Skip to content
MIT Better World

By Kara Baskin

Most were opioid overdoses, according to the Centers for Disease Control and Prevention. On Staten Island, the statistics are especially grim: historically, the area has suffered the highest rate of unintentional opioid deaths out of New York City’s five boroughs and double the rate of the United States overall, with 28.7 unintentional overdoses per 100,000 people each year.

Jónas Oddur Jónasson, an assistant professor of operations management at the MIT Sloan School of Management, and Nikolaos Trichakis PhD ’11, the Zenon Zannetos Career Development Professor of Operations Management, are working to address this crisis. They have partnered with the Staten Island Performing Provider System on a proactive program designed to allocate resources to high-need patients based on data. The program is called Hotspotting the Opioid Crisis.

Through this partnership, the MIT team was able to develop a computer model that helps providers predict a range of adverse opioid-related events. Researchers began by accessing data from more than 70 Staten Island care providers, gathering electronic health records and prescription data on 251,781 patients who were either on Medicaid or uninsured. They then applied their artificial intelligence-based analytics system to predict which patients were most at risk of overdosing.

For example, by considering 107 behavioral variables, the system can stratify patients by their risk of opioid overdoses through examining their history of prior prescriptions and interactions with the Staten Island health care system. It can also capture the number of short-acting hydrocodone prescriptions filled in the past 90 days or the number of benzodiazepine refills. In this way, the team’s algorithm can identify the top 1% of the highest-risk patients, who in turn account for 69% of adverse opioid events, Jónasson says.

Their model can help health care teams conduct targeted interventions, steering limited resources to the most vulnerable. It’s a potent example of the type of impactful work funded by the MIT Sloan Health Systems Initiative (HSI), according to HSI Director Anne Quaadgras ’85, SM ’86.

A program within the MIT Sloan School of Management, HSI funds and amplifies research, convenes experts, and advances networking opportunities to tackle urgent issues in health care. Research centers on analytics, operations, and incentives that promote healthier behaviors to reduce costs.

“Our goal is to bring researchers and practitioners together to innovate and implement systemic health care solutions,” Quaadgras says. “There are lots of places that do health research, but they’re usually in medical schools and public health schools. There are far fewer that are really focused on health systems and delivery systems.”

HSI has invested in projects ranging from developing analytics to support liquid biopsy for cancer detection to analyzing the health care costs of postmortem genetic testing. Approximately 30 Sloan researchers with roughly 80 working papers are affiliated with HSI.

In another HSI-supported effort, Jónasson has been conducting behavioral analytics research on tuberculosis (TB) patients in Kenya. He and his team are assessing the benefits of a treatment-adherence support platform called Keheala, which offers automated medication reminders, motivational messages, and personal outreach from peer sponsors (people who have overcome TB themselves).

The lack of adherence to treatment protocols is a major barrier to global efforts to eradicate TB, Jónasson says, so it’s important to find interventions that work. In the Keheala study, researchers found that outreach to patients from peer sponsors increased the odds of verified next-day treatment adherence by 35%. This work helps to validate the costly but impactful program, Jónasson says.

“This type of work illustrates how data science and analytics can have a material positive impact to societal welfare and can help improve the world we live in,” Trichakis adds.

Jónasson says that such work on behavioral issues is the next step for the burgeoning field of precision medicine, which uses artificial intelligence and machine learning to improve medical decision making and to personalize medicines. “It’s interesting to try to increase precision—doing the right thing for the right person at the right time,” he says. “The idea is to bring precision to behavioral health and social services, which contribute to health.”