In the not-too-distant future, a “planetary” computer will be able to create a computational model of your body, with the ability to run simulations of your health and to anticipate chronic disease before you show any symptoms.
This is the direction we’re headed, according to Larry Smarr, founding director of the California Institute of Telecommunications & Information Technology at UC San Diego. While his expertise lies in computer networks and infrastructure, Smarr has emerged as a de facto leader in quantified health—largely due to his relentless curiosity about his own health. In 2011, Smarr diagnosed his own Crohn’s disease long before he showed any symptoms.
Smarr laid out his vision for a looming transformation in healthcare at the recent Xconomy San Diego Forum on the Human Impact of Innovation. The advances making this transformation possible include next-generation genome sequencing (where the cost of sequencing has been dramatically plunging), combined with new biomedical data from a host of sensors and blood tests.
Smarr’s work has contributed to a growing realization that the trillions of microbes in the human gut play a crucial role in maintaining human health. In his talk, Smarr dryly noted that the density of information in a fecal sample works out to roughly 1 billion microbes per gram of stool, encompassing a treasure trove of data that can be collected non-invasively every day. (The microbiome collectively contains 3.3 million genes—roughly 100 times more than the 23,000 genes of the human genome.)
This massive river of data would overwhelm conventional computing technology, but Smarr said a revolution in machine learning—based on a radically new (non-von Neumann) chip design—is now making it possible to analyze big data streams and monitor human health in unprecedented ways.
Google’s first machine-learning chip is 30 times faster than conventional computer processors and graphics processors, Smarr said. As a result, Google and companies like San Diego-based KnuEdge, which is poised to roll out its neural-based AI technology sometime this year, and Intel’s San Diego-based Artificial Intelligence Products Group (headed by former Nervana Systems CEO Naveen Rao) are moving toward the development of what he called “planetary” computers with Web-based AI capabilities.
Smarr said the promise of microbiome data as a way to predict chronic disease was demonstrated late last year by scientists at UC San Diego. Using machine learning technology, the team said, they had successfully trained a computer system to tell the difference between a person with a “healthy” intestine from someone with inflammatory bowel disease by analyzing the genetic makeup of the microbes in their gut.
Machine learning techniques also are being applied to X-rays and other medical imaging used in medical diagnostics. With such technology, Smarr said, “the worst radiologist with this kind of machine support would be better than the best radiologist without this support.”
Following Smarr’s presentation, other luminaries discussed their progress in healthtech:
—Jeff Hawkins, vice president for reproductive and genetic health at San Diego-based Illumina (NASDAQ: ILMN), discussed how genome sequencing is being used to analyze fetal DNA in maternal blood samples—providing a more accurate and less-risky replacement for amniocentesis.
—Steven Steinhubl, a cardiologist and director of digital medicine at the Scripps Translational Science Institute, gave an overview of a 40-year effort to gather medical and genomic data on 1 million Americans “to better understand what makes each of us unique.”
—Clayton Lewis, the co-founder and CEO of Seattle-based Arivale, described how Arivale combines personal genomic and biomedical data with health, nutrition, and fitness “coaching” to help individuals avoid chronic diseases like diabetes. At some point in the future, Lewis said, Arivale will scan individuals for 60 diseases, assess how close they are to the tipping point for each disease, and prepare a plan for treating or reversing disease in a meaningful way.
Machine learning was a recurring theme throughout the forum, with Intel’s Urs Köster predicting that AI technology already is transforming healthcare, finance, retailing, transportation, energy, government, and other industries.
AI technology also has enabled the robotics technologies behind self-driving vehicles to advance more rapidly than expected, according to Henrik Christensen, director of the Institute for Contextual Robotics at UC San Diego.
In a presentation on the roadmap for self-driving cars, Christensen predicted that autonomous, self-driving cars would be driving on U.S. roads by 2025. But significant technology hurdles must still be overcome, including the ability of machine vision systems to “see” through snowstorms and other tough weather conditions.
“People are being way too optimistic about self-driving technology, and how quickly it’s going to be in place,” Christensen said.
KnuEdge founder and CEO Dan Goldin said he realized a critical need for a new type of machine intelligence while considering a manned mission to Mars during the nine years he led NASA (from 1992 to 2001). The 40-minute lag in communications between the Earth and Mars-bound spacecraft meant “you have no mission control,” Goldin said.
After retiring from NASA, Goldin set out to create low-power processing technology that was modeled on the highly efficient neural processes of the human brain. The first KnuEdge chip, built with an older manufacturing technology, features 256 cores—inspired by neuron-like cells—on a single chip. The company, stealthy for more than a decade, said last year it also had developed “LambdaFabric” technology that is ideally suited for massively parallel processing applications. KnuEdge says its fabric, based on neurobiological principles, enables its multicore Hermosa processor to connect to as many as 512,000 other Hermosa processors with a latency (interaction lag) of just 400 nanoseconds.
KnuEdge is now preparing to provide cloud-based machine intelligence as a service. As Goldin put it, “We want to enable anyone to provide an AI experience to anyone, anywhere,” and he predicted that San Diego will be at the center of “real machine intelligence” within five years. Asked in an e-mail afterward for an explanation of what he meant by “real machine intelligence,” Goldin declined to elaborate.
(Special thanks to Joyce Thorne for providing her notes from the forum.)
Bruce V. Bigelow was the editor of Xconomy San Diego from 2008 to 2018. Read more about his life and work here.