natal surgery and the standard practice (i.e. post-natal surgery). So: unfortunately, the results of this trial do not generalize to your situation.
However, many academic medical institutions with neurosurgical and rehab departments have ongoing studies related to spinal cord injuries. There are a variety of surgeries that can help people with spinal cord dysfunction who have specific clinical problems (e.g. bladder control, contractures, etc.).
There are also clinical trials that have looked at restorative neurotechnologies for people with spinal cord dysfunction. For instance, check out:
Essentially, this was a prospective randomized trial looking at prenatal surgery vs. standard management with mothers who knew they were going to have a child with spina bifida.
This work stems from pre-clinical experiments using sheep:
A brain computer interface is something that records neural activity in a person, and then decodes that neural activity to allow them to control objects in their environment. When we think about sources of neural information, it's useful to divide them into non-invasive (EEG, fMRI, MEG, NIRS) technology, and to invasive technology (intracortical electrodes, SEEG, ECoG) systems.
Fundamentally, the DARPA grant is about acquiring new sources of neural information. As the article points out, the latter category is happening almost exclusively as people are hanging out in epilepsy monitoring units with nothing better to do, or as part of early clinical trials looking at safety profiles of implanted devices.
Simply speaking, the divide between invasive and non-invasive systems is about the signal-to-noise ratio of neural information. The closer you get to single neurons, the closer you get to the source of neural activity, the higher the signal quality you get to decode with. Consider that EEG systems (the most commonly used non-invasive methods) average neural activity over the range of several 10^-3 to 10^-2 meters, through the attenuation of spinal fluid, dura mater, skull, five scalp layers, and the electrode interface. Electrodes used in modern intracortical BCI systems are 10^-5 meters in diameter.
To my mind, the DARPA announcement is extraordinary -- these technologies have the possibility of upping the SNR for non-invasive methods, and acquiring fundamentally different sources of neural activity. Importantly, the methods seem to have the ability to work at the bedside, without needing very large and very expensive devices (e.g. fMRI, MEG).
Here are some reasons why this would be important:
1) Someone has been admitted to the intensive care unit, and needs a breathing tube on a ventilator for life saving purposes. Some people in this situation are awake and alert. Literature suggests that being intubated and awake is incredibly scary, since you're aware but can't communicate. Having a non-invasive beside method that could decode your thoughts would be a dramatic increase to the person's quality of life, given their temporary stay. Note that eye-trackers may not be feasible, given the amount of stuff in the ICU room, the fact that they're lying down, etc.
2) If you ask people with locked in syndrome them about their priorities for improving their quality of life, the biggest need is improving their ability to communicate (as opposed to spelling boards, or looking up/down to communicate). Note these people cannot use eye-trackers, due to their neurologic condition.
3) Ongoing clinical trials have had volunteers undergo implantation of electrodes into their brain. The devices implanted have been percutaneous (i.e. had a portion go through their skin). These devices have allowed some participants to feed themselves, either using muscle stimulation or through robotic arm control. Not only would having a non-invasive method obviate the need for a surgery, but having a method for "writing into" the brain would provide closed-loop control of the arm/robotic limb they were controlling, putatively increasing the quality of control.