Workshop Proceedings of the 19th International AAAI Conference on Web and Social Media

Workshop: #SMM4H-HeaRD 2025: Joint 10th Social Media Mining for Health and Health Real-World Data Workshop and Shared Tasks

DOI: 10.36190/2025.67

Published: 2025-06-05
NoviceTrio at #SMM4H-HeaRD 2025: Leveraging LLMs and Ensemble Methods for Classifying Dementia Family Caregivers on Twitter
Abir Naskar, Liuliu Chen, Jemima Kang, Mike Conway

Traditional interventions for dementia caregivers remain inaccessible and research has yet to explore the use of social media platforms to provide internet-based support. Task 3 of the 10th Social Media Mining for Health Workshop (SMM4H 2025) focuses on the task of binary classification of dementia family caregivers on Twitter. This paper presents a system that evaluates multiple pretrained language models, including BERT variants and LLMs, using ensembling and prompting strategies. Our LLM ensemble achieved the best performance with an F1 score of 0.957. We discuss the strengths and limitations of LLMs in this context.