Assessment and treatment of visuospatial neglect using active learning with Gaussian processes regression

• DOI: 10.48550/arxiv.2310.13701
• Publication Date: 2023-01-01
• Authors: Ivan De Boi; Elissa Embrechts; Quirine Schatteman; Rudi Penne; Steven Truijen; Wim Saeys
• Keywords: Computer. Automation; FOS: Computer and information sciences; Computer Science - Machine Learning; Computer Science - Artificial Intelligence; 05 social sciences; Computer Science - Human-Computer Interaction; Reproducibility of Results; Human-Computer Interaction (cs.HC); Machine Learning (cs.LG); Perceptual Disorders; Stroke; Treatment Outcome; Artificial Intelligence (cs.AI); Artificial Intelligence; Humans; 0501 psychology and cognitive sciences; Human medicine; Engineering sciences. Technology

Visuospatial neglect is a disorder characterised by impaired awareness for visual stimuli located in regions of space and frames reference. It often associated with stroke. Patients can struggle all aspects daily living community participation. Assessment methods are limited show several shortcomings, considering they mainly performed on paper do not implement the complexity life. Similarly, treatment options sparse only small improvements. We present an artificial intelligence solution designed to accurately assess patient's visuospatial three-dimensional setting. active learning method based Gaussian process regression reduce effort it takes patient undergo assessment. Furthermore, we describe how this model be utilised oriented opens way gamification, tele-rehabilitation personalised healthcare, providing promising avenue improving engagement rehabilitation outcomes. To validate our assessment module, conducted clinical trials involving patients real-world compared results obtained using AI-based widely used conventional tests currently employed practice. The validation serves establish accuracy reliability model, confirming its potential as valuable tool diagnosing monitoring neglect. Our VR application proves more sensitive, while intra-rater remains high.