Biothreats are disease-causing pathogens that spread quickly through a population. In many ways, they're far sneakier and difficult to contain than human terrorists.
"Unlike a bomb or natural disaster, biothreats don't usually announce themselves," says Nathan Sutton, a data scientist at Vituity. "An outbreak can spread extensively before our current monitoring systems detect it."
Sutton and his colleagues have an idea that might change that.
Advancing to the Finals
On Feb. 14, Vituity's data team was named one of five finalists in the U.S. Department of Homeland Security's Hidden Signals Challenge.
Hidden Signals challenged data teams with developing early biothreat warning systems using existing free or low-cost data sources. A panel of expert judges evaluated 37 entries on originality, feasibility, impact, sustainability, and scalability.
The Vituity team members split a $20,000 reward and advanced to the final phase of competition.
Vituity's Chief Data and Analytics Officer Dipti Patel-Misra, PhD, MBA, couldn't be more thrilled for her team.
"To create this group, we sought out individuals who could really push the edge of innovation," she says. "They've been together less then two years, and they're already delivering brilliant, first-to-market solutions. So I had no doubt they had it in them."
Keeping a Pulse on the Flu
Interestingly, several 2018 Hidden Signals finalists proposed new ways to track the flu virus.
Flu is a perennial problem that's both potentially deadly and extremely difficulty to detect in its early stages.
From 2002 to 2015, Google tried to "nowcast" outbreaks of both flu and dengue fever by correlating morbidity and mortality data with user's online search behavior.
Unfortunately, the project proved to be a flop. "The data they were using wasn't close enough to the outcome," Sutton says. "A lot of people who search for info on the flu don't actually have symptoms."
On the flip side, Sutton and his team hypothesized that emergency department (ED) wait times might be a sensitive early indicator of a flu outbreak. To explore the idea, they leveraged Vituity's in-house database, which compiles a wealth of clinical data (including ED wait times) from Vituity EDs across the country.
They first created a reference model by layering Vituity's wait-time data over the CDC's historical influenza mortality data. (The Vituity data was abstracted to protect patient privacy.) This provided an excellent picture of how ED wait times change and spike as a flu outbreak spreads through a population.
A Gold Mine of Free Data
To put their model into practice, the team still needed a way to monitor ED wait times across the country in real time. That's when team members David Yue and Vivek Bhansali came up with a brilliant solution.
Many hospitals publish average ED wait times on their websites as a marketing tool. Yue and Bhansali designed an ingestion engine that "listens" continuously to websites at 142 hospitals in 19 states and 16 major metropolitan areas. The system refreshes hourly, providing an excellent map of ED wait times across the country.
By referencing current wait times against the historical model, the team believes it can detect flu outbreaks far before our current monitoring systems can. And the Department of Homeland Security agrees.
"The judges were amazed by the feasibility of the idea," Sutton says. "They were asking each other, 'Why aren't we doing this already?'"
The judges also liked the fact that the model comes with a live feed (the ingestion engine) built in. "There's no additional data collection or extraction required to get it up and running," Sutton says. "They can just switch it on and watch it go."
The Dream Team
Patel-Misra couldn't personally participate in Hidden Signals because of her connection to one of the judges. But when she floated the idea to her team, they jumped at the chance to participate.
All team members worked several hours a week off-the-clock for two months to design their concept.
In a way, the challenge was business as usual. Though the team has been together for less than two years, they've already developed several exciting data products to help Vituity's hospital clients improve care.
In 2016, they launched a two-way practice management solution that displays clinical data in real time. And just last month, they delivered a secure phone app that assists ED providers with patient callbacks.
Both projects were challenging because they were completely novel. "There was nothing out there for us to replicate or build upon," Patel-Misra explains. "We created everything from the design phase up."
To facilitate innovation, Vituity's data executives have worked hard to create a culture that encourages risk-taking. "We embrace failure as part of the process," Patel-Misra says. "We believe that very often, failure translates into opportunity."
In fact, the earlier a team fails, the faster it can often narrow in on the best solution. "When we use this agile approach, our customers see benefits and ROI sooner," Patel-Misra says. "It's a fast win for them."
"Many of the finalists were from small tech companies," she adds. "But I think this win shows that you can absolutely foster creativity and innovation in a large organization like Vituity."
The Next Phase
The five Hidden Signals finalists now move on to phase 2 of the competition, in which they will enter a virtual accelerator to create a detailed system design of their concept.
But at least one team doesn't plan to stop there.
"As engineers and data scientists, writing documentation doesn't fulfill our deepest dreams," Sutton says. "So while it's not required for our submission, we actually plan to build this thing."
The Vituity team hopes that creating a plug-and-play solution the government can use immediately will boost their competitive edge as they vie for the $200,000 grand prize.
The winner will be announced in April. And at least one industry expert is definitely betting on them.
"I was very proud and thrilled when they were announced as finalists," Patel-Misra says. "But I wasn't really surprised. I'm probably a little biased, but I think they're one of the best data teams I've ever seen."
Congratulations to Vituity's Nathan Sutton, Amy Baer, Jeff Lieske, David Yue, Vivek Bhansali, and Prasad Mantri on being named Hidden Signals finalists. We're all rooting for you as you move into phase 2.