AUTHOR=Formica Caterina , Giambò Fabio Mauro , Latella Desiree , Bonanno Lilla , Lombardo Marco , Tomarchio Orazio , Marra Angela , Alagna Antonella , Bonanno Carmen , Quartarone Angelo , Marino Silvia 

TITLE=Humanoid robots for psychological assessment in mild cognitive impairment: from evaluation to the future of AI-driven data prediction systems

JOURNAL=Frontiers in Psychology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2025.1579626

DOI=10.3389/fpsyg.2025.1579626

ISSN=1664-1078

ABSTRACT=IntroductionNeurodegenerative diseases, such as Alzheimer’s Disease (AD), are increasingly prevalent, emphasizing the need for early diagnosis and effective intervention. This study explores the feasibility of using the humanoid robot Pepper to administer cognitive assessments for Mild Cognitive Impairment (MCI). Specifically, it evaluates the usability, accuracy, and patient experience of robot-administered cognitive testing compared to traditional assessments conducted by neuropsychologists.MethodsA total of 100 MCI patients were randomly assigned to two groups: one undergoing the Mini-Mental State Examination (MMSE) with Pepper and the other receiving the same test administered by a neuropsychologist. After that participants were submitted to a Satisfaction Questionnaire (SQ) designed to assess their emotional and experiential response to the testing procedure, whether administered by a human or a robot.ResultsThe intergroup analysis (EG vs. CG) reveals significant differences in age (p = 0.003) and Total SQ (p = 0.01), and in SQ2 (χ2 = 9.76; df = 1; p = 0.002), SQ4 (χ2 = 5.02; df = 1; p = 0.02), SQ5 (χ2 = 25.35; df = 1; p < 0.001), SQ6 (χ2 = 7.68; df = 1; p = 0.006) and SQ7 (χ2 = 7.56; df = 1; p = 0.006). Results indicate no significant differences in MMSE scores between the two groups, suggesting comparable cognitive evaluation accuracy. However, participants assessed by Pepper reported lower frustration levels and higher satisfaction (90% vs. 40%) compared to those tested by a neuropsychologist. Additionally, 92% of patients in the robot-assisted group expressed willingness to retake the test in the same manner, indicating high acceptability and engagement.DiscussionThese findings suggest that robot-assisted cognitive assessments may enhance patient comfort and accessibility to neuropsychological testing. The integration of Artificial Intelligence (AI) further supports diagnostic accuracy and predictive potential, offering promising avenues for early intervention in neurodegenerative diseases.