Focusing on the problems of heterogeneous indicators, fuzzy evaluation boundaries and scattered improvement basis in the quality monitoring of elderly care services, this paper proposed a fuzzy comprehensive evaluation model for continuous improvement of social work. Based on the digital platform of elderly care service, the work order log, check-in record, return visit questionnaire, case file and supervision account are incorporated into the unified computing framework, and a multi-layer index system covering service safety, service process, service experience and improvement effect is constructed. Weight calculation, fuzzy evaluation matrix and comprehensive scoring mechanism are combined. To realize the quantitative identification and classification of the quality of pension service. The results show that the comprehensive score of the sample ranges from 48.6 to 92.4, and the average score is 76.8. The consistency rate of the model level discrimination reaches 92.5%, and the level maintenance rate is 85.0% under the condition of ±10% data perturbation. Compared with the traditional analytic Hierarchy process and simple weighted average method, the proposed model performs better in boundary sample identification, result stability and fuzzy information processing. The research shows that this method can adapt to the monitoring needs of multi-source data continuous access in the smart elderly care scenario, and can synchronize the output of comprehensive score, level affiliation and dimensional shortcomings, which provides a quantitative basis for social work supervision, case review and service rectification.
