Background: Maintaining cognitive efficiency and independence is a central goal of healthy aging. Socially assistive robots (SARs) are increasingly proposed as scalable digital health solutions to support daily activities in older adults and to facilitate aging-in-place. However, concerns remain regarding whether robot-mediated assistance reduces or inadvertently increases cognitive load, potentially undermining usability, user acceptance, and long-term real-world adoption, particularly in aging populations. Objective: This study aimed to examine how robot-assisted (human-robot interaction [HRI]) and human-assisted (human-human interaction [HHI]) support influences cognitive load during task performance in younger and older adults. A multimodal assessment framework integrating behavioral, subjective, and physiological measures was used to identify age-related differences in cognitive effort and stress associated with different forms of assistance. Methods: A total of 60 healthy adults (30 younger adults: mean age 34.8, SD 10.1 years; and 30 older adults: mean age 72.3, SD 5.5 years) completed a modified Trail Making Test under 7 within-subject conditions: independent performance (baseline), 3 robot-assisted conditions, and 3 human-assisted conditions, each corresponding to low, medium, and high cognitive load levels. Performance accuracy and completion time were recorded as behavioral indicators. Perceived cognitive load was assessed using the National Aeronautics and Space Administration Task Load Index, and physiological stress was evaluated via pre- and postcondition salivary cortisol concentrations. Linear mixed-effects models were applied to examine main effects and interactions of age group, assistance type, cognitive load level, and time. Results: Significant interactions between age group and assistance type were observed for accuracy (=6.50; =.01) and perceived cognitive load (=4.58; =.03). Older adults demonstrated lower accuracy and higher perceived cognitive load during robot-assisted conditions compared with human-assisted conditions, whereas no such differences were observed in younger adults. Across age groups, human assistance improved performance at low and medium cognitive load levels. Physiological analysis revealed a significant age×assistance× time interaction (=5.16; =.02), with older adults showing increased posttask cortisol concentrations during robot-assisted interaction, indicating higher physiological stress. Conclusions: While both human and robotic assistance enhanced task performance relative to independent completion, the type of support critically shaped cognitive load responses in older adults. Robot-assisted interaction was associated with increased behavioral errors, higher perceived workload, and elevated physiological stress, suggesting that current SAR implementations may impose additional extraneous cognitive load in older users. These findings highlight the importance of designing adaptive, age-sensitive digital assistive systems that minimize cognitive burden through simplified interaction, responsive pacing, and multimodal support. Multimodal cognitive load assessment provides a valuable framework for optimizing the usability and effectiveness of assistive digital health technologies for aging populations.
Hospitalists Are Already Using AI—Why Implementation Will Determine Its Impact
The adoption of artificial intelligence (AI) into clinical practice is accelerating, outpacing the development of organizational guidance, training, and governance. A recent study indicated that