Can Robots Heal an Injured Brain?

The last words that Liviana Sulzer spoke, 18 months ago, were very much in character: “Now it’s time for a song.” This was often how she felt, living as she did inside a toddler movie-musical, where even just a spilled cup of milk could get her up onto a chair, twirling with her arms out wide and singing as loud as she could manage: We just spilled our milk … It was messy on the table, and then we cleaned it up … And noooow it’s aaaaall cleeeaaaned up! When the song was over, she’d bend toward her brothers, ages 6 and 1, in a deep and gracious bow.

It was May 2020—a week before Livie’s fourth birthday—and the kids were playing in the yard. Throughout the Sulzers’ quiet neighborhood in Austin, Texas, the Persian silk trees had begun to bloom in pink-tipped puffs. There were flowers in their backyard, too. Livie had a favorite one, purple and about as tall as she was. She called it Dr. Iris and, trapped at home by the COVID-19 shutdown, she’d made a game of scooting over to it in her push-car and spilling all her problems. (She often couldn’t think of any when she got there.)

But the loneliest phase of the pandemic, with its makeshift games and spotty child care, was nearly over. Things were getting back to normal. A nanny had started just over a week before, and Livie’s mother, Lindsay—a bioengineer and expert in regenerative medicine—was headed to the office for her first day back at work, at a local cell-therapy start-up. Livie’s father, James, an assistant professor at the University of Texas at Austin who specializes in rehabilitation robotics, was grading papers in the walk-in closet that he’d turned into a home office. He’d asked his graduate students to propose studies or devices that might one day help a patient recover from a nervous-system injury.

The sky was clear and calm and sunny. Livie stood near the center of the yard, 30 feet below the overhanging branches of a pecan tree. Her two brothers were nearby.

“Now it’s time for a song,” she announced.

There was a crack, a whoosh, a scream.

Livie was unconscious when the EMTs arrived, her eyelids fluttering. In the ambulance, James overheard someone say “blown right pupil.” He didn’t know what that implied. A falling tree branch had knocked his daughter out, but there wasn’t any blood. How bad could it be?

At Dell Children’s Medical Center, a neurosurgeon named Winson Ho knew right away that it was bad—really bad. Livie’s blown pupil, the fact that it was dilated and unresponsive to light, told him that her brain had been swelling for some time, pressed against the inside of her skull. Without rapid intervention, she could die. A CT scan showed a thick fracture across her crown that forked into a pair of smaller, twig-like cracks—as if the branch’s shape had been imprinted in her bone. In the operating room, Ho carved out a piece of Livie’s skull, four inches wide and six long, to give her brain more space to bloat. If an older person had come in with that same degree of injury, he’d later say, doctors and family might have chosen not to intervene.

Livie spent the next two weeks in a coma. She had lesions on her right sensorimotor cortex, her left orbitofrontal cortex, and the tract of fibers called the corpus callosum that connects the brain’s two hemispheres. James remembers being told that Livie might end up with a little trouble walking, and some difficulty concentrating on math problems. “It was devastating to think that she’d be permanently injured,” he said.

In a grim coincidence, he’d spent his whole career devising ways to fix a damaged nervous system, and Lindsay had once worked on isolating stem cells out of body fat, an approach that has been used to treat traumatic brain injuries (TBIs) such as Livie’s. The two had met as graduate students at Northwestern University in 2004; James had been training at the affiliated Rehabilitation Institute of Chicago, one of the world’s leading hospitals for physical medicine and rehab. Between the two of them, James and Lindsay had decades of experience in biomedicine, and a large network of professional connections. “There’s no treatment out there that we don’t have access to,” James told me.

While Livie was in the ICU, James reached out to friends and colleagues and asked them for advice: Which treatments should they try with Livie in the weeks ahead? Which technologies might help? He also started coming up with notions of his own. At the hospital, James and Lindsay had to stretch out Livie’s joints three times a day to help prevent contractures, a shortening of the muscle fibers that can result in lasting disability. There must be some way to automate the stretching, James thought—perhaps he could design a robot that did it better and more often. “I was looking for opportunities to apply what I knew to help her,” he said. But eventually he abandoned the idea. Building the stretching bot would take months, he realized—and even then it might end up pushing Livie’s tiny joints too far and hurting her.

From our November 2021 issue

Check out the full table of contents and find your next story to read.

See More

The risk of contractures soon gave way to other, graver ones. After two weeks, Livie came out of her coma, though only to a point. Her eyes were open now and she was breathing on her own, but she wasn’t making any sounds or responding to the world around her. “When Livie starts talking again, what will she sound like?” her older brother, Noah, asked his parents at one point. “She had such a cute voice.”

Her voice returned, a few weeks later, in the form of wounded, mouselike shrieks—a high-pitched din of pain or maybe fear as she emerged from a semiconscious state. In mid-July, James flew with Livie to the Kennedy Krieger Institute in Baltimore, well known for its work with TBI in children; Lindsay took the boys to stay with family in Cleveland. (The parents would trade places twice a month.) In Baltimore, they rarely left their daughter’s side, sleeping on a fold-out chair in her room, haunted by the sounds of her discomfort. The only way to make her shrieking stop, James found, was by giving her koala hugs: She was the bear; he was the tree.

As the founder of UT’s Cellular to Clinically Applied Rehabilitation Research and Engineering Initiative, James had seen a thousand clever fixes for a person’s brain: neural stimulators and electrode caps; full-body exoskeletons; sleek, motorized contraptions that facilitated movement at a single joint. At his lab there was a split-belt treadmill that could measure the force of each footstep. His colleague had a robot that could assist the movement of both arms at once.

James has always been a builder. In high school he was into woodshop, making coffee tables and armoires; later, he interned at Alcoa, the aluminum company, where he saw a massive forging press make wheels for trucks. As a college sophomore, he learned about advances in prosthetics, and it occurred to him that tinkering could serve a greater good. A boom was under way in the field of rehabilitation robotics; in graduate school, James made a powered brace to help stroke survivors bend their knees. Later, he shifted his focus to the brain itself, designing tools for neuro-feedback that used multimillion-dollar MRI machines to nudge a person’s cortex into making new connections.

Working in an engineering lab, one tends to fixate on the engineering challenge: building the device. Whom exactly the device is for and what sorts of injuries it can help address are secondary concerns. Now that logic had flipped around as James sat beside his daughter. He knew that Livie’s brain could still send signals to her muscles, even if those signals weren’t strong enough—or clear enough—to make her muscles work. So he came up with a way for Livie to exercise her neurons while her body remained still. With the help of a graduate student, he attached electrodes to her limbs and neck, to pick up even feeble spurts of muscle activation; then he linked them to a music playlist. Whenever Livie twitched her biceps or her triceps, even just a tiny bit, a favorite song, such as “Baby Shark,” would play a little louder.

The electrode gadget checked all of James’s boxes for design: It allowed Livie to participate in her own recovery; it encouraged her to practice; it made use of neural information that a doctor or a parent might never see during a normal course of treatment. Yet it proved as useless for Livie as the stretching bot that James had only pictured in his head. For one thing, placing the muscle sensors took too much time, time that could be spent on other forms of therapy; he also couldn’t tell whether Livie understood the point, that she should try to make the music louder; and even if these other problems could be solved, Livie’s muscles were so small that some activation might be missed.

Livie was making progress now, but in slow motion. In September, she moved back to Austin, where she started doing therapy sessions at home (nine hours every week, plus 30 more of exercise and “therapeutic recreation”) and as an outpatient at Dell Children’s (another seven hours). Still, she wasn’t quite responsive; her eyes were misaligned; her head was cocked off to the left and couldn’t seem to straighten up. She had trouble swallowing and had to take her meals through a tube connected to her stomach. Her right arm worked a bit, and her left leg too, but she hadn’t figured out a way of rolling over. James and Lindsay knew about the crucial, early window for reshaping and remapping the brain—in many cases, a person’s progress in the first few weeks or months after an injury can predict how things play out in the long term. The level of recovery they’d once imagined now seemed like foolish optimism.

In the months that followed, Lindsay took on the Herculean tasks of arranging Livie’s care—hiring nurses and personal attendants, procuring equipment such as wheelchairs, and setting up a never-ending carousel of feedings, medicine, and exercise, all while wrestling the Hydra of insurance claims. Lindsay also puts in a few hours a week at her cell-therapy start-up, and attends to her younger son, Reed, now 2 years old and almost always seeking her attention.

From the November 2000 issue: Children’s products and risk

James assumed the role of in-house rehab scientist: the family’s principal investigator into Livie’s injury, and its chief adviser on how to treat her most effectively. Nature’s whim had put his daughter in this awful place. Technology would help to bring her back. “I feel this weight of responsibility,” he told me, “given what I feel I should know about the field.”

Shortly after Livie’s accident, while she was still unconscious in the ICU, James reached out to another dad he knew in Austin—a guy in commercial real estate named Barney Sinclair whose own daughter Charley had been injured several years before, when she was roughly Livie’s age. Barney had been headed out to Oklahoma with three kids in the car. The highway was wet with rain; another car hydroplaned across the median and Barney smashed into its side. Charley’s brain, like Livie’s, started swelling in her skull; surgeons had to drill a hole to reduce the pressure.

Charley was treated at the same hospital as Livie would be. Barney, feeling helpless, started asking questions of her doctors and nurses: If I were Bill Gates, he’d say to them, what would I be doing to help my daughter? You know, like, if resources were not an issue? Eventually he landed on robotics, and in 2018 he started a nonprofit—he called it Project Charley—with a plan to purchase gait-training bots and other high-tech tools for rehab clinics in the Austin area. Charley would get the benefit of using them, and so would other people like her.

That’s when James and Barney met. “I’m a real-estate guy, right? I build warehouses,” Barney told me. “I don’t know what equipment to buy, but I know how to tap into the smartest people in Austin and let them help me make smart decisions.” So he visited James’s lab and saw the split-belt treadmill and the two-armed robot; the dads had lunch and talked about virtual-reality therapies and rehab gadgets that seemed to have potential. “He was so gracious with his time. It’s just tragically ironic that this happened to him [a few] years later, and he was calling me,” Barney said. “The thing that I kept going back to was that it was going to be okay, that we were happy, and that he’s going to get there but it’s going to be just unbelievably tough.”

Charley, now 10 years old and half a decade past her accident, doesn’t walk or talk, but even early on she had a way of saying yes (looking up) and a way of saying no (a shake of the head). At Dell Children’s, she began using an eye-tracking device, selecting icons with her gaze and forming rudimentary sentences that way. She’s since learned to read and write, and now sends texts to Barney while he’s working. “She tells you what she did that day; she tells you what made her mad or what was funny,” he said. “That’s how Charley communicates.”

For Barney and his wife, Shannon, the Bill Gates approach to rehabilitation has been successful. Two years ago, their nonprofit arranged to purchase one of the most expensive and widely used rehabilitation robots on the market—a half-million-dollar machine called the Lokomat, meant to teach people with brain injuries how to walk again—and installed it at a clinic several miles up the road from James’s lab at UT. Charley has been training with it ever since. “We know it’s been good for her,” Barney told me.

James and Lindsay aren’t wealthy, but resources haven’t been an issue. Given their backgrounds and milieu, they can choose among a wide variety of interventions: stem-cell treatments, “diving” sessions in an oxygen tank, infrared-laser therapy, robotic exoskeletons. But as scientists, they’ve been discouraged by the paucity of data on whether any of these approaches really work. Clinical studies in the field are pretty scarce, even when it comes to the most common neural injuries, in adults who suffer strokes. Far less research has been done on injured children; for those like Livie, with damage spread across the brain, delivered by a violent blow, there’s almost nothing.

“How do you make an informed, educated decision?” Lindsay said to me. “It’s a huge challenge, and I think we have a harder time because we want to have some sort of scientific rationale.” James agreed. “It’s very easy, as a scientist, to just be skeptical of everything,” he said. “But as a parent, you need to have some optimism, and you need to take leaps of faith.”

I first spoke with James and Lindsay in the spring, as the first anniversary of Livie’s accident approached. They’d been thinking back across their 12-month stretch of impossible calamity. The pandemic had meant that Livie (and her parents) couldn’t have any visitors when she was in the hospital, and that every nurse or therapist who came to see her was also, to some degree, a mortal threat. Then came the winter storms in February and the power crisis and blackouts. Livie’s medications grew warm inside the fridge, and the pump they used to give her feedings almost sapped its battery.

Now the Persian silk trees in the neighborhood were flowering again, along with Dr. Iris in the Sulzers’ yard. On James’s and Lindsay’s smartphones, auto-generated galleries of snapshots taken “One Year Ago Today” approached a crushing turning point: There was Livie riding on her wheeled giraffe; Livie playing in the yard; Livie with her brothers; Livie in a coma. It was time for an accounting of all the things they’d tried to do to help with her recovery, and of how far she’d really come.

During Livie’s year of rehabilitative therapy, she’d cycled through dozens of commercial products and devices from James’s lab—bungee-cord harnesses, wireless electrodes, eye-tracking games, and so on—but none of them was perfect. None of them was even close. Her disabilities remain both diverse and severe: Like Charley, Livie cannot speak. She has a way of saying yes—she pumps her right arm up and down, like she’s hitting an imaginary button labeled MORE—but her no, a left-foot stomp, is somewhat less reliable. She can walk, a bit, when someone’s holding her, but her limbs have been weakened by osteoporosis. Her cognitive disabilities appear to be significant, but they’re tricky to assess given the limitations of her movements. “We don’t know what she’ll recover,” Lindsay said. “We don’t know to what point she’ll get, when that will be, will she ever talk again, any of those things. We don’t know.”

Read: The brain that wasn’t supposed to heal

James, in particular, began to fixate on the mounting failures and uncertainties, and the ways his field had come up short. He began to wonder if the whole idea of rehab engineering—its deepest motivations—might be off the mark. Many of the problems Livie encountered had to do with a gadget’s usability: It might be cumbersome to set up, or hard to learn, or prone to breaking. Locating one of Livie’s nerves with an electrical stimulator took 10 minutes, for example, and then you couldn’t really tell whether the pulse was doing much to help her straighten out her foot. Given Livie’s crowded schedule of care and treatments, even modest hiccups of this kind could make an intervention useless. “It’s so frustrating, because all these ideas that I think are awesome wind up sucking,” James told me. “I went into this to build devices to help people, but I never considered that building devices might not be the answer.”

When they were at Northwestern, James and Lindsay had been trained to think of failure as a skill: To fail correctly—to do it “in an interesting way”—you had to sift through the rubble of your disappointment and ask, How did things go wrong, and why did we expect a different outcome? As the couple’s fear and frustration over Livie’s progress grew, these very questions came into their minds: What happened here, and why? The field of rehab engineering hadn’t done that much for Livie. But now it seemed like maybe Livie could do something for the field.

By the time we first spoke, James and Lindsay had written up their observations. They were trying, as James would later say, to draw some meaning—and perhaps a fragment of relief—from what had been “this huge, loud, overwhelming noise” inside their heads. On April 7, the Journal of NeuroEngineering and Rehabilitation had published the remarkable result: a co-authored manifesto on the principles of technology design, but also a peer-reviewed portrait of their suffering. The paper, “Our Child’s TBI: A Rehabilitation Engineer’s Personal Experience, Technological Approach, and Lessons Learned,” starts with brief biographies of James and Lindsay, and then a recitation of what happened to their daughter, referred to only as “B”—a private reference to her nickname, Boogie.

The text that follows is staggeringly personal. In a section called “Technologies Explored,” James and Lindsay note how “the feeling of helplessness and rollercoaster of emotions is often temporarily assuaged with new treatments and devices.” A subsection on “Emotional Trauma” begins like this:

The heart of the paper, though, its message and its purpose, comes later on, where it turns from pain to disillusionment. There the language slips from “we” to “I,” from Livie’s parents’ point of view to her father’s, the rehab engineer’s. Like many in the field, he wrote, he’d never bothered to understand the very tasks that he was trying to automate. At the Rehabilitation Institute of Chicago, patients were seen only one or two floors away from James’s lab, yet he did not take that walk downstairs; he did not watch the therapists at work, or see the role they played in troubleshooting problems and providing motivation, empathy, and guidance. “We kind of have this attitude that therapists are just too stupid to know the technology, or they’re afraid of losing their jobs, or they just don’t understand that robots can do what they do,” James told me. “But that’s not true. Therapists do a lot of things that robots will never be able to do.” It was just as wrong, the paper said, for engineers to disregard the most basic tools of rehab: a mat, a ball, a table. These possess the virtues of sturdiness and simplicity, virtues that robotics engineers too often overlook.

It wasn’t that James had given up on making gadgets—not at all. But he’d come to realize that innovation should be tailored to human needs, not dictated merely by technological possibility. This was the central lesson, as the paper put it, of James and Lindsay’s “immersive experience” of traumatic brain injury. In Figure 1 they provide a set of guidelines for rehab engineering: 11 threshold questions that should be answered for any new device. “Can the task be accomplished using simpler technology?” reads the first. “Can it be set up and cleaned up as quickly as a bench and some toys?” “Does it require expertise to operate properly?” “Can it be combined with other therapies?” If a new technology cannot pass these tests, it may not be worth the time and effort to develop.

Read: I know the secret to the quiet mind. I wish I’d never learned it.

“I’ve been one of those people promoting high-tech stuff for a really long time, just to explore and see if it works, because it could be revolutionary,” James said. Now he was pleading for a new approach, and his message was getting through. Two prominent professors in the field told me that they’ve made “Our Child’s TBI” required reading in their labs. James and Lindsay have been asked to give several talks related to their paper; one invitation, for them to serve as keynote speakers at the discipline’s most prestigious annual meeting, RehabWeek Virtual ’21, said that “Figure 1 of this paper should be printed out and hung up over the desk or bed or kitchen table of every person working in our field. Eventually, it should become ingrained into our brains and become second nature.”

At the Spero Rehab Clinic in central Austin, an elderly man stepped over tiny hurdles as he made his way around the gym. There were other basic tools for clients’ use: a tilting table, parallel bars, a skateboard. But the site’s clinical director, Brooke Aarvig, was showing me the bigger-ticket items. She took me over to an exoskeleton for training movements of the arm and hand, called the ArmeoSpring, and then to a VR setup and a motorized “hippotherapy” bench with stirrups, which resembled the world’s tamest mechanical bull. And then finally to the clinic’s prize machine, in the center of the workout floor: the Lokomat from Project Charley, an eight-foot-high, 2,200-pound marvel of rehabilitative tech—a treadmill with a hanging harness and robotic legs.

A young woman with chin-length hair and severe weakness on one side of her body was being strapped into the harness by a technician with a clipboard. He wrapped her left hand to the armrest with a bandage and braced her legs in the metal frames of the robotic legs; the machine then hoisted her up about six inches. A moment later, she was walking—or the Lokomat was walking. You couldn’t really tell. When the movement started, she was still briefly suspended well above the treadmill’s belt, loping queerly through the air.

“Interactive robotic therapists” emerged in the early 1990s, when growing interest in the science of “neuroplasticity”—the idea that the brain forms new connections in the course of learning or recovery—buoyed hope that the same process could be mechanized and made efficient. Robotic therapists could, in theory, help patients move in ways they couldn’t on their own and then repeat those movements many times, while also measuring their progress. Each repetition would trigger currents in the brain and feed into a rehabilitative flow of neural signals. Eventually, the theory went, these would carve out new channels in the cortex—or reopen ones that had closed.

By the end of the decade, a team at MIT led by the mechanical engineers Neville Hogan and Hermano Igo Krebs was running small clinical trials with what they called the MIT-Manus: a robotic arm that could assist (and challenge) a patient’s own arm through different exercises. In a 1999 paper describing their accomplishments, the researchers boasted that “robotics and information technology can provide an overdue transformation of rehabilitation clinics from primitive manual operations to more technology-rich operations.” The Lokomat arrived a couple of years later.

Larger trials of the rehab robots turned out disappointing findings, though. A major study from 2010, in The New England Journal of Medicine, looked at patients who had suffered strokes and had impairment in their upper limbs. Across a 12-week intervention, those who’d been treated with the MIT-Manus robot did no better—though, to be fair, no worse—than those who’d gotten standard care. Another large study, published in The Lancet in 2019, reached a similar conclusion: 12 weeks of training on an MIT-Manus provided no improvement in upper-limb function for people recovering from a stroke compared with normal therapy. Results have been slightly better for lower-limb machines: A comprehensive survey of the research literature, published in 2020, looked at 62 studies of “electromechanical- and robot-assisted gait-training devices” for people who had trouble walking after a stroke—including 25 trials involving the Lokomat—and concluded that the use of these machines (especially in the first few months post-impairment) increased people’s odds of being able to walk independently.

“The hype that robots are going to fix everything has not borne out at this point,” says Theresa Hayes Cruz, the director of the National Center for Medical Rehabilitation Research at the National Institutes of Health, and one of James’s former grad-school classmates. “I think what we’ve learned is that therapists bring a lot more than just the physical movement to a patient. There’s that psychosocial interaction, motivation, things like that.”

But rehab roboticists suggest that some clinicians may have been too quick to abandon a good idea. Physical therapists sometimes worry that the technology is “too complex” for them, says Arun Jayaraman, the director of the Max Näder Center for Rehabilitation Technologies and Outcomes Research at Shirley Ryan AbilityLab (as the Rehabilitation Institute of Chicago, where James did his graduate studies, is now known). They may also think of robots as a threat. “It’s the same issue with any automation,” he said. “Factory workers are scared of robots in the car industry, or in any industry, because they think the robots are taking away their jobs.” Jayaraman concedes that the first generation of these devices was a little glitchy, but he says that’s just how innovation works: You start with something, and then you make it better. “If you didn’t do that first version of the iPhone, then you’re not going to get the iPhone 12 Pro Max.”

At Spero Rehab, the Lokomat seems to be working well for certain clients. A therapist might get tired while assisting a patient on a single lap around the gym, whereas the robot can help the same person do the equivalent of six to 10 laps. You can also dial up the robot’s body-weight support over the course of a session, to keep the client going even as they tire out. It’s like getting spotted when you’re lifting weights—a way to land some extra reps.

But over the past two years, some therapists have grown wary of the Lokomat’s power, and its magnetism. “This machine could be walking for you,” one told me. Its motors can take on much of the exercise themselves. Indeed, the woman who was on the treadmill when I made my visit to Spero had been training in this very configuration with the guidance level turned up all the way, to 100 percent. That meant she had to use her muscles only to help support her body weight, while the robot worked her legs on the treadmill like a marionette’s. There’s little reason to believe that this form of repetition, especially when it’s limited to movement in one direction and of one specific type, leads to any substantive motor learning or recovery, James told me at one point. “What we’re supposed to know, as scientists, is that if you guide someone to do a movement that you already know how to do, it doesn’t help.”

Even when that guidance knob is turned down, and even when a person is really moving for themself, expectations for the Lokomat can run too high. According to a clinician who has worked with the device, people sometimes get so enamored with the mere idea of using a fancy robot—so addicted to its coolness and its automation—that they barely notice when their training doesn’t yield results. Some people “really, really” benefit from using the Lokomat, the clinician told me, but others have been working with it for years without any signs of progress. “We ask them to decrease how often they do it,” or else tell the insurance provider it’s time to discontinue therapy. Yet these people still end up coming back, and paying out of pocket for another hour on the robot, and then another, and another—more time spent without the hands-on work of traditional therapy. “People think this machine is a magic fix,” the clinician said, “and it’s not.”

The first thing you see, upon entering the Sulzers’ house, is a nine-by-nine metal frame suspended above a pair of sofas with a harness hanging down and space for Livie to walk. There’s a stander, too—a large device that helps Livie practice being upright. (James has been working on pressure sensors for the stander’s foot plate, hooked up to his smartphone via Bluetooth, to measure how Livie is balancing her weight.) When I first arrived, Livie was positioned on the rug, working with a physical therapist. She looked up—as she always does whenever someone new comes in—and gave a slow and happy-sounding greeting, a cross between a groan and a laugh. She has curly brown hair and thick, dramatic eyebrows of the sort that people try to fake. She smiles all day long, with mischief and enthusiasm, and delights in having James lean in to kiss her cheeks. “I’m going to chomp you,” he says. “I’m going to give you the chomps!”

Several of James’s students from UT also came by that day, to test out the latest prototype of Livie’s ride-on car. It’s a jeep for kids, pink with big black wheels—the kind of thing you might get at Walmart for $300, but with its accelerator pedal rewired to a hand-operated console. The students carried it out into the yard, 15 or 20 feet from the screened-in porch, almost to the very spot where the tree branch had fallen the year before. James lifted Livie, in her rainbow dress and sparkle sneakers, into the seat.

The team designed the console to make it work in lots of different ways. When Livie grasps and yanks a yellow ball, the car moves forward. With a different module in place, the controller rotates like a dial, to help Livie practice supination of her wrist. (“If a therapy is going to be really useful, it’s got to do multiple things at once,” James told me. “Everything’s got to be a Swiss Army knife.”) But after grabbing at the ball a few times without success, Livie lost energy, or patience. Her head began to droop a little farther to the left. She let go.

The idea they’re working from—to hack a ride-on car for use by kids with disabilities—comes from Cole Galloway, a professor of physical therapy at the University of Delaware, whom James consulted after Livie’s injury. In 2007, Galloway started GoBabyGo, a project to teach families how to create their own mobility devices at very low cost. Galloway’s cars aren’t supposed to help a kid with a brain injury get “back to normal” in their basic motor functions, he told me. They’re not designed, like fancy rehab robots, to fix problems within a narrow frame of brain function. Rather, they’re meant to give kids a way to move around and interact with their peers, without the stigma that might attach to being in a powered wheelchair. A medical model of recovery often gives way to a social one, he said. “Most people eventually go, ‘I no longer care as much about impairment level or function. I want my kid to be a citizen. I want them to be invited to a birthday party.’ ” But he wondered whether James was ready for that step. “He’s very much an engineer. He’s a builder at heart, meaning ‘I fix things, and Livie is fixable.’ ” To Galloway, that mindset is a trap.

James and Lindsay’s goal right now, they told me, is for Livie to develop “functional independence,” by which they seem to mean being able to take care of herself, more or less, with the help of some assistive technologies. They acknowledged to me that this may not be realistic, but they’ll keep pushing for that outcome—with new treatments and devices, and more therapy—until they’re certain that her window of recovery has closed.

Livie’s ride-on car, as James conceived it, reflects a focus on improving basic skills. It’s not designed to be a way for Livie to move across the yard (even if that would be nice). Instead it’s meant to make her more excited about an exercise that could, in theory, help her brain to fix itself. James and his students have made the car into a fancy version of the “Baby Shark” machine: If she pulls or twists the knob, she gets to zoom around! And if she gets to zoom around, she’ll want to pull or twist the knob some more. The problem was, Livie didn’t seem that interested. “I don’t know,” James said after several failed attempts. “Maybe she doesn’t care about driving the car.”

Back in the house, it was time for Livie’s lunch. A nurse put her in the stander in the living room and set up an iPad for her entertainment while a pouch of chickpea formula emptied into her feeding tube.

James disappeared into his office and came out with a bin. He spilled its contents out onto the rug: an assortment of devices that he’d tried so far. He showed me a headband with symbols on it, meant to help a computer track Livie’s head motion, and a harness called an Upsee, in which your kid is strapped to your hips and legs and feet in such a way that the two of you can walk in tandem. There was a VR headset, a spinal-cord stimulator, and a wireless, force-tracking computer interface called FitMi. At his lab I’d seen a sensor he was working on with a colleague in the textiles-and-apparel department—it can be sewn into the collar of a shirt. James wants to use it to measure Livie’s swallowing as she relearns to eat. He also had a plan to build a sheath that gently strokes her back, on the theory that it might reduce excessive muscle contraction.

I asked him whether he might decide, at any point, that enough is enough—that after having tried so many gadgets without significant success, it was time to stop. “Am I gonna give up? No. From what I’ve seen, there’s always improvements to be made,” he said. “At the same time, I do have to consider at what point does this stop benefiting Livie, and that’s a tough question to answer, and I haven’t confronted that yet.” He went on, after a moment’s thought: “That’ll be a tough, tough question to ask myself, and hopefully, I can have the courage to say, ‘Okay, you know what? We’re better off doing other things rather than trying to help her that way.’ ”

During lunch, Livie’s iPad cycled through a set of family photographs and videos. Like the pictures on the Sulzers’ walls, they were all from before the accident. Many showed Livie herself. That’s what she likes to see when she’s strapped into the stander, waiting for her food, James said. He’d found that the pictures often work as motivation for her therapy, so he’d programmed some of them into her rehab devices too.

“What’s sad is we’ve watched these videos so many times, the memories are the videos,” he said. At first, he used to look at them and think, Oh, hey, there’s Livie. Now he feels as though he’s looking at a different person. “It’s like I’ve developed a new relationship with her,” he told me. “Now it’s like I have two daughters, in a way. One that passed away, and now this one.”

He’d said something similar the first time we talked, about seeing Livie split in half—as one girl in the pictures and videos, and another in the stander. “If I thought of it as that she died, and that there’s this new child here, it felt like a release,” he said. “I guess it’s hard to explain why, but it made it easier.” And yet James and Lindsay aren’t quite ready to let go of the girl in the pictures. “Emotionally, we’re not resolved in this at all,” James said. “There are flashes she shows of her former self, and we still hold out hope that she’s going to come back.”

The new Livie finished up her meal. The old Livie hovered like a ghost.

At dinner that evening, Livie’s brother Noah, now 7, sat to my right. He was biting into the crust of his pizza, making holes that looked like stars and boats and other shapes—a constant game for him, his parents said, finding pictures in his food. He pointed to a little glob of mozzarella that had several slices of black olive sticking out of it at different angles. “Hey, it’s like that building in Australia,” he observed.

Before the accident, Livie was Noah’s best friend. He still plays with her, and they like to lie in bed together and watch TV. But Noah struggles to understand what’s going on: whether Livie will get back to how she was before, and how long that getting-back might take. “He’s really gotten the shit end of the stick in this whole deal. He’s out of control of everything,” James had told me. Now Noah sat there while the grown-ups talked, studying his cheese-and-olives. “It’s the Sydney Opera House,” he said. “It’s the Austin Olive House.”

The month before, at one of James and Lindsay’s academic talks, Lindsay had brought up something Noah had asked her: “How come in stories there’s always a happy ending, and that’s not true in real life, like with Livie?”

No one had an answer to that question. “I’m worried that you don’t have a story,” James had told me earlier that day. “You don’t have an ending.” He’d hoped to find some resolution in sending out a message to his field. “It does feel good that people say, ‘I read your paper and I really liked it and I’m integrating it into my research,’ ” he said after dinner, “but I would say that in the end, it still feels kind of empty, because it doesn’t change our situation.” He wondered if he’d ever really had a message to begin with. “I mean, it’s like we’re all just putzing around trying to find what might work, and no one knows what might work, and it’s just a big mess. That’s not a story, and that’s what bothers me.”

James and I had spent many hours talking about Livie, and all the different ways he’d tried to help her. But Livie is not the only person who needs rehabilitation. Everyone in the family was traumatized by the accident, James and Lindsay wrote in their paper. Everyone in the family needs to get better; everyone is moving through their own windows of recovery, unsure of their prognoses, trying everything they can. There might never be a magic fix for Livie’s injury, and there might never be a magic fix for James’s, or Lindsay’s, or her brothers’, either. Each of them can only gird themselves for the grueling journey to a better place—or for the grueling task of making peace with where they are. Each of them can only use whatever tools they have.

For James, those tools are in his lab. Making gadgets has been therapeutic, even when those gadgets break or go awry. “It helps me feel like I have agency over her recovery, and that gives me hope,” he said.

It was almost time for the kids to go to sleep. James and Lindsay were busy attending to the boys, so I sat down next to Livie in her room. She was in her bed, with green and pink pool noodles pressed up against the rail to prevent a fall. Her feeding tube was hanging from a pole. The nurse had said that Livie really likes a book called Ada Twist, Scientist, so I took it from the shelf and read aloud.

I looked up from the page and saw that Livie was staring at the wall. James had said that she sometimes has spells of inattention, and that he and Lindsay think they might be seizures. But when I paused to ask Livie if she wanted me to read some more, a smile quickly stretched across her face. She groaned at me and pumped her right arm up and down.

Yes, she signaled from the bed. Go on. The story isn’t over yet.

This article appears in the November 2021 print edition with the headline “The Engineers’ Daughter.”