The Center for Machine Learning and Health (CMLH) at Carnegie Mellon University recently announced the 2024 Generative AI in Healthcare Fellows. This round, two Robotics Institute Ph.D. students, Angela Chen and Bardienus (Bart) Duisterhof, earned their place among the seven recipients.
Chen and Duisterhof both utilize artificial intelligence (AI) in unique ways to expand their research at the intersection of robotics and healthcare. The fellowship includes one year of tuition and stipend support alongside $3,000 in funding to support conference travel for paper presentations, equipment and human-subject experiments.
Angela Chen
Angela’s research, “Harnessing Generative AI To Enhance Mental Health Provider Training: Developing Adaptive Virtual Patients With Multi-Session Capabilities,” focuses on advancing mental healthcare by creating AI-based virtual patients to help improve the training of mental health providers. Chen aims to create AI patients capable of simulating the complex evolution of a client’s mental state and therapy needs over time.
“Without sufficient professional support, many people seek help through online mental health communities, which can provide peer support. In addition to didactic instruction, peer counselors need a safe environment to practice their counseling skills. Training methods used by professional therapists, such as role-playing with trained actors, are too expensive and hard to apply on a large scale,” said Chen. “I’m trying to address this limitation in my research by using AI to build virtual patients for practice and feedback.”
Chen emphasized that it can be difficult to ensure that virtual patients accurately represent the diverse needs of real-world mental health patients and providers. A large part of her mission relies on creating virtual patients that can present realistic, nuanced behaviors reflective of the complexities of human mental health.
“My goal is to use generative AI in healthcare to complement current practice, not to replace anyone,” said Chen. “We want our training programs to help reduce ethical issues in mental healthcare and make it more accessible.”
As a robotics Ph.D., Chen brings a unique perspective to the table, combining her deep knowledge of robotics systems with a passion for improving mental health outcomes. The funding from the Generative AI in Healthcare fellowship supports Chen’s mission to bridge distinct disciplines and highlight the growing importance of interdisciplinary approaches when solving complex, real-world problems.
Chen is advised by professor Haiyi Zhu and collaborates with professors Sherry Wu and Robert Kraut.
Bardienus (Bart) Duisterhof
Bart, advised by professor Jeffrey Ichnowski, focuses on equipping robots with superhuman perception to improve their ability to perceive objects and process information. His research, “Generative 4D Foundation Models for Robust Robot Assistance in Hospitals,” centers on helping robots better understand flexible or transparent objects that have been traditionally difficult to perceive, by using foundation models to track, render, and predict their movements and positioning.
“Tracking the movement of objects, rendering the objects accurately in 3D, and predicting how these objects will move or change shape based on what the robot does can help ensure that robots operate in critical environments with precision and reliability,” said Duisterhof.
Creating accurate reconstructions that are above human comprehension allows robots to be particularly useful in healthcare. Generative AI can be used to create large-scale, realistic simulations of deformable and transparent objects, such as IV bags or surgical tools. These simulations help recreate diverse scenarios that are difficult or expensive to replicate in the real world.
“Generative AI can improve how robots assist with tasks such as surgical preparation,” said Duisterhof. “My work focuses on building new representations and inputs to enable superhuman sensing for robots in safety-critical settings.”
Duisterhof’s research distinctly aligns with the Generative AI in Healthcare fellowship initiative to improve accessibility and ethical implementation of AI in healthcare.
“Robotic assistance in the medical domain will only improve upon human precision,” said Duisterhof. “There is a lot of unique work happening in robotics to help quell healthcare anxieties, and generative AI is improving upon how we can train and prepare systems for critical tasks.”
For More Information: Aaron Aupperlee | 412-268-9068 | aaupperlee@cmu.edu