Whenever Eric Karl Oermann has to tell a patient about a terrible prognosis, their first question is always: "how long do I have?" Oermann would like to offer a precise answer, to provide some certainty and help guide treatment. But although he's one of the country's foremost experts in medical artificial intelligence, Oermann is still dependent on a computer algorithm that's often wrong.
Doctors are notoriously terrible at guessing how long their patients will live.
Artificial intelligence, now often called deep learning or neural networks, has radically transformed language and image processing. It's allowed computers to play chess better than the world's grand masters and outwit the best Jeopardy players. But it still can't precisely tell a doctor how long a patient has left – or how to help that person live longer.
Someday, researchers predict, computers will be able to watch a video of a patient to determine their health status. Doctors will no longer have to spend hours inputting data into medical records. And computers will do a better job than specialists at identifying tiny tumors, impending crises, and, yes, figuring out how long the patient has to live. Oermann, a neurosurgeon at Mount Sinai, says all that technology will allow doctors to spend more time doing what they do best: talking with their patients. "I want to see more deep learning and computers in a clinical setting," he says, "so there can be more human interaction." But those days are still at least three to five years off, Oermann and other researchers say.
Doctors are notoriously terrible at guessing how long their patients will live, says Nigam Shah, an associate professor at Stanford University and assistant director of the school's Center for Biomedical Informatics Research. Doctors don't want to believe that their patient – whom they've come to like – will die. "Doctors over-estimate survival many-fold," Shah says. "How do you go into work, in say, oncology, and not be delusionally optimistic? You have to be."
But patients near the end of life will get better treatment – and even live longer – if they are overseen by hospice or palliative care, research shows. So, instead of relying on human bias to select those whose lives are nearing their end, Shah and his colleagues showed that they could use a deep learning algorithm based on medical records to flag incoming patients with a life expectancy of three months to a year. They use that data to indicate who might need palliative care. Then, the palliative care team can reach out to treating physicians proactively, instead of relying on their referrals or taking the time to read extensive medical charts.
But, although the system works well, Shah isn't yet sure if such indicators actually get the appropriate patients into palliative care. He's recently partnered with a palliative care doctor to run a gold-standard clinical trial to test whether patients who are flagged by this algorithm are indeed a better match for palliative care.
"What is effective from a health system perspective might not be effective from a treating physician's perspective and might not be effective from the patient's perspective," Shah notes. "I don't have a good way to guess everybody's reaction without actually studying it." Whether palliative care is appropriate, for instance, depends on more than just the patient's health status. "If the patient's not ready, the family's not ready and the doctor's not ready, then you're just banging your head against the wall," Shah says. "Given limited capacity, it's a waste of resources" to put that person in palliative care.
The algorithm isn't perfect, but "on balance, it leads to better decisions more often."
Alexander Smith and Sei Lee, both palliative care doctors, work together at the University of California, San Francisco, to develop predictions for patients who come to the hospital with a complicated prognosis or a history of decline. Their algorithm, they say, helps decide if this patient's problems – which might include diabetes, heart disease, a slow-growing cancer, and memory issues – make them eligible for hospice. The algorithm isn't perfect, they both agree, but "on balance, it leads to better decisions more often," Smith says.
Bethany Percha, an assistant professor at Mount Sinai, says that an algorithm may tell doctors that their patient is trending downward, but it doesn't do anything to change that trajectory. "Even if you can predict something, what can you do about it?" Algorithms may be able to offer treatment suggestions – but not what specific actions will alter a patient's future, says Percha, also the chief technology officer of Precise Health Enterprise, a product development group within Mount Sinai. And the algorithms remain challenging to develop. Electronic medical records may be great at her hospital, but if the patient dies at a different one, her system won't know. If she wants to be certain a patient has died, she has to merge social security records of death with her system's medical records – a time-consuming and cumbersome process.
An algorithm that learns from biased data will be biased, Shah says. Patients who are poor or African American historically have had worse health outcomes. If researchers train an algorithm on data that includes those biases, they get baked into the algorithms, which can then lead to a self-fulfilling prophesy. Smith and Lee say they've taken race out of their algorithms to avoid this bias.
Age is even trickier. There's no question that someone's risk of illness and death goes up with age. But an 85-year-old who breaks a hip running a marathon should probably be treated very differently than an 85-year-old who breaks a hip trying to get out of a chair in a dementia care unit. That's why the doctor can never be taken out of the equation, Shah says. Human judgment will always be required in medical care and an algorithm should never be followed blindly, he says.
Experts say that the flaws in artificial intelligence algorithms shouldn't prevent people from using them – carefully.
Researchers are also concerned that their algorithms will be used to ration care, or that insurance companies will use their data to justify a rate increase. If an algorithm predicts a patient is going to end up back in the hospital soon, "who's benefitting from knowing a patient is going to be readmitted? Probably the insurance company," Percha says.
Still, Percha and others say, the flaws in artificial intelligence algorithms shouldn't prevent people from using them – carefully. "These are new and exciting tools that have a lot of potential uses. We need to be conscious about how to use them going forward, but it doesn't mean we shouldn't go down this road," she says. "I think the potential benefits outweigh the risks, especially because we've barely scratched the surface of what big data can do right now."
Seven years ago, mountain biking near his home in Whitefish, Montana, Jeff Marquis felt confident enough to try for a jump he usually avoided. But he hesitated just a bit as he was going over. Instead of catching air, Marquis crashed.
Researchers' major new insight is that recovery is still possible, even years after an injury.
After 18 days on a ventilator in intensive care and two-and-a-half months in a rehabilitation hospital, Marquis was able to move his arms and wrists, but not his fingers or anything below his chest. Still, he was determined to remain as independent as possible. "I wasn't real interested in having people take care of me," says Marquis, now 35. So, he dedicated the energy he formerly spent biking, kayaking, and snowboarding toward recovering his own mobility.
For generations, those like Marquis with severe spinal cord injuries dreamt of standing and walking again – with no realistic hope of achieving these dreams. But now, a handful of people with such injuries, including Marquis, have stood on their own and begun to learn to take steps again. "I'm always trying to improve the situation but I'm happy with where I'm at," Marquis says.
The recovery Marquis and a few of his fellow patients have achieved proves that our decades-old understanding of the spinal cord was wrong. Researchers' major new insight is that recovery is still possible, even years after an injury. Only a few thousand nerve cells actually die when the spinal cord is injured. The other neurons still have the ability to generate signals and movement on their own, says Susan Harkema, co-principal investigator at the Kentucky Spinal Cord Injury Research Center, where Marquis is being treated.
"The spinal cord has much more responsibility for executing movement than we thought before," Harkema says. "Successful movement can happen without those connections from the brain." Nerve cell circuits remaining after the injury can control movement, she says, but leaving people sitting in a wheelchair doesn't activate those sensory circuits. "When you sit down, you lose all the sensory information. The whole circuitry starts discombobulating."
Harkema and others use a two-pronged approach – both physical rehabilitation and electrical stimulation – to get those spinal cord circuits back into a functioning state. Several research groups are still honing this approach, but a few patients have already taken steps under their own power, and others, like Marquis, can now stand unassisted – both of which were merely fantasies for spinal cord injury patients just five years ago.
"This really does represent a leap forward in terms of how we think about the capacity of the spinal cord to be repaired after injury," says Susan Howley, executive vice president for research for the Christopher & Dana Reeve Foundation, which supports research for spinal cord injuries.
Jeff Marquis biking on a rock before his accident.
This new biological understanding suggests the need for a wholesale change in how people are treated after a spinal cord injury, Howley says. But today, most insurance companies cover just 30-40 outpatient rehabilitation sessions per year, whether you've sprained your ankle or severed your spinal cord. To deliver the kind of therapy that really makes a difference for spinal cord injury patients requires "60-80-90 or 150 sessions," she says, adding that she thinks insurance companies will more than make up for the cost of those therapy sessions if spinal cord injury patients are healthier. Early evidence suggests that getting people back on their feet helps prevent medical problems common among paralyzed people, including urinary tract infections, which can require costly hospital stays.
"Exercise and the ability to fully bear one's own weight are as crucial for people who live with paralysis as they are for able-bodied people," Howley notes, adding that the Reeve Foundation is now trying to expand the network of facilities available in local communities to offer this essential rehabilitation.
"Providing the right kind of training every day to people could really improve their opportunity to recover," Harkema says.
It's not entirely clear yet how far someone could progress with rehabilitation alone, Harkema says, but probably the best results for someone with a severe injury will also require so-called epidural electrical stimulation. This device, implanted in the lower back for a cost of about $30,000, sends an electrical current at varying frequencies and intensities to the spinal cord. Several separate teams of researchers have now shown that epidural stimulation can help restore sensation and movement to people who have been paralyzed for years.
Epidural stimulation boosts the electrical signal that is generated below the point of injury, says Daniel Lu, an associate professor and vice chair of neurosurgery at the UCLA School of Medicine. Before a spinal cord injury, he says, a neuron might send a message at a volume of 10 but after injury, that volume might drop to a two or three. The epidural stimulation potentially trains the neuron to respond to the lower volume, Lu says.
Lu has used such stimulators to improve hand function – "essentially what defines us" – in two patients with spinal cord injuries. Both increased their grip strength so they now can lift a cup to drink by themselves, which they couldn't do before. He's also used non-invasive stimulation to help restore bladder function, which he says many spinal cord injury patients care about as much as walking again.
A closeup of the stimulator.
Not everyone will benefit from these treatments. People whose injury was caused by a cut to the spinal cord, as with a knife or bullet, probably can't be helped, Lu says, adding that they account for less than 5 percent of spinal cord injuries.
The current challenge Lu says is not how to stimulate the spinal cord, but where to stimulate it and the frequency of stimulation that will be most effective for each patient. Right now, doctors use an off-the-shelf stimulator that is used to treat pain and is not optimized for spinal cord patients, Harkema says.
Swiss researchers have shown impressive results from intermittent rather than continuous epidural stimulation. These pulses better reflect the way the brain sends its messages, according to Gregoire Courtine, the senior author on a pair of papers published Nov. 1 in Nature and Nature Neuroscience. He showed that he could get people up and moving within just a few days of turning on the stimulation. Three of his patients are walking again with only a walker or minimal assistance, and they also gained voluntary leg movements even when the stimulator was off. Continuous stimulation, this research shows, actually interferes with the patients' perception of limb position, and thus makes it harder for them to relearn to walk.
Even short of walking, proper physical rehabilitation and electrical stimulation can transform the quality of life of people with spinal cord injury, Howley and Harkema say. Patients don't need to be able to reach the top shelf or run a marathon to feel like they've been "cured" from their paralysis. Instead, recovering bowel, bladder and sexual functions, the ability to regulate their temperature and blood pressure, and reducing the breakdown of skin that can lead to a life-threatening infection can all be transformative – and all appear to improve with the combination of rehabilitation and electrical stimulation.
Howley cites a video of one of Harkema's patients, Stefanie Putnam, who was passing out five to six times a day because her blood pressure was so low. She couldn't be left alone, which meant she had no independence. After several months of rehabilitation and stimulation, she can now sit up for long periods, be left alone, and even, she says gleefully, cook her own dinner. "Every time I watch it, it brings me to tears," Howley says of the video. "She's able to resume her normal life activity. It's mind-boggling."
The work also suggests a transformation in the care of people immediately after injury. They should be allowed to stand and start taking steps as soon as possible, even if they cannot do it under their own power, Harkema says. Research is also likely to show that quickly implanting a stimulator after an injury will make a difference, she says.
There may be medications that can help immediately after an injury, too. One drug currently being studied, called riluzole, has already been approved for ALS and might help limit the damage of a spinal cord injury, Howley says. But testing its effectiveness has been a slow process, she says, because it needs to be given within 12 hours of the initial injury and not enough people get to the testing sites in time.
Stem cell therapy also offers promise for spinal cord injury patients, Howley says – but not the treatments currently provided by commercial stem cell clinics both in the U.S. and overseas, which she says are a sham. Instead, she is carefully following research by a California-based company called Asterias Biotherapeutics, which announced plans Nov. 8 to merge with a company called BioTime.
Asterias and a predecessor company have been treating people since 2010 in an effort to regrow nerves in the spinal cord. All those treated have safely tolerated the cells, but not everyone has seen a huge improvement, says Edward Wirth, who has led the trial work and is Asterias' chief medical director. He says he thinks he knows what's held back those who didn't improve much, and hopes to address those issues in the next 3- to 4-year-long trial, which he's now discussing with the U.S. Food and Drug Administration.
So far, he says, some patients have had an almost complete return of movement in their hands and arms, but little improvement in their legs. The stem cells seem to stimulate tissue repair and regeneration, he says, but only around the level of the injury in the spinal cord and a bit below. The legs, he says, are too far away to benefit.
Wirth says he thinks a combination of treatments – stem cells, electrical stimulation, rehabilitation, and improved care immediately after an injury – will likely produce the best results.
While there's still a long way to go to scale these advances to help the majority of the 300,000 spinal cord injury patients in the U.S., they now have something that's long been elusive: hope.
"Two or three decades ago there was no hope at all," Howley says. "We've come a long way."