Doctoral student’s system eases patient-discharge process

Every day, millions of people are discharged after extended hospital stays, but matching these patients with appropriate care facilities can be arduous, often reliant on months-old, inaccurate data.

Now, a text message-based, hybrid computer-human system that regularly updates both patients' and care facilities' availability statuses, developed by a Cornell doctoral student, is smoothing that time-consuming process. The system was tested at a hospital in Hawaii for 14 months, beginning in early 2022, and helped place nearly 50 patients in care facilities.

In fact, the system worked so well, the hospital is still using it.

"I worked closely with the people who had the problem, one on one, and not just, 'Here's a technology, maybe it'll help you.' It was a more tailored approach, and that helped us get off the ground faster," said Vince Bartle, doctoral student in the field of information science at Cornell Tech and lead author of "Faster Information for Effective Long-Term Discharge: A Field Study in Adult Foster Care," originally published on May 2 in Proceedings of the Association of Computing Machinery on Human-Computer Interaction.

The work won a Best Paper award at the ACM SIGCHI Conference on Computer-Supported Cooperative Work and Social Computing, Oct. 18-22 in Bergen, Norway. Bartle had another paper on his system, regarding preferences and incentives for placing long-term patients, that received Impact recognition at the conference.

Senior authors are Nicola Dell, associate professor of information science at Cornell Tech, the Jacobs Technion-Cornell Institute and the Cornell Ann S. Bowers College of Computing and Information Science; and Nikhil Garg, assistant professor of operations research and information engineering at Cornell Tech, the Jacobs Technion-Cornell Institute and Cornell Engineering.

The seeds of this project were planted in 2020, right as the pandemic hit, when Bartle learned of "a whole host of problems" at Queen's Medical Center in Oahu, including placing long-term-care patients following discharge from the hospital.

Following numerous discussions with Ashley Shearer, director of care coordination, and Alexandra Wroe, chief operating officer, both at the Queen's Health Systems in Honolulu, Bartle determined the important questions that needed to be addressed with his text-based system. The message network was then tailored to meet the needs of all stakeholders.

After receiving an opt-in message, each care facility received regular survey messages, generated every 21 days, asking them to confirm their vacancy status and patient preferences. Hospital staff handled responses from care facilities and coordinated the placement of patients.

Before Bartle's system was deployed, the state of Hawaii would provide updated information on care-home availability every 105 days, meaning the data was often obsolete by the time hospital staff inquired. Shearer said her team used to make "cold call after cold call" to potential caregivers.

"We would often reach disconnected numbers or get no answer," she said. "Even when they connected, the match wasn't always right for the patient. Everything was tracked by hand, which made the whole process slow and frustrating."

After testing his system on a small set of care homes, Bartle deployed it in February 2022 to a total of 1,047 homes that could be reached via text message.

Bartle's system sent out a total of 16 surveys - more than 37,000 individual text messages - during the test period. The hospital received more than 8,000 total responses, achieving a typical response rate of between 35% and 44% for each survey.

Out of the 155 long-term-care patients the hospital received during the survey period, 127 were discharged. Care coordinators confirmed that at least one-third of those were placed in homes that were first contacted via Bartle's system.

Bartle said the response was better than he expected.

"Every step of this has been, to some extent, surprising," he said. "We started by sending out 10 messages, and I thought, 'I hope someone responds' and 'What if they hate me?' But there is a real need, a real problem that's actually happening. And I think that's contributed to how they engage with it."

Garg praised Bartle's years of work developing and fine-tuning the platform.

"That Vince was able to form a relationship with the hospital, and then prototype, build and deploy this platform, is nothing short of incredible," Garg said. "Many academics hope to have this sort of impact once over their career, and Vince was able to do it as a graduate student by himself."

In fact, Bartle said, the hospital informed him that they now rely solely on his platform, and no longer the state, for updated patient and facility information. The hospital is sponsoring the project, Bartle said, which is a major reason for its continued success.

Said Shearer: "Vince worked closely with our team to design a solution that actually fits the way we work. It's a more efficient, more thoughtful approach that helps with patient and caregiver satisfaction, and staff morale."

Shearer and Wroe are co-authors of the paper. In addition to Queen's Medical Center, financial support for this work came from the Gates Millenium Scholar Program, the National Science Foundation and Amazon.