“People have tried to classify personality types since Hippocrates’ time, but previous scientific literature has found that to be nonsense,” says William Revelle, professor of psychology and a self-proclaimed skeptic when it comes to personality types. So when his Northwestern colleagues Luís Amaral of the McCormick School of Engineering and Martin Gerlach, a postdoctoral fellow in Amaral’s lab, proposed a study to outline new personality types, Revelle balked.
When hospital leaders across Chicago wanted to know if providing housing to the city’s homeless individuals reduced their use of the emergency room and increased their use of primary care services, they turned to a collaboration brought together by the Feinberg School of Medicine’s Center for Health Information Partnerships (CHIP).
“We knew from prior work that homeless patients are the most likely to seek care across multiple institutions,” says CHIP director Abel Kho, “and this fragmented care and lack of social support leads to poor health outcomes.”
By linking data on Chicago’s homeless individuals with clinical data across multiple hospital and health care centers citywide, the collaborative team was able to determine the immense impact of housing on health.
“We learned that we need to expand our focus,” Kho says. “Rather than giving homeless individuals more health care, we should give them supportive housing, and they do much better and make better use of our health care system.”
Data with Purpose
Kho believes big data can improve lives — and lots of them.
Bringing together expertise across medicine, artificial intelligence, computer science, machine learning and more, Kho’s CHIP works to increase use of electronic health records and improve both health care and health outcomes nationwide.
Kho, who earned his undergraduate degree in engineering, took a year off and then went to medical school. His training in informatics — using massive amounts of information to improve health and health care — made clear for him what’s possible when medicine embraces data science.
“Informatics is the perfect blend of problem-solving and population-level impact,” Kho says. “And that’s really exciting, because it gives us a chance to get novel insights about a population or condition.”
Ten years ago, Kho says, there was no way to uniquely and anonymously identify a patient in Chicago, meaning simple questions — like how many diabetics there were in the city — went unanswered.
“So, we started working with stakeholders, figured out a way to identify patients and then built the infrastructure to do that on a broad scale,” Kho says.
A Central Hub
One central component of CHIP’s work is the Chicago Area Patient-Centered Outcome Research Network, or CAPriCORN. Bringing together the health data records of 11 institutions in Chicago, from Veteran Affairs offices to academic medical centers, CAPriCORN acts as a data warehouse that allows the institutions to share data while maintaining patient privacy.
“We developed software that allows us to link people across institutions, and our method is now being used by a national network linking the private identities of 60 to 80 million Americans,” Kho says. “That creates a really powerful tool to bring data together to do all sorts of interesting analysis.”
This data will allow many different doctors and organizations to gain insights into, and develop solutions for, pressing public health and health care issues.
For example, CAPriCORN is being used to predict whether cancer patients given a particular type of treatment, known as checkpoint-inhibitor therapy, will go on to experience autoimmune disease. Led by immunology expert and CHIP associate director Theresa Walunas, researchers are using machine learning to identify cancer patients who received this therapy and were subsequently diagnosed with rheumatoid arthritis, diabetes and other autoimmune diseases.
By comparing these patients with autoimmune disease patients who have not had cancer, the researchers think they will be able to shed new light on how autoimmune diseases develop and who is at increased risk. Importantly, this work could also lead to improved treatment regimens — and outcomes — for cancer patients.
“We’re in a place now where we can explore autoimmune and other rare diseases,” Walunas says. “We couldn’t do that before, but electronic health data has changed the game.”
Eager for Expansion
In addition to ongoing projects, CHIP — housed within Feinberg’s Institute for Public Health and Medicine — also welcomes research proposals submitted by other institutions, foundations and even pharmaceutical firms.
“We’ve always believed that insights are stronger when you have multiple people involved in discovering them,” Walunas says. “It’s rare that one source can give you everything you need.”
Opening up access in this way, while maintaining a database that’s expanding by the day, might seem like a daunting task. Kho says CHIP is certainly up to the challenge.
“We’re happy to be the ones to roll up our sleeves because we believe this work not only enables our own research but lots of other research too.”