Year: 2026
Type: article
Abstract: Staffing shortages in nursing homes directly affect the quality of care received by residents, including the development of pressure ulcers, a dangerous and costly condition exacerbated by staffing limitations. Current technological solutions are Alternating Pressure Air Mattresses (APAMs) that change inflation level of air cells within the mattress, operating on a passive cycle with fixed rules that do not account for the diversity of patient needs and body morphologies that can influence pressure injury risk. This paper outlines the development of a proof-of-concept using reinforcement learning to create an autonomous and adaptive APAM, where a Soft Actor-Critic agent trained in a MuJoCo physics simulation to optimize air cell inflation levels across varying patient positions and orientations to determine the effectiveness of reinforcement learning in pressure ulcer prevention. The initial model did not outperform static or randomized inflation level changes. Future improvements include more training episodes, an enhanced reward function, and a simpler initial demonstration with fewer air cells. Following a successful demonstration of this proof of concept, additional clinical and biometric data will be incorporated to train a more complex, adaptive, and autonomous policy for a more comprehensive pressure injury care system.
Authors: Trey Hansen, Stephen McGill
Language: English
Subfield: Dermatology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W7234336804
