Related Experiment Video
Updated: Jun 27, 2026
07:12
A Gaze-Contingent Display Framework for Perceptual Learning Research with Simulated Central Vision Loss
Published on: April 11, 2025
Reinforcement learning-driven adaptive game therapy for cognitive impairment patients with improved vision
Youseef Alotaibi1, Surendran Rajendran2
1Department of Software Engineering, College of Computing, Umm Al-Qura University, Makkah, 21955, Saudi Arabia.
BMC Psychology
|June 26, 2026
Summary
This study introduces a novel deep learning framework for early Alzheimer's disease diagnosis using Magnetic Resonance Imaging (MRI). The model achieves high accuracy by integrating advanced neural networks and adaptive game therapy, improving patient engagement and diagnostic precision.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Medical Imaging
Background:
- Cognitive impairment, particularly Alzheimer's disease, necessitates early diagnosis and personalized management.
- Neuroimaging, such as MRI, offers insights into brain changes related to cognitive decline.
- Current deep learning models struggle with long-range spatial dependencies in neuroimaging data, limiting diagnostic accuracy.
Purpose of the Study:
- To develop an advanced deep learning framework for improved diagnosis of neurodegenerative conditions.
- To enhance the capture of spatial and temporal dependencies in neuroimaging data.
- To integrate adaptive game therapy for cognitive rehabilitation and patient engagement.
Main Methods:
- A novel framework combining Improved Vision Transformer (Im-ViT) and Residual Simple Recurrent Unit (ResNet-SRU) based Multilayer Perceptron (MLP).
- Preprocessing neuroimaging data using Multiscale Gaussian Filter (MGF) for enhanced feature clarity.
- Integration of Visual Working Memory (VWM)-based game therapy with Iterative Hiking-based Reinforcement Learning (ItHRL) for adaptive difficulty adjustment.
Main Results:
- The proposed model achieved high diagnostic performance with an accuracy of 99.62%.
- Key performance metrics include Recall (99.33%), Precision (98.97%), F-Score (99.56%), and Specificity (99.62%).
- A low Mean Squared Error (MSE) of 0.018 indicates precise predictions.
Conclusions:
- The combined approach of advanced neuroimaging analysis and adaptive game therapy significantly improves classification accuracy.
- The model demonstrates faster convergence and enhanced patient engagement.
- This framework offers a promising direction for early diagnosis and management of cognitive impairment.
