Scholarlink

<p>As the population ages, there will be a significant increase in the need for intelligent systems that can comprehend individuals’ social interactions and anticipate the services they want. It is impossible for traditional methods of determining what older people need, based on static demographic or clinical data, to reflect the dynamic, ever-changing social networks that significantly impact health and support. The researchers developed a Graph Neural Network–based Elderly Service Prediction model (GNN-ESP) specifically to predict services for older adults and address this need. To demonstrate how the elderly are linked to their loved ones and to society as a whole, the technique presented uses weighted edges. Data on various aspects, including health, engagement frequency, and activity patterns, is compiled using a graph convolutional architecture. The next step is to identify key social connections using an aggregation approach that emphasizes attentiveness. It is necessary to make adjustments over time, as people’s needs for services and social interactions are constantly evolving. Compared with the F1-scores of normal machine learning and non-graph deep learning models, the GNN-ESP model achieved an 8–12% improvement. This model accurately forecasted service demand for senior social care in 94.1% of cases when applied to a real-world dataset. The fact that fraudulent service alerts occurred 17.6% less often is yet another indication that the newly recommended method is more trustworthy. As a result of the GNN-ESP design, intelligent systems can make decisions about expanding social services and elder care. The reason it is effective is that it keeps a record of social contacts and how those interactions develop into logical requests for help. Easy to set up and modify to meet this work changing requirements.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A graph neural network-based model for analyzing social relationships and predicting service needs among the elderly

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月2日

查看全文
  • Yi Wen,
  • Zhen Deng

摘要

As the population ages, there will be a significant increase in the need for intelligent systems that can comprehend individuals’ social interactions and anticipate the services they want. It is impossible for traditional methods of determining what older people need, based on static demographic or clinical data, to reflect the dynamic, ever-changing social networks that significantly impact health and support. The researchers developed a Graph Neural Network–based Elderly Service Prediction model (GNN-ESP) specifically to predict services for older adults and address this need. To demonstrate how the elderly are linked to their loved ones and to society as a whole, the technique presented uses weighted edges. Data on various aspects, including health, engagement frequency, and activity patterns, is compiled using a graph convolutional architecture. The next step is to identify key social connections using an aggregation approach that emphasizes attentiveness. It is necessary to make adjustments over time, as people’s needs for services and social interactions are constantly evolving. Compared with the F1-scores of normal machine learning and non-graph deep learning models, the GNN-ESP model achieved an 8–12% improvement. This model accurately forecasted service demand for senior social care in 94.1% of cases when applied to a real-world dataset. The fact that fraudulent service alerts occurred 17.6% less often is yet another indication that the newly recommended method is more trustworthy. As a result of the GNN-ESP design, intelligent systems can make decisions about expanding social services and elder care. The reason it is effective is that it keeps a record of social contacts and how those interactions develop into logical requests for help. Easy to set up and modify to meet this work changing requirements.