Blog Posts for Erich S. Huang, MD, PhD

Hard Lessons: What the World of Health AI Can Learn from Aviation

January 6, 2021

Erich S. Huang, MD, PhD
Director, Duke Forge & Duke Crucible

A Boeing 737 Max 8 airliner in flight amid blue sky and clouds at a British airshow. Image credit: Oleg V. Belyakov via Wikipedia (CC BY-SA 3.0)

The morning of October 29, 2018, was a balmy 80 degrees in Jakarta, Indonesia, with a 5-knot wind blowing out of the southwest. Under clear skies, Lion Air Flight 610 pushed back from the gate at Soekarno-Hatta International Airport, bound for Depati Amir Airport on Bangka Island. On board were 188 souls: two pilots, five flight attendants, and 181 passengers.

During takeoff, as the Boeing 737 Max 8 accelerated to rotation speed and the pilots pulled back on the control column to lift the nose of the plane to the sky, the left control column “stick shaker” activated and continued shaking for most of the few remaining minutes of the flight—a warning sign that the plane was at risk of stalling, or losing lift.

Twelve minutes later, the aircraft plummeted into the Java Sea 21 miles offshore, killing all aboard. Five months later (a timespan during which Boeing claimed that it would be able to identify and remedy whatever the problem was), Ethiopian Airlines Flight 302, also a Boeing 737 Max 8, crashed en route from Addis Ababa to Nairobi under suspiciously similar circumstances, with the loss of all 157 passengers and crew.

Despite initial protestations from Boeing’s then-CEO Dennis Muilenburg that the aircraft were “properly designed” and crew error was responsible for the total 346 lives lost, the 387 operational 737 Max 8s across the world remain grounded today, more than two years later, although the FAA recently announced a series of actions that could return the U.S. fleet to service.

Since then, investigations have raised many questions. Chief among them: how the aerodynamic characteristics of the 737 Max 8 differs from its predecessors, the workings of the plane’s Maneuvering Characteristics Augmentation System (MCAS) flight control system, what “airmanship” means, and the Federal Aviation Authority’s (FAA) ability to adequately oversee the certification of new aircraft. Meanwhile, Boeing’s course for readying the plane for its return to service has been rocky. Along the way, revisions of the MCAS have elicited a state deemed “catastrophic” for a passenger airliner, there have been errors with indicator lights, wiring bundles have been noted to be too close together, and a software glitch that “prevents the flight control computers from powering up and verifying they are ready for flight” appeared in testing....Read more


Forging Tools to Fit the Hand

January 29, 2020

Erich S. Huang, MD, PhD
Director, Duke Forge

A blacksmith (mostly out of frame) hammer a piece of hot, glowing iron on an anvil. Image credit: Jeff Kubina via Wikimedia Commons (CC BY-SA 2.0)

In 1949, J. Deryl Hart, MD, then Chair of Surgery and future President of Duke University, opened Duke’s Surgical Instrument Shop, which was charged with building “anything to aid [healthcare practitioners] in their work and studies.” Alongside myriad surgical instruments, its products included a well-known “Duke University Inhaler” for administering trichloroethylene anesthesia and a folding operating table. As a medical student, I walked by the original site of the Instrument Shop every day, and as a trained surgeon I’ve appreciated how surgical instruments extend my ability to reach into difficult places, dissect tissues at odd angles, and rejoin tissues after injury.

But while the Instrument Shop shaped (and still shapes) metals that extend our capabilities in the operating theater, we now live in a world where data—all those bits of information about our physiologic state, our demographics, our social environments, our genetics and epigenetics—increasingly comprise the raw materials for making new tools for helping our patients. In fact, one could argue that data scientists are the metallurgists and instrument makers of our time.

A little more than two years ago, when we founded Duke’s center for actionable health data science, we named it “Duke Forge” with this metaphor in mind. We are, in fact, instrument makers—not so very different than the craftsmen of the Surgical Instrument Shop. This may sound quaintly artisanal and not remotely “tech” or “data science-y”, but it is very deliberate. “Artificial intelligence,” “machine learning,” “statistics,” “data science”—whatever you want to call it—is an instrument. It’s an extension to a fundamentally moral activity: healing our fellow humans...Read more


Approaching Health Data Science with Humility

October 3, 2019

Erich S. Huang, MD, PhD
Co-Director, Duke Forge

“One, two, three, four, five, six, seven, eight…”

Surgical team caring for a patient in a brightly-lit operating room

Elbows locked, scrubs damp with sweat, glasses sliding down your nose, you keep counting the chest compressions. Amid the vortex of urgency and adrenaline, klaxon blaring in the background, a team has materialized: respiratory therapist at the head of the bed, a nurse by the code cart, sticky pads to the chest and side, wires, tubing, junior residents on either side placing chest tubes…

A “code” — the medical response to a person going into cardiopulmonary arrest — at once represents both the triumph and failure of modern medicine. Viewed from one angle, we see an impressively universal, uniform approach that rigorously applies life-saving techniques, allowing us to yank a patient back from the precipice thanks to teamwork and an extraordinary suite of technology. But there’s another side: the sheer, violent unexpectedness of a code. Pounding on someone’s chest and shocking them is not an expected outcome for just about any intervention – by definition, something has gone awry.

Even when everyone has done everything right, the unexpected still happens. You may have applied the most up-to-date knowledge to select a treatment, done a thorough preoperative assessment, consulted the right specialists, and executed the procedure perfectly, but the complexity of human pathophysiology and our all-too-fallible efforts to alter its course come with no guarantees of success.

Practicing medicine is an exercise in humility. And in the context of healthcare, practicing data science should be no different. As a data scientist, you might employ your craft to the best of your knowledge and abilities, but as the volume of data swells and data products are bound up with the ways we deliver healthcare, there will inevitably be unexpected effects—including unintended harm....Read more


En Garde!

May 20, 2019

Cluttersnap_Saber.png

Erich S. Huang, MD, PhD
Co-Director, Duke Forge

Every Thursday night, I find myself facing an opponent brandishing a fencing sabre on a narrow strip, 14 meters long by 2 meters wide, called a piste. Hooked up to an electrical circuit, wearing high-tensile fencing whites and metallic jackets, we face each other. My adversary’s intent is oblique, hidden in shadows behind the mesh of a fencing mask—does he expect to make a “first intention” attack? Or is he inviting me to attack, allowing him to score a touch with a “second intention” parry, then riposte? Once the judge commands “fence!” initiative and intention may change many times in the course of a single touch between us.

Fencing is a fluid, “open-skilled” sport in which one must constantly modulate between offense and defense in response to an opponent who is also constantly sliding between these modes. On top of that, good fencers are excellent at waiting for a split second (“preparing”) and deciding whether to attack or defend based on cues from their opponent. The South Korean “K-Sabre” style emphasizes speed and deep lunging, while the Eastern European “old school” emphasizes small footwork and precise bladework. However, you can’t get away with pursuing just one approach. Competitive fencers must be multidimensional, capable of changing styles and able to go on offense or defense instantly. This dynamism is the challenge and joy of being “on piste”—arriving at the correct combination of offensive and defensive actions over a 5-, 10-, or 15-touch bout to emerge with a win.

Also? Getting to (legally) chase someone with a sword is fun, too...Read more


What Health Data Science and Raising Chickens Have in Common

April 29, 2019

Chicken.jpg

Erich S. Huang, MD, PhD
Co-Director, Duke Forge

In my most recent blog post, I wrote about Susan Jones*, a 65 year-old patient and Medicare recipient, and described how a multidisciplinary team at Duke constructed an AI-powered workflow to help patients like her.

Building that workflow was a laborious, 18-month-long process. We assembled a band of explorers comprising clinicians, nurses, pharmacists, machine learning experts, statisticians, and informatics experts, and blazed a path into an only partially-understood data wilderness. Clearing this path and developing the machine learning logic to navigate it allows us to predict unplanned hospitalization 1.6 million times a month.

This is the state of data science in health: our current electronic health record systems were never designed for data science, which requires us to “hack” a system that is optimized for billing and documentation to present data in a manner appropriate for statistical analysis or machine learning. This means that each new data science use case is a new expedition into the unknown, with new trails to be blazed. This is a far different state of affairs than we see when Google Assistant tells you about the traffic on the way to work, or Amazon suggests a purchase—where we’re still clearing undergrowth and breaking dirt paths, the tech giants have completely mapped and built modern data highways because they’ve truly become “data science first” companies.

If we are to make Duke a place that will build robust, rapidly testable, and iterative data science and artificial intelligence for the benefit of our patients and clinicians, we must quickly acquire the culture and technical underpinnings to make this feasible...Read more


Front entrance to Davison Building at Duke University

Staying Focused on What Matters

March 11, 2019

Erich S. Huang, MD, PhD
Co-Director, Duke Forge

If you ever feel the urge to play amateur archeologist and prowl Duke University’s Davison Building, built in the 1930s, you will find the remnants of the original Duke Hospital.

Peer down an anonymous corridor of offices, and the stretcher-height stainless steel lining the walls and stretcher-width doors bear witness to their past lives as operating theaters. Peek through the square window in a door, and it's quite possible that you’ll be looking into the operating room where in 1968 William Sealy was the first person to surgically treat an aberrant cardiac electrical pathway called “Wolff-Parkinson-White syndrome.” Elsewhere, between the former Osler and Long wards on the second floor, was the medical intensive care unit, and on the third floor across from the former Cushing ward was the surgical ICU.

It was in these units during the late sixties and early seventies that Frank Starmer, an electrical engineer and a founding faculty member of Duke’s Computer Science Department, worked with Department of Medicine Chair Eugene A. Stead, Jr. and other faculty to build machines, each the size of a small refrigerator, that captured data from critically ill heart attack patients.

These forgotten appliances, these corridors of history, represent Duke’s long and continuing tradition of using data to solve clinical problems...Read more


APIs and Health Data: Chipping Away at the Tower of Babel

The Tower of Babel, c 1563.

September 20, 2018

Erich S. Huang, MD, PhD
Co-Director, Duke Forge

In the kickoff symposium for Duke Forge in 2017, I noted that that a 2012 issue of the Harvard Business Review heralded “data scientist” as the “sexiest job of the 21st century,” “sexy” in this case to be construed as “having rare qualities that are much in demand.” Indeed, considering this summer’s oversubscription of Duke’s Summer Course in Machine Learning, many see working familiarity with TensorFlow, convolutional neural networks (CNNs), long short-term memory (LSTM), and generative adversarial networks (GANs) as the attributes that will prompt prospective employers to “swipe right” on their resumés.

In contrast to the Harvard Business Review, however, the New York Times reported that good data science requires a lot of “janitor work”. In the article, a computer scientist stated, “it’s an absolute myth that you can send an algorithm over raw data and have insights pop up.”

I heartily agree.

More often, it requires a laborious slog into the raw muck and vagaries of such a dataset before you can fire up and tune an algorithm to wring meaning from it. Many cite a rule-of-thumb of 80% time spent cleaning data and 20% applying machine learning to it. But without discounting the real labor that, say, an AirBnB or Stitch Fix data scientist puts into pointing personalized recommendations at customers, our colleagues in the tech industry have made it easier to store data of diverse types at scale, distribute it to computing resources, and push results to individual consumers. 

Another thing tech folk do well is create “ecosystems” for their data science to flourish. AirBnB and Lyft would not work nearly as well without preexisting services like Google Maps. Braintree Payments provides technology for clearing credit card transactions for Stitch Fix; Warby Parker uses Stripe for the same. In turn, these businesses make their services available to other businesses—note how in Apple Maps you can request a Lyft, or how your local restaurants use Uber Eats to deliver, and that you can watch HBO NOW on an AppleTV.

Such ecosystems depend on application programming interfaces, or APIs...Read more