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."
When David M. Kurtz was doing his clinical fellowship at Stanford University Medical Center in 2009, specializing in lymphoma treatments, he found himself grappling with a question no one could answer. A typical regimen for these blood cancers prescribed six cycles of chemotherapy, but no one knew why. "The number seemed to be drawn out of a hat," Kurtz says. Some patients felt much better after just two doses, but had to endure the toxic effects of the entire course. For some elderly patients, the side effects of chemo are so harsh, they alone can kill. Others appeared to be cancer-free on the CT scans after the requisite six but then succumbed to it months later.
"Anecdotally, one patient decided to stop therapy after one dose because he felt it was so toxic that he opted for hospice instead," says Kurtz, now an oncologist at the center. "Five years down the road, he was alive and well. For him, just one dose was enough." Others would return for their one-year check up and find that their tumors grew back. Kurtz felt that while CT scans and MRIs were powerful tools, they weren't perfect ones. They couldn't tell him if there were any cancer cells left, stealthily waiting to germinate again. The scans only showed the tumor once it was back.
Blood cancers claim about 68,000 people a year, with a new diagnosis made about every three minutes, according to the Leukemia Research Foundation. For patients with B-cell lymphoma, which Kurtz focuses on, the survival chances are better than for some others. About 60 percent are cured, but the remaining 40 percent will relapse—possibly because they will have a negative CT scan, but still harbor malignant cells. "You can't see this on imaging," says Michael Green, who also treats blood cancers at University of Texas MD Anderson Medical Center.
The new blood test is sensitive enough to spot one cancerous perpetrator amongst one million other DNA molecules.
Kurtz wanted a better diagnostic tool, so he started working on a blood test that could capture the circulating tumor DNA or ctDNA. For that, he needed to identify the specific mutations typical for B-cell lymphomas. Working together with another fellow PhD student Jake Chabon, Kurtz finally zeroed-in on the tumor's genetic "appearance" in 2017—a pair of specific mutations sitting in close proximity to each other—a rare and telling sign. The human genome contains about 3 billion base pairs of nucleotides—molecules that compose genes—and in case of the B-cell lymphoma cells these two mutations were only a few base pairs apart. "That was the moment when the light bulb went on," Kurtz says.
The duo formed a company named Foresight Diagnostics, focusing on taking the blood test to the clinic. But knowing the tumor's mutational signature was only half the process. The other was fishing the tumor's DNA out of patients' bloodstream that contains millions of other DNA molecules, explains Chabon, now Foresight's CEO. It would be like looking for an escaped criminal in a large crowd. Kurtz and Chabon solved the problem by taking the tumor's "mug shot" first. Doctors would take the biopsy pre-treatment and sequence the tumor, as if taking the criminal's photo. After treatments, they would match the "mug shot" to all DNA molecules derived from the patient's blood sample to see if any molecular criminals managed to escape the chemo.
Foresight isn't the only company working on blood-based tumor detection tests, which are dubbed liquid biopsies—other companies such as Natera or ArcherDx developed their own. But in a recent study, the Foresight team showed that their method is significantly more sensitive in "fishing out" the cancer molecules than existing tests. Chabon says that this test can detect circulating tumor DNA in concentrations that are nearly 100 times lower than other methods. Put another way, it's sensitive enough to spot one cancerous perpetrator amongst one million other DNA molecules.
"It increases the sensitivity of detection and really catches most patients who are going to progress," says Green, the University of Texas oncologist who wasn't involved in the study, but is familiar with the method. It would also allow monitoring patients during treatment and making better-informed decisions about which therapy regimens would be most effective. "It's a minimally invasive test," Green says, and "it gives you a very high confidence about what's going on."
Having shown that the test works well, Kurtz and Chabon are planning a new trial in which oncologists would rely on their method to decide when to stop or continue chemo. They also aim to extend their test to detect other malignancies such as lung, breast or colorectal cancers. The latest genome sequencing technologies have sequenced and catalogued over 2,500 different tumor specimens and the Foresight team is analyzing this data, says Chabon, which gives the team the opportunity to create more molecular "mug shots."
The team hopes that that their blood cancer test will become available to patients within about five years, making doctors' job easier, and not only at the biological level. "When I tell patients, "good news, your cancer is in remission', they ask me, 'does it mean I'm cured?'" Kurtz says. "Right now I can't answer this question because I don't know—but I would like to." His company's test, he hopes, will enable him to reply with certainty. He'd very much like to have the power of that foresight.
The white two-seater car that rolls down the street in the Sorrento Valley of San Diego looks like a futuristic batmobile, with its long aerodynamic tail and curved underbelly. Called 'Sol' (Spanish for "sun"), it runs solely on solar and could be the future of green cars. Its maker, the California startup Aptera, has announced the production of Sol, the world's first mass-produced solar vehicle, by the end of this year. Aptera co-founder Chris Anthony points to the sky as he says, "On this sunny California day, there is ample fuel. You never need to charge the car."
If you live in a sunny state like California or Florida, you might never need to plug in the streamlined Sol because the solar panels recharge while driving and parked. Its 60-mile range is more than the average commuter needs. For cloudy weather, battery packs can be recharged electronically for a range of up to 1,000 miles. The ultra-aerodynamic shape made of lightweight materials such as carbon, Kevlar, and hemp makes the Sol four times more energy-efficient than a Tesla, according to Aptera. "The material is seven times stronger than steel and even survives hail or an angry ex-girlfriend," Anthony promises.
Co-founder Steve Fambro opens the Sol's white doors that fly upwards like wings and I get inside for a test drive. Two dozen square solar panels, each the size of a large square coaster, on the roof, front, and tail power the car. The white interior is spartan; monitors have replaced mirrors and the dashboard. An engineer sits in the driver's seat, hits the pedal, and the low-drag two-seater zooms from 0 to 60 in 3.5 seconds.
It feels like sitting in a race car because the two-seater is so low to the ground but the car is built to go no faster than 100 or 110 mph. The finished car will weigh less than 1,800 pounds, about half of the smallest Tesla. The average car, by comparison, weighs more than double that. "We've built it primarily for energy efficiency," Steve Fambro says, explaining why the Sol has only three wheels. It's technically an "auto-cycle," a hybrid between a motorcycle and a car, but Aptera's designers are also working to design a four-seater.
There has never been a lack of grand visions for the future of the automobile, but until these solar cars actually hit the streets, nobody knows how the promises will hold up.
Transportation is currently the biggest source of greenhouse gases. Developing an efficient solar car that does not burden the grid has been the dream of innovators for decades. Every other year, dozens of innovators race their self-built solar cars 2,000 miles through the Australian desert.
More effective solar panels are finally making the dream mass-compatible, but just like other innovative car ideas, Aptera's vision has been plagued with money problems. Anthony and Fambro were part of the original crew that founded Aptera in 2006 and worked on the first prototype around the same time Tesla built its first roadster, but Aptera went bankrupt in 2011. Anthony and Fambro left a year before the bankruptcy and went on to start other companies. Among other projects, Fambro developed the first USDA organic vertical farm in the United Arab Emirates, and Anthony built a lithium battery company, before the two decided to buy Aptera back. Without a billionaire such as Elon Musk bankrolling the risky process of establishing a whole new car production system from scratch, the huge production costs are almost insurmountable.
But Aptera's founders believe they have found solutions for the entire production process as well as the car design. Most parts of the Sol's body can be made by 3D printers and assembled like a Lego kit. If this makes you think of a toy car, Anthony assures potential buyers that the car aced stress tests and claims it's safer than any vehicle on the market, "because the interior is shaped like an egg and if there is an impact, the pressure gets distributed equally." However, Aptera has yet to release crash test safety data so outside experts cannot evaluate their claims.
Instead of building a huge production facility, Anthony and Fambro envision "micro-factories," each less than 10,000 square feet, where a small crew can assemble cars on demand wherever the orders are highest, be it in California, Canada, or China.
If a part of the Sol breaks, Aptera promises to send replacement parts to any corner of the world within 24 hours, with instructions. So a mechanic in a rural corner in Arkansas or China who never worked on a solar car before simply needs to download the instructions and replace the broken part. At least that's the idea. "The material does not rust nor fatigue," Fambro promises. "You can pass the car onto your grandchildren. When more efficient solar panels hit the market, we simply replace them."
More than 11,000 potential buyers have already signed up; the cheapest model costs around $26,000 USD and Aptera expects the first cars to ship by the end of the year.
Two other solar carmakers are vying for the pole position in the race to be the first to market: The German startup Sono has also announced it will also produce its first solar car by the end of this year. The price tag for the basic model is also around $26,000, but its concept is very different. From the outside, the Sion looks like a conservative minivan for a family; only a closer look reveals that the dark exterior is made of solar panels. Sono, too, nearly went bankrupt a few years ago and was saved through a crowdfunding campaign by enthusiastic fans.
Meanwhile, Norwegian company Lightyear wants to produce a sleek solar-powered luxury sedan by the end of the year, but its price of around $180,000 makes it unaffordable for most buyers.
There has never been a lack of grand visions for the future of the automobile, but until these solar cars actually hit the streets, nobody knows how the promises will hold up. How often will the cars need to be repaired? What happens when snow and ice cover the solar panels? Also, you can't park the car in a garage if you need the sun to charge it.
Critics, including students at the Solar Car team at the University of Michigan, say that mounting solar panels on a moving vehicle will never yield the most efficient results compared to static panels. Also, they are quick to point out that no company has managed to overcome the production hurdles yet. Others in the field also wonder how well the solar panels will actually work.
"It's important to realize that the solar mileage claims by these companies are likely the theoretical best case scenario but in the real world, solar range will be significantly less when you factor in shading, parking in garages, and geographies with lower solar irradiance," says Evan Stumpges, the team coordinator for the American Solar Challenge, a competition in which enthusiasts build and race solar-powered cars. "The encouraging thing is that I have seen videos of real working prototypes for each of these vehicles which is a key accomplishment. That said, I believe the biggest hurdle these companies have yet to face is successfully ramping up to volume production and understanding what their profitability point will be for selling the vehicles once production has stabilized."
Professor Daniel M. Kammen, the founding director of the Renewable and Appropriate Energy Laboratory at the University of California, Berkeley, and one of the world's foremost experts on renewable energy, believes that the technical challenges have been solved, and that solar cars have real advantages over electric vehicles.
"This is the right time to be bullish. Cutting out the charging is a natural solution for long rides," he says. "These vehicles are essentially solar panels and batteries on wheels. These are now record low-cost and can be built from sustainable materials." Apart from Aptera's no-charge technology, he appreciates the move toward no-conflict materials. "Not only is the time ripe but the youth movement is pushing toward conflict-free material and reducing resource waste....A low-cost solar fleet could be really interesting in relieving burden on the grid, or you could easily imagine a city buying a bunch of them and connecting them with mass transit." While he has followed all three new solar companies with interest, he has already ordered an Aptera car for himself, "because it's American and it looks the most different."
After taking a spin in the Sol, it is startling to switch back into a regular four-seater. Rolling out of Aptera's parking lot onto the freeway next to all the oversized gas guzzlers that need to stop every couple of hundreds of miles to fill up, one can't help but think: We've just taken a trip into the future.