Utilizing offline reinforcement learning (RL) with real-world clinical data is getting increasing attention in AI for healthcare. However, implementation poses significant challenges. Defining direct rewards is difficult, and inverse RL (IRL) struggles to infer accurate reward functions from expert behavior in complex environments. Offline RL also encounters challenges in aligning learned policies with observed human behavior in healthcare applications. To address challenges in applying offline RL to physical activity promotion for older adults at high risk of falls, based on wearable sensor activity monitoring, we introduce Kolmogorov-Arnold Networks and Diffusion Policies for Offline Inverse Reinforcement Learning (KANDI). By leveraging the flexible function approximation in Kolmogorov-Arnold Networks, we estimate reward functions by learning free-living environment behavior from low-fall-risk older adults (experts), while diffusion-based policies within an Actor-Critic framework provide a generative approach for action refinement and efficiency in offline RL. We evaluate KANDI using wearable activity monitoring data in a two-arm clinical trial from our Physio-feedback Exercise Program (PEER) study, emphasizing its practical application in a fall-risk intervention program to promote physical activity among older adults. Additionally, KANDI outperforms state-of-the-art methods on the D4RL benchmark. These results underscore KANDI's potential to address key challenges in offline RL for healthcare applications, offering an effective solution for activity promotion intervention strategies in healthcare.

Patient transfer devices allow to move patients passively in hospitals and care centers. Instead of hoisting the patient, it would be beneficial in some cases to assist their movement, enabling them to move by themselves. However, patient assistance requires devices capable of precisely controlling output forces at significantly higher speeds than those used for patient transfers alone, and a single motor solution would be over-sized and show poor efficiency to do both functions. This paper presents a dual-motor actuator and control schemes adapted for a patient mobility equipment that can be used to transfer patients, assist patient in their movement, and help prevent falls. The prototype is shown to be able to lift patients weighing up to 318 kg, to assist a patient with a desired force of up to 100 kg with a precision of 7.8%. Also, a smart control scheme to manage falls is shown to be able to stop a patient who is falling by applying a desired deceleration.

Deepak Gaddipati is founder and chief technology officer at VirtuSense, an artificial intelligence company that aims to transform healthcare from reactive to proactive by alerting care teams to adverse events such as falls, sepsis and heart attacks before they occur. Gaddipati invented the first commercial full-body, automated, AI-powered scanning system, which is widely deployed across most U.S. airports. He is steeped in the power of AI. Healthcare IT News sat down with Gaddipati to discuss some of his work in healthcare with AI and where he sees the technology headed. You invented the full-body scanning system.