It doesn’t take a data scientist to calculate the impact the SDSU Data Science Symposium is having on the region, , according to the chair of the organizing committee Semhar Michael.
The fifth version of the annual conference was held on the SDSU campus Feb. 6-7 with nearly 200 participants from academia and industry. Kurt Cogswell, head of the SDSU mathematics and statistics department, said, “Having industry is a key and unique aspect of this conference. It serves as a motivator for students, and it provides an opportunity for industry to meet and recruit students.”
Michael, associate professor of statistics at SDSU, said 97 students registered, the most ever, which she attributed to industry sponsorships that provided them free admission. In addition, travel reimbursement was offered to students who presented papers. Six of the 14 undergraduate presenters came from other schools, including Iowa State and University of Minnesota.
Their topics ranged from weather (tropical cyclone prediction) to health (detecting lung abnormalities) to social issues (comparing crime rates before and after COVID).
The main focus of the conference is data science in the realms of finance, health care and agriculture. In addition to breakout sessions led by faculty and industry, there were two keynote presentations.
Ryan Nichols, data science and analytics adviser with TransUnion, spoke on “Unraveling Complex Problems: Applying Systems Thinking in Data Science” at the Monday night banquet.
From health data to health outcomes
The Tuesday morning keynote was from Emily Griese, chief operating officer with Sanford Health Plan and an assistant professor in pediatrics at the University of South Dakota Sanford School of Medicine.
Her message was “How does data science help us achieve better health outcomes for our communities?”
Griese, a previous symposium speaker and an annual classroom guest for Cogswell, said the health care industry creates massive amounts of patient data, but the challenge for analysts is to make that “super messy” information meaningful. “Health care data isn’t lab research. Nothing about health care data science is controlled.
“It takes science to make it meaningful and actionable. Ninety percent of your time will be spent cleaning up the data, but the other 10% can have an impact at the population level,” Griese said.
Change of address key indicator
She explained that the Sanford Health system sees more than a million patients annually. Each one is asked for date of birth, address, proof of insurance and other chart-tracking questions in addition to the question the patient wants to answer: Why are you here? Oftentimes the information is asked more than once during a visit and in varying ways. The answers may be recorded slightly different.
Hence, some super messy data, but it can yield a treasure trove of data for the insurance executive trying to control costs and ultimately, better the health outcomes of the patients and members.
Griese said Sanford looked at its stacks of patient information to find what would be the best predictor of unnecessary use of emergency rooms. Lack of insurance and a patient’s work schedule certainly were factors, but “we have found that frequent change of address is the top predictor of inappropriate use of ER,” she said.
Patients who have a new address every visit not only often lack insurance or an established relationship with a provider, they probably don’t know where the neighborhood clinic is, Griese said.
So, do patient data profiles change the look of a patient visit? Drivers such as appointment no-shows and prescription refill rates can create a risk factor indicating these patients need additional support to avoid inappropriate cycling back through the emergency room. This is important data informing healthcare practices, Griese said.
Staying relevant to participants
In post-keynote comments, she heralded conference organizers for growing the event and bringing in speakers who take deeper dive at data application.
Michael said the symposium scientific committee has diversified the content of the event and tried to have more recent topics as data science is a fast-growing field. Artificial intelligence was addressed by several session speakers. In addition to the 97 students, there were 29 faculty members and 72 from industry in attendance.
That threefold perspective fosters collaboration and opens the door for student internships, Michael said. Between Tuesday’s lunch and afternoon session, there is a recruiting fair. Also, “industry shares with faculty and students what they look for in new graduates and what courses and training academia should focus on,” Michael said.
The symposium had 11 sponsors, led by longtime platinum sponsors Capital Services and First Bank & Trust.
- Contact:
- Telephone number: 605-688-4538