In his sophomore year, biomedical engineering major Bikram Bains, Engr ’26, was determined to find a meaningful volunteer opportunity at The Johns Hopkins Hospital.
Enter Chaplain Elizabeth Tracey in the medical intensive care unit (MICU), who created the This is My Story (TIMS) Program during the COVID-19 pandemic. She wanted to use AI to streamline the program’s audio editing process, which involved condensing detailed interviews into brief, soulful windows into a patient’s life.
For the last six years, Tracey and her team have been interviewing patients or their loved ones (if they are unable to verbally communicate) to get to know them on a personal level. These interviews seek to capture the patient’s identity by asking about sources of joy, essential care information for the medical team, and sources of peace. The responses typically generate human interest insights—about cherished pets, beloved grandchildren, interesting hobbies—that can help clinicians connect with the patient. These connections can transform the health care experience, helping the patient to feel seen and understood while encouraging the provider to deliver care with more empathy and intention.
The interviews are edited and integrated into the patient’s medical records so that the entire care team can listen to them at any time.
“The TIMS Program interested me because of its focus on improving human-centered care even in the most intense medical situations,” says Bains. “The unique opportunity to use my technical skills in AI to improve a process while maintaining a personal approach to care was very appealing to me.”
Bains recruited his roommates Sampath Rapuri and Edgar Robitaille—both biomedical engineering and computer science double majors—to join him. The students developed an AI model that can automate the editing process of interviews and recognize subtle details that can be highlighted, such as a love of dogs, to help a clinician and patient connect. They were also challenged with training the model to remove disfluencies such as “um.”
The team’s AI model supports a larger effort to improve TIMS efficiency by using an app.
“After seeing the benefits of TIMS recording both for the medical team and patients and families, it was clear that we needed an app to streamline and standardize the process. Bains, Rapuri, and Robitaille have developed the process to edit TIMS recordings into their final form as part of the app’s functionality,” Tracey says. “Without their contribution, we would not have a functional app.”
“The unique opportunity to use my technical skills in AI to improve a process while maintaining a personal approach to care was very appealing to me.”
— Bikram Bains
By uniting empathy and AI in their work, the students are working toward improved efficiency while keeping the program’s focus on the “personal touch.” In fact, their AI model can edit the interviews to a comparable level to that achieved by a medical student trained for one hour by Tracey and her team.
“It is not hard to get an output from an AI model, but it is hard to get the right output, especially when you have such sensitive information. You must have a lot of thoughtfulness, and we had high standards on what we wanted the output to be, which took a lot of experimenting,” Bains says.
Fresh off a publication in JMIR Medical Informatics, the team is now finalizing a second paper analyzing the application of their model throughout the hospital.
“We have grown as engineers through this opportunity to work on a project that impacts real patients and care providers at the hospital. Having that instant feedback with a direct pipeline from the classroom to the clinic is extremely gratifying,” says Rapuri.
That pipeline was supported by a network of Johns Hopkins resources. Utilizing HopGPT (formerly the Hopkins AI Lab) allowed them to run their sophisticated models, while a Catalyst Award from the Johns Hopkins Office for Undergraduate Research funded the computational resources needed to turn the team’s ideas into working solutions.
With this infrastructure in place, the trio was able to adapt as quickly as the field itself. “As AI evolved over the last two years, we kept our ear to the ground and stayed informed on advances that could improve our project,” says Bains.
This hands-on experience has done more than solve a technical challenge; it has shaped the students’ futures. Now seniors, all three plan to attend medical school after graduation with an interest in specializing in critical care. They hope to stay involved with TIMS even from afar and pass the project to a new cohort of students to continue.
— CATHERINE GRAHAM AND HOLLY PAESCH
Illustration by Helena Pallarés