# Identifying vulnerable groups of community-dwelling older adults with a strong willingness to receive volunteer services based on machine learning methods

## TL;DR

This study uses machine learning to identify which older adults are most likely to benefit from volunteer services, based on factors like living alone and health status.

## Contribution

The study introduces a novel approach using machine learning to identify vulnerable older adults with high service demand and willingness to receive volunteer support.

## Key findings

- Logistic regression and random forest models effectively predicted willingness to receive volunteer services.
- Key predictors include living alone, poor health, and lack of caregiving support.
- Depression and absence of government care were significant predictors of service demand.

## Abstract

Although volunteer services play an increasingly important role in addressing the aging crisis, there is still insufficient evidence on which community-based older adult groups should be prioritized for service provision. This study aims to identify the key characteristics of vulnerable community-dwelling older adults with strong willingness and high demand for volunteer services.

This study employed a cross-sectional descriptive design. A total of 852 community-dwelling older adults from four central cities in China were surveyed between March 25 and May 5, 2022, using a convenience sampling method. Data were analyzed using SPSS 26.0 and advanced machine learning techniques in Python.

Logistic regression (AUC = 0.975) and random forest (AUC = 0.970) achieved the best performance in predicting willingness, with the most critical predictors being absence of cohabiting family care and caregiving support, living alone, poor health and depressive status, having basic medical insurance, and advanced age. For demand level, multiple linear regression (R² = 0.230) performed best, identifying depression status, poor health, willingness to help others, and absence of government care as significant predictors. Model comparisons demonstrated robust and consistent variable importance rankings across algorithms.

This study provides a scientific basis for developing more targeted and efficient volunteer service strategies, enabling closer alignment with older adults’ needs and more effective allocation of resources, thereby minimizing public resource waste. Its implementation has significant theoretical and practical value for optimizing service distribution.

The online version contains supplementary material available at 10.1186/s12889-025-24902-7.