Abstract

Background and Motivation. The global rise in Mild Cognitive Impairment (MCI) and dementia among aging populations presents an urgent healthcare challenge. With over 55 million people worldwide living with dementia and effective pharmacological treatments remaining elusive, non-pharmacological interventions have become a critical research priority. Human-Computer Interaction (HCI) and Human-Robot Interaction (HRI) offer innovative avenues for scalable, engaging, and evidence-based cognitive care. This presentation synthesizes findings from two complementary empirical studies that examine the efficacy and usability of interactive technologies—multisensory virtual reality (VR) serious gaming and socially assistive robots (SARs)—in supporting cognitive health among older adults.

Study 1: SENSO™ – A Multisensory Motion-Captured Serious Game. The first study presents the development and usability evaluation of SENSO™, a novel HCI system that integrates markerless motion capture, non-immersive virtual reality, and synchronized olfactory stimulation into a culturally grounded serious game. The system features three teahouse-themed tasks—Dim Sum (object selection), Steamer (timing and sequencing), and Cashier (counting and transactions)—designed to train instrumental activities of daily living (IADLs) while simultaneously engaging cognitive and motor functions (dual-task training). A usability study was conducted with 41 healthy older adults (aged 60+), stratified into three age groups: 60–69 (n=22), 70–79 (n=17), and ≥80 (n=2). System Usability Scale (SUS) scores averaged 82/100, indicating high acceptability and intuitive design. Performance analysis revealed age-neutral results in the Dim Sum task, statistically significant age-related declines in the Steamer task, and notable trends in the Cashier task, providing normative baselines for future MCI risk prediction. The integration of olfactory cues—leveraging the olfactory system’s direct connection to the hippocampus and amygdala—is proposed as a key enhancement for memory encoding and reminiscence in therapeutic contexts.

Study 2: Socially Assistive Robots in Cognitive and Reminiscence Therapy. The second study, published in IEEE Transactions on Affective Computing, investigates the neurophysiological effects of robot-led versus human-led cognitive interventions on elderly individuals with MCI. Using a randomized controlled trial (RCT) design, eight MCI participants (mean age: 70.1 years) were assigned to either a NAO robot-led or human-led group, receiving both Cognitive Training (CT) and Reminiscence Therapy (RT). Functional Near-Infrared Spectroscopy (fNIRS) was employed to measure oxyhemoglobin (HbO₂) concentration changes in the dorsolateral prefrontal cortex (DLPFC)—a region critical for working memory and executive function. Key findings indicate no significant difference in DLPFC activation between robot-led and human-led interventions, suggesting that SARs can serve as viable alternatives to human therapists. Critically, RT elicited distinct brain activation patterns compared to CT: during the retrieval phase, RT produced positive DLPFC activation, while CT showed deactivation, particularly in the human-led group. This suggests that SARs may reduce performance anxiety in elders with MCI during memory recall tasks, a notable clinical advantage.

Synthesis and Implications. Collectively, these studies advance the evidence base for HCI and HRI in elderly cognitive care. SENSO™ demonstrates that multisensory, motion-captured VR gaming can achieve high usability and provide ecologically valid dual-task training, while the SAR study establishes that robot-led therapy produces neurophysiologically comparable outcomes to human-led sessions. Both studies highlight the importance of user-centered, culturally sensitive design and objective neurophysiological measurement in evaluating therapeutic efficacy. Future work will integrate biomarkers, expand sample sizes, and explore longitudinal outcomes to validate clinical transfer and establish these technologies as standard tools in non-pharmacological MCI management.

Keywords

Human-Computer Interaction (HCI); Human-Robot Interaction (HRI); Mild Cognitive Impairment (MCI); Virtual Reality; Serious Games; Olfactory Stimulation; Socially Assistive Robots; Reminiscence Therapy; Cognitive Training; fNIRS; Elderly Care; Non-pharmacological Intervention

鳴謝

The authors gratefully acknowledge the participants, occupational therapy consultants, and institutional partners at Pathfinder Technology Limited, UOW College Hong Kong, THEi, The Hong Kong Polytechnic University, and South China University of Technology. Special thanks to Professor Sam Chan, Professor Will Chien, Ms. Stella Cheng, Ms. Vera Lam, and Ms. Anita Ngan for their guidance and support.