A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care

Background and objective: Diabetes is a global health concern, affecting millions of adults worldwide and exhibiting a growing prevalence. Managing the disease highly relies on continuous glucose monitoring, yet the dense and complex nature of electronic devices data streams poses significant challe...

ver descrição completa

Detalhes bibliográficos
Autores: Gaitán-Guerrero, Juan F., Martínez-Cruz, Carmen, Espinilla, Macarena, Díaz-Jiménez, David, López, José L.
Formato: artículo
Estado:Versión publicada
Fecha de publicación:2025
País:España
Recursos:Universidad de Jaén
Repositorio:RUJA. Repositorio Institucional de la Producción Científica de la Universidad de Jaén
OAI Identifier:oai:ruja.ujaen.es:10953/7428
Acesso em linha:https://www.sciencedirect.com/science/article/pii/S0169260725002950
https://hdl.handle.net/10953/7428
Access Level:acceso abierto
Palavra-chave:IoT-data fuzzy summarization
Large language models
Fine-tuning process
Continuous glucose monitoring
Prompt engineering
Evaluation methodology
004
004.3
004.4
004.7
id ES_b65e6545a72a45cad1dbb8d16effcebd
oai_identifier_str oai:ruja.ujaen.es:10953/7428
network_acronym_str ES
network_name_str España
repository_id_str
spelling A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes careGaitán-Guerrero, Juan F.Martínez-Cruz, CarmenEspinilla, MacarenaDíaz-Jiménez, DavidLópez, José L.IoT-data fuzzy summarizationLarge language modelsFine-tuning processContinuous glucose monitoringPrompt engineeringEvaluation methodology004004.3004.4004.7Background and objective: Diabetes is a global health concern, affecting millions of adults worldwide and exhibiting a growing prevalence. Managing the disease highly relies on continuous glucose monitoring, yet the dense and complex nature of electronic devices data streams poses significant challenges for efficient interpretation. Large Language Models are being widely applied across different domains for their ability to generate human-like text, but still fall short in producing accurate and meaningful text from raw data. To address this limitation, this study proposes a fine-tuning methodology tailored specifically to glucose data, but scalable to other expert-guided domains, enabling the models to generate concise, relevant and safe summaries, bridging the gap between raw data and efficient medical attention. Methods: This study introduces a novel continuous glucose monitoring framework that involves fine-tuned GPT models using structured datasets generated through an expert-guided data modeling based on Fuzzy Logic and prompt engineering for task contextualization. A new evaluation methodology is defined to assess the performance of the Large Language Models across different critical domains where expert knowledge is fundamental to characterize temporally dependent data and ensure valuable insights. Results: Fine-tuned GPT-4o achieved the highest performance, with an average score of 96% across all metrics. GPT-4o-mini followed with 76% score, while GPT-3.5 scored 72%. The use of fuzzy knowledge-based prompts proved more effective in scenarios with full data availability, or in scenarios with a simplified data availability when the models are not fine-tuned; domain-guided prompts improved output relevance and stability in fine-tuned models with less data availability. Conclusions: These results indicate the capability of our methods to align Large Language Models with the task of generating human-like text from raw data, highlighting their potential to manage diabetes by complex glucose patterns interpretation, alleviating the burden on healthcare systems.Elsevier202620262025info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfhttps://www.sciencedirect.com/science/article/pii/S0169260725002950https://hdl.handle.net/10953/7428reponame:RUJA. Repositorio Institucional de la Producción Científica de la Universidad de Jaéninstname:Universidad de JaénInglésComputer Methods and Programs in Biomedicine 2025;269;108878Attribution-NonCommercial-NoDerivs 3.0 Spainhttp://creativecommons.org/licenses/by-nc-nd/3.0/es/info:eu-repo/semantics/openAccessoai:ruja.ujaen.es:10953/74282026-06-24T12:41:07Z
dc.title.none.fl_str_mv A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
title A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
spellingShingle A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
Gaitán-Guerrero, Juan F.
IoT-data fuzzy summarization
Large language models
Fine-tuning process
Continuous glucose monitoring
Prompt engineering
Evaluation methodology
004
004.3
004.4
004.7
title_short A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
title_full A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
title_fullStr A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
title_full_unstemmed A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
title_sort A novel fine-tuning and evaluation methodology for large language models on IoT raw data summaries (LLM-RawDMeth): A joint perspective in diabetes care
dc.creator.none.fl_str_mv Gaitán-Guerrero, Juan F.
Martínez-Cruz, Carmen
Espinilla, Macarena
Díaz-Jiménez, David
López, José L.
author Gaitán-Guerrero, Juan F.
author_facet Gaitán-Guerrero, Juan F.
Martínez-Cruz, Carmen
Espinilla, Macarena
Díaz-Jiménez, David
López, José L.
author_role author
author2 Martínez-Cruz, Carmen
Espinilla, Macarena
Díaz-Jiménez, David
López, José L.
author2_role author
author
author
author
dc.subject.none.fl_str_mv IoT-data fuzzy summarization
Large language models
Fine-tuning process
Continuous glucose monitoring
Prompt engineering
Evaluation methodology
004
004.3
004.4
004.7
topic IoT-data fuzzy summarization
Large language models
Fine-tuning process
Continuous glucose monitoring
Prompt engineering
Evaluation methodology
004
004.3
004.4
004.7
description Background and objective: Diabetes is a global health concern, affecting millions of adults worldwide and exhibiting a growing prevalence. Managing the disease highly relies on continuous glucose monitoring, yet the dense and complex nature of electronic devices data streams poses significant challenges for efficient interpretation. Large Language Models are being widely applied across different domains for their ability to generate human-like text, but still fall short in producing accurate and meaningful text from raw data. To address this limitation, this study proposes a fine-tuning methodology tailored specifically to glucose data, but scalable to other expert-guided domains, enabling the models to generate concise, relevant and safe summaries, bridging the gap between raw data and efficient medical attention. Methods: This study introduces a novel continuous glucose monitoring framework that involves fine-tuned GPT models using structured datasets generated through an expert-guided data modeling based on Fuzzy Logic and prompt engineering for task contextualization. A new evaluation methodology is defined to assess the performance of the Large Language Models across different critical domains where expert knowledge is fundamental to characterize temporally dependent data and ensure valuable insights. Results: Fine-tuned GPT-4o achieved the highest performance, with an average score of 96% across all metrics. GPT-4o-mini followed with 76% score, while GPT-3.5 scored 72%. The use of fuzzy knowledge-based prompts proved more effective in scenarios with full data availability, or in scenarios with a simplified data availability when the models are not fine-tuned; domain-guided prompts improved output relevance and stability in fine-tuned models with less data availability. Conclusions: These results indicate the capability of our methods to align Large Language Models with the task of generating human-like text from raw data, highlighting their potential to manage diabetes by complex glucose patterns interpretation, alleviating the burden on healthcare systems.
publishDate 2025
dc.date.none.fl_str_mv 2025
2026
2026
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv https://www.sciencedirect.com/science/article/pii/S0169260725002950
https://hdl.handle.net/10953/7428
url https://www.sciencedirect.com/science/article/pii/S0169260725002950
https://hdl.handle.net/10953/7428
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Computer Methods and Programs in Biomedicine 2025;269;108878
dc.rights.none.fl_str_mv Attribution-NonCommercial-NoDerivs 3.0 Spain
http://creativecommons.org/licenses/by-nc-nd/3.0/es/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv Attribution-NonCommercial-NoDerivs 3.0 Spain
http://creativecommons.org/licenses/by-nc-nd/3.0/es/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:RUJA. Repositorio Institucional de la Producción Científica de la Universidad de Jaén
instname:Universidad de Jaén
instname_str Universidad de Jaén
reponame_str RUJA. Repositorio Institucional de la Producción Científica de la Universidad de Jaén
collection RUJA. Repositorio Institucional de la Producción Científica de la Universidad de Jaén
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869417441015103488
score 15.812429

Registos relacionados