Early Detection of Alzheimer's Disease Assisted by AI-Powered Human-Robot Communication

Abstract

Dementia, more recently called a major Neuro-cognitive disorder, affects memory, behavior, thinking, social autonomy, and social abilities severely enough to interfere with the activity of daily life. With an aging population, the number of people suffering from it, especially in its most common form, Alzheimer's disease (AD), will drastically increase. With an associated increase in financial costs and negative social impacts, AD presents dramatic challenges for health services as well as affected individuals and their family caregivers.Various cognitive screening tests and questionnaires have been developed for the diagnosis of AD, but they are rarely accurate or reliable enough for all early-stage diagnoses and may be associated with false-positives. They are often used to monitor the severity of functional loss if Alzheimer's disease is detected. With such limitations, artificial intelligence (AI) has a potential to help in detection of AD. Indeed, AI is currently playing a key role in facilitating the diagnosis of some diseases using pharmacological and non-pharmacological markers. AI models are capable of handling enormous amounts of clinical/patient data, learning robustly, and acting with high performance for identification of novel markers of AD risk and early detection. To date, early detection of AD has mostly employed markers coming from image processing of magnetic resonance imaging (MRI).There is a need to expand markers to provide opportunities for early detection of AD by measuring symptoms or behaviors that are associated with AD. Examples of such non-pharmacological markers include certain specific forms of oral communication or behaviors. Interaction between human and robot is one approach to observe these indicators. This chapter discusses the potential for early detection of AD based on human-robot communication, which uses a supervisor robot conducting direct interactions with people during a cognitive assessment. A scoping review of that subject was conducted. The results obtained will be summarized according to existing proposed methods of AI-powered robots using PRISMA-Scr (Preferred Reporting Items for Systematic Reviews and Meta-Analysis-Scoping Reviews) reporting guidelines. An information expert did a detailed search in the Scopus database from the start of the project until January 2022. This review selected the qualitative or quantitative studies that used AI-powered robots for early detection of AD in a process where human and robot communicated. All studies that proposed, tested, or both were included. In the study selection stage, title and abstract of the extracted records were screened by a reviewer according to the eligibility criteria. Titles and abstracts of all bibliographies of the studies were also examined for our inclusion criteria. Bibliographies that met the inclusion criteria were considered for full text review, full texts of the included articles from the previous two steps were reviewed and categorized as inclusion or exclusion. One reviewer performed the three steps of the study selection, and a second reviewer confirmed the results of the screening step done by the first reviewer.We retrieved 416 documents. After three levels of screening, five studies were included in the scoping review. Data including population characteristics, type of intervention, and length of the intervention. Outcomes were collected. The details are reported in the results section of this chapter. The findings are discussed, along with recommendations for further research in this field to improve the early detection of AD through human-robot communication.