University of Twente Student Theses
Exploration of Different Large Language Models for Retrieval-Augmented Generation in Analyzing Wearable Running Data for Sports Physiotherapy
Chiras, Marios (2024) Exploration of Different Large Language Models for Retrieval-Augmented Generation in Analyzing Wearable Running Data for Sports Physiotherapy.
|
PDF
656kB |
| Abstract: | Large Language Models have shown great results in analyzing and processing different types of datasets. In this study, different parameter-size LLMs will be evaluated in a physiotherapist agent application, that process biomechanical running data from wearable devices. The research aims to determine the optimal size of the state-of-the-art LLMs used in this research that balances accuracy and computational resources. The study will explore and evaluate the models in a Retrieval-Augmented Generation (RAG) application. The different models used in this study were divided into 3 categories Small, Medium, and Large models based on their parameter size. In the experiment, the Larger category was shown to offer the best balance of the highest accuracy when it comes to detecting outliers within the biomechanical data while balancing computational resources. The study highlights a unique approach to developing physiotherapist applications with the integration of RAG and Text-To-SQL methodologies for processing and retrieving running data. The goal of this application is to provide valuable insights into the trade-offs between the accuracy and computation efficiency of different models. These findings contribute to the development of applications in sports physiotherapy, offering insights into the appropriate selection of LLMs in specific circumstances. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/101017 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page