Deep learning-driven multidimensional feature extraction and early warning method for digital addiction behavior in the elderly

The influence that addiction to technology has on the mental and behavioral health of older adults is a matter of grave concern that warrants considerable concern. Therefore, individuals have a more difficult time communicating and thinking critically. Conventional diagnostic approaches are, in most cases, unsuccessful due to the intricate and diverse patterns of digital dependence observed among older people. This study proposes the Deep Learning-Driven Multidimensional Feature Extraction and Early Warning (DL-MFE-EW) framework to identify, assess, and anticipate indicators of addiction to digital media. To discover patterns in behavior and time within large datasets, the recommended strategy will use Long Short-Term Memory (LSTMs) and Convolutional Neural Networks (CNNs). On some of these websites, one will get information on the amount of time spent in front of a screen, the total number of interactions, and psychological indicators. Utilize a method of attention that considers mental, emotional, and temporal dimensions simultaneously if interested in gaining an understanding of the connections that exist between elements. DL-MFE-EW achieved 0.947 prediction accuracy and 0.93 early-detection sensitivity on real-world datasets of older adults' digital activities, representing a significant improvement over baseline models. It is possible that multidimensional modeling based on deep learning could be an effective method for protecting the digital health of older people and implementing preventive measures.