An elderly care service demand analysis model based on MLP-Self-Attention

Abstract

Purpose:  With the intensification of the aging population trend, the demand for elderly care services is steadily rising. Therefore, accurately understanding the actual needs of the elderly is a prerequisite for formulating effective elderly care policies and providing high-quality services. Through demand analysis, specific information about the elderly's health status, lifestyle, and mental state can be gathered, providing a basis for personalized services. Additionally, demand analysis can assist elderly care institutions or governments in planning suitable service offerings, enhancing service quality, and meeting the diverse needs of the elderly population. Moreover, timely understanding and addressing the actual needs of the elderly can improve their quality of life and sense of happiness, promoting social harmony and stability.

Theoretical framework: Demand analysis for elderly care services is a crucial component in the construction of the elderly care service system, holding significant importance for ensuring the well-being of the elderly and the sustainable development of society. This study proposes a multi-label model MAC for elderly care service demand analysis, incorporating a Multi-Layer Perceptron (MLP), Self-Attention mechanism, and Cosine distance.

Methods: Using survey results and feedback from individuals aged 65 and above in Jiangsu Province as the dataset, MAC is employed to derive demand analysis results for elderly care services.

Findings: The model utilizes the MLP as the foundation for linear fitting, effectively reducing individual differences between data. It employs the cosine value of the angle between two vectors in vector space as a metric for measuring the magnitude of differences between individuals, thereby enhancing the classification performance of elderly care services.

Practical implications: The model incorporates a Self-Attention mechanism to strengthen the fitting of the importance of all factors in the survey questionnaire to the final result. Experimental results demonstrate that the MAC model can provide more rational planning suggestions for elderly care service providers. As a fusion algorithm utilized in elderly care service analysis, MAC exhibits commendable performance.

Originality/Value The originality lies in its emphasis on the main approaches, analysis methods, and recently used models in the study of aging population trend, the demand for elderly care services is steadily rising.

Keywords: Elderly Care Service Analysis, MLP, Attention, Cosine, Multi-Label.