Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación
En la gestión hospitalaria, confiar en las predicciones de una inteligencia artificial (IA) tiene un riesgo asociado, en entornos de alto riesgo como la salud, es esencial entender el por qué los modelos llegan a sus conclusiones. La explicabilidad hace posible identificar las características más re...
- Autores:
-
Banquez Cabarcas, Andrés
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2025
- Institución:
- Universidad de Antioquia
- Repositorio:
- Repositorio UdeA
- Idioma:
- spa
- OAI Identifier:
- oai:bibliotecadigital.udea.edu.co:10495/48393
- Acceso en línea:
- https://hdl.handle.net/10495/48393
- Palabra clave:
- Computers, Pipeline
Pipelines
Administración Hospitalaria
Hospital Administration
Inteligencia artificial
Artificial intelligence
http://id.loc.gov/authorities/subjects/sh85029569
https://id.nlm.nih.gov/mesh/D006739
- Rights
- openAccess
- License
- http://creativecommons.org/licenses/by-nc-sa/4.0/
| id |
UDEA2_bd8faae56ed5ae149f23ed5cd70bece6 |
|---|---|
| oai_identifier_str |
oai:bibliotecadigital.udea.edu.co:10495/48393 |
| network_acronym_str |
UDEA2 |
| network_name_str |
Repositorio UdeA |
| repository_id_str |
|
| dc.title.spa.fl_str_mv |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| title |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| spellingShingle |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación Computers, Pipeline Pipelines Administración Hospitalaria Hospital Administration Inteligencia artificial Artificial intelligence http://id.loc.gov/authorities/subjects/sh85029569 https://id.nlm.nih.gov/mesh/D006739 |
| title_short |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| title_full |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| title_fullStr |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| title_full_unstemmed |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| title_sort |
Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de Investigación |
| dc.creator.fl_str_mv |
Banquez Cabarcas, Andrés |
| dc.contributor.advisor.none.fl_str_mv |
Santana Velásquez, Angelower Salazar Sanchez, María Bernarda |
| dc.contributor.author.none.fl_str_mv |
Banquez Cabarcas, Andrés |
| dc.contributor.researchgroup.none.fl_str_mv |
Intelligent Information Systems Lab. |
| dc.subject.lcsh.none.fl_str_mv |
Computers, Pipeline Pipelines |
| topic |
Computers, Pipeline Pipelines Administración Hospitalaria Hospital Administration Inteligencia artificial Artificial intelligence http://id.loc.gov/authorities/subjects/sh85029569 https://id.nlm.nih.gov/mesh/D006739 |
| dc.subject.decs.none.fl_str_mv |
Administración Hospitalaria Hospital Administration |
| dc.subject.lemb.none.fl_str_mv |
Inteligencia artificial Artificial intelligence |
| dc.subject.lcshuri.none.fl_str_mv |
http://id.loc.gov/authorities/subjects/sh85029569 |
| dc.subject.meshuri.none.fl_str_mv |
https://id.nlm.nih.gov/mesh/D006739 |
| description |
En la gestión hospitalaria, confiar en las predicciones de una inteligencia artificial (IA) tiene un riesgo asociado, en entornos de alto riesgo como la salud, es esencial entender el por qué los modelos llegan a sus conclusiones. La explicabilidad hace posible identificar las características más relevantes tanto para las predicciones globales, como para las predicciones locales de un sistema, lo que permite a los usuarios comprender la razón detrás de cada decisión hecha por el modelo, ya sea a nivel de un caso individual o la totalidad del sistema. Este proyecto propone un pipeline para seleccionar el mejor algoritmo de explicabilidad, enfocándose en el contexto de la gestión hospitalaria, considerando la importancia de la transparencia de la inteligencia artificial contextualizada en entornos médicos. La metodología incluye una revisión bibliográfica de las técnicas existentes y su uso en diferentes aplicaciones, incluyendo el ámbito de la salud, con este propósito, se definirán métricas para evaluar la calidad de sus explicaciones y comparar las distintas opciones disponibles. Luego, se desarrollará el pipeline para poder seleccionar e implementar la alternativa que cumpla con los criterios y objetivos establecidos para cada trabajo en particular. Finalmente, se probarán los resultados obtenidos con datos cuyos valores esperados sean conocidos, de manera que pueda confirmarse que los cálculos obtenidos son correctos. Tras el desarrollo del pipeline, se utilizó un modelo de Random Forest con datos de MIMIC IV, SHAP demostró mayor fidelidad y un comportamiento ideal para entornos que requieren trazabilidad. Mientras que LIME ofreció explicaciones simples, rápidas, útiles para decisiones rápidas. |
| publishDate |
2025 |
| dc.date.accessioned.none.fl_str_mv |
2025-11-25T21:03:06Z |
| dc.date.issued.none.fl_str_mv |
2026 |
| dc.type.none.fl_str_mv |
Trabajo de grado - Pregrado |
| dc.type.coar.none.fl_str_mv |
http://purl.org/coar/resource_type/c_7a1f |
| dc.type.redcol.none.fl_str_mv |
http://purl.org/redcol/resource_type/TP |
| dc.type.content.none.fl_str_mv |
Text |
| dc.type.coarversion.none.fl_str_mv |
http://purl.org/coar/version/c_b1a7d7d4d402bcce |
| dc.type.driver.none.fl_str_mv |
info:eu-repo/semantics/bachelorThesis |
| dc.type.version.none.fl_str_mv |
info:eu-repo/semantics/draft |
| format |
http://purl.org/coar/resource_type/c_7a1f |
| status_str |
draft |
| dc.identifier.uri.none.fl_str_mv |
https://hdl.handle.net/10495/48393 |
| url |
https://hdl.handle.net/10495/48393 |
| dc.language.iso.none.fl_str_mv |
spa |
| language |
spa |
| dc.relation.references.none.fl_str_mv |
M. Christopher, G. Casalicchio and B. Birschl, "Interpretable Machine Learning - A Brief History, State-of-the-Art and Challenges," in Joint European conference on machine learning and knowledge discovery in databases, 2020, pp. 417-431. P. A, "Asking 'Why' in AI: Explainability of intelligent systems - Perspectives and challenges," Intelligent Systems In Accounting Finance & Management, vol. 25, no. 2, pp. 63-71, 2018. S. Lumbreras, "Inteligencia Artificial y medicina: la necesidad de modelos interpretables," TECHNO REVIEW International Techonology Science and Society Review, vol. 9, no. 2, pp. 97-102, 2021. C. Molnar, Interpretable Machine Learning, 2020. C. Maußner, M. Oberascher, A. Antengruber, A. Kahl and R. Sitzenfrei, "Explainable Artificial Intelligence for Reliable Water Demand Forecasting to Increase Trust in Predictions," Water Research, vol. 268, p. 122, 2024. H. Tang and Et. al, "Analysis and evaluation of explainable artificial intelligence on suicide risk assessment," Scientific Reports, vol. 14, no. 1, 2024. dinisurunisal, "Suicide-Risk-Prediction-Project," 04 Mayo 2020. [Online]. Available: https://github.com/dinisurunisal/Suicide-Risk-Prediction-Project. P. N. Snirivasu, U. Sirisha, K. Sandeep, S. P. Praveen, L. P. Maguluri and T. Bikku, "An Interpretable Approach with Explainable AI for Heart Stroke Prediction," Diagnosis, vol. 14, no. 2, p. 128, 2024. Á. Torres Martos and et al, "Multiomics and eXplainable artificial intelligence for decision support in insulin resistance early diagnosis: A pediatric population-based longitudinal study," Artificial Intelligence In Medicine, vol. 156, 2024. B. Davazdahemami, H. M. Zolbanin and D. Delen, "An explanatory analytics framework for early detection of chronic risk factors in pandemics," Healthcare Analytics, vol. 2, 2022. D. Bernard and et al, "Explainable machine learning framework to predict personalized physiological aging," Aging Cell, no. 8, p. 22, 2023. H. Kang, "The prevention and handling of the missing data," Korean J. Anesthesiol, vol. 64, no. 5, pp. 402-406, 2013. S. A. A. Kharis and A. H. A. Zili, "Predicting life expectancy of lung cancer patients after thoracic surgery using SMOTE and machine learning approaches," J. Nat, vol. 23, no. 3, pp. 152-161, 2023. W. Zhu, R. Qiu and Y. Fu, "Comparative Study on the Performance of Categorical Variable Encoders in Classification and Regression Tasks," arXiv, 2024. S. Kumari and B. Jayaram, "Measuring Concentration of Distances - An Effective and Efficient Empirical Index," IEEE Transactions on Knowledge and Data Engineering, vol. 29, no. 2, pp. 373-386, 2017. Scikit-learn, "StandardScaler," Scikit-learn, [Online]. Available: https://sckit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html. Scikit-learn, "MinMaxScaler," Scikit-learn, [Online]. Available: https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html. Scikit-learn, "1.10. Decision Trees," Scikit-learn, [Online]. Available: https://scikit-learn.org/stable/modules/tree.html. O. Rainio, J. Teuho and R. Klén, "Evaluation metrics and statistical test for machine learning," Scientific Rep, vol. 14, no. 1, 2024. S. Lumberg and S. I. Lee, "A unified approach to interpreting model predictions," arXiv, 2017. |
| dc.rights.uri.none.fl_str_mv |
http://creativecommons.org/licenses/by-nc-sa/4.0/ |
| dc.rights.accessrights.none.fl_str_mv |
info:eu-repo/semantics/openAccess |
| dc.rights.license.en.fl_str_mv |
Attribution-NonCommercial-ShareAlike 4.0 International |
| dc.rights.coar.none.fl_str_mv |
http://purl.org/coar/access_right/c_abf2 |
| rights_invalid_str_mv |
http://creativecommons.org/licenses/by-nc-sa/4.0/ Attribution-NonCommercial-ShareAlike 4.0 International http://purl.org/coar/access_right/c_abf2 |
| eu_rights_str_mv |
openAccess |
| dc.format.extent.none.fl_str_mv |
39 páginas |
| dc.format.mimetype.none.fl_str_mv |
application/pdf |
| dc.publisher.none.fl_str_mv |
Universidad de Antioquia |
| dc.publisher.program.none.fl_str_mv |
Bioingeniería |
| dc.publisher.place.none.fl_str_mv |
Medellín, Colombia |
| dc.publisher.faculty.none.fl_str_mv |
Facultad de Ingeniería |
| dc.publisher.branch.none.fl_str_mv |
Campus Medellín - Ciudad Universitaria |
| publisher.none.fl_str_mv |
Universidad de Antioquia |
| institution |
Universidad de Antioquia |
| bitstream.url.fl_str_mv |
https://bibliotecadigital.udea.edu.co/bitstreams/07542e42-1db2-496e-a67b-564196ef3b98/download https://bibliotecadigital.udea.edu.co/bitstreams/49fb266f-a729-4b0a-a5bf-650fd8bbd40a/download https://bibliotecadigital.udea.edu.co/bitstreams/b7d99f7b-d60e-4bf9-b729-92b62704d92a/download https://bibliotecadigital.udea.edu.co/bitstreams/26ca0d99-ecf6-4a06-a919-2dc8faf9490d/download https://bibliotecadigital.udea.edu.co/bitstreams/523652ed-f8d3-47f1-8d5f-fe7158b879b8/download https://bibliotecadigital.udea.edu.co/bitstreams/6cc48cdb-a2aa-482c-a661-4e61d6c464c2/download https://bibliotecadigital.udea.edu.co/bitstreams/104c32d7-654b-423c-b654-5032fdd79bdf/download https://bibliotecadigital.udea.edu.co/bitstreams/3f893c67-dd75-4add-a032-e2e3c24e0568/download |
| bitstream.checksum.fl_str_mv |
8c42cc3d6cd517eacfb91446dfa35c7d e10838b9808c406b8d57c9dc42b3394b b76e7a76e24cf2f94b3ce0ae5ed275d0 5643bfd9bcf29d560eeec56d584edaa9 02d7c9aafe7d53a7468e7026f2baeee8 fdf250596dd73ac5645f04b55eb5c3f7 6a6ca98e692d3476e2d28f72814e97ee b65aff6670eaa5a15936afd8a9a825cb |
| bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 MD5 MD5 MD5 MD5 |
| repository.name.fl_str_mv |
Repositorio Institucional de la Universidad de Antioquia |
| repository.mail.fl_str_mv |
aplicacionbibliotecadigitalbiblioteca@udea.edu.co |
| _version_ |
1851052418575368192 |
| spelling |
Santana Velásquez, AngelowerSalazar Sanchez, María BernardaBanquez Cabarcas, AndrésIntelligent Information Systems Lab.2025-11-25T21:03:06Z2026https://hdl.handle.net/10495/48393En la gestión hospitalaria, confiar en las predicciones de una inteligencia artificial (IA) tiene un riesgo asociado, en entornos de alto riesgo como la salud, es esencial entender el por qué los modelos llegan a sus conclusiones. La explicabilidad hace posible identificar las características más relevantes tanto para las predicciones globales, como para las predicciones locales de un sistema, lo que permite a los usuarios comprender la razón detrás de cada decisión hecha por el modelo, ya sea a nivel de un caso individual o la totalidad del sistema. Este proyecto propone un pipeline para seleccionar el mejor algoritmo de explicabilidad, enfocándose en el contexto de la gestión hospitalaria, considerando la importancia de la transparencia de la inteligencia artificial contextualizada en entornos médicos. La metodología incluye una revisión bibliográfica de las técnicas existentes y su uso en diferentes aplicaciones, incluyendo el ámbito de la salud, con este propósito, se definirán métricas para evaluar la calidad de sus explicaciones y comparar las distintas opciones disponibles. Luego, se desarrollará el pipeline para poder seleccionar e implementar la alternativa que cumpla con los criterios y objetivos establecidos para cada trabajo en particular. Finalmente, se probarán los resultados obtenidos con datos cuyos valores esperados sean conocidos, de manera que pueda confirmarse que los cálculos obtenidos son correctos. Tras el desarrollo del pipeline, se utilizó un modelo de Random Forest con datos de MIMIC IV, SHAP demostró mayor fidelidad y un comportamiento ideal para entornos que requieren trazabilidad. Mientras que LIME ofreció explicaciones simples, rápidas, útiles para decisiones rápidas.In hospital management, relying on AI predictions carries inherent risks, in high-risk environments such as healthcare, it is essential to understand why models arrive at their conclusions . Explainability makes it possible to identify the most relevant features both in global and local predictions, allowing users to comprehend the rationale behind each decision made by the system , whether on an individual case or at a systemic level. This project proposes a pipeline to select the most suitable explainability algorithm, focusing on the context of hospital management and emphasizing the importance of transparency in AI applied to medical settings. The methodology includes a literature review of existing techniques and their applications, particularly in healthcare. Metrics will be defined to evaluate the quality of the explanations and compare different options. Subsequently, the pipeline will be developed to implement the alternative that meets the criteria and objectives for each particular work. Finally, the system will be tested using data with known outcomes to validate the correctness of the results. After developing the pipeline, a Random Forest model was trained using MIMIC-IV data. SHAP showed higher fidelity and proved suitable for contexts requiring traceability, while LIME provided simpler and faster explanations, making it more useful for real-time clinical decision making.PregradoBioingeniero39 páginasapplication/pdfspaUniversidad de AntioquiaBioingenieríaMedellín, ColombiaFacultad de IngenieríaCampus Medellín - Ciudad Universitariahttp://creativecommons.org/licenses/by-nc-sa/4.0/info:eu-repo/semantics/openAccessAttribution-NonCommercial-ShareAlike 4.0 Internationalhttp://purl.org/coar/access_right/c_abf2Computers, PipelinePipelinesAdministración HospitalariaHospital AdministrationInteligencia artificialArtificial intelligencehttp://id.loc.gov/authorities/subjects/sh85029569https://id.nlm.nih.gov/mesh/D006739Análisis de códigos fuente de métodos de explicabilidad de inteligencia artificial. Proyecto de InvestigaciónTrabajo de grado - Pregradohttp://purl.org/coar/resource_type/c_7a1fhttp://purl.org/redcol/resource_type/TPTexthttp://purl.org/coar/version/c_b1a7d7d4d402bcceinfo:eu-repo/semantics/bachelorThesisinfo:eu-repo/semantics/draftM. Christopher, G. Casalicchio and B. Birschl, "Interpretable Machine Learning - A Brief History, State-of-the-Art and Challenges," in Joint European conference on machine learning and knowledge discovery in databases, 2020, pp. 417-431.P. A, "Asking 'Why' in AI: Explainability of intelligent systems - Perspectives and challenges," Intelligent Systems In Accounting Finance & Management, vol. 25, no. 2, pp. 63-71, 2018.S. Lumbreras, "Inteligencia Artificial y medicina: la necesidad de modelos interpretables," TECHNO REVIEW International Techonology Science and Society Review, vol. 9, no. 2, pp. 97-102, 2021.C. Molnar, Interpretable Machine Learning, 2020.C. Maußner, M. Oberascher, A. Antengruber, A. Kahl and R. Sitzenfrei, "Explainable Artificial Intelligence for Reliable Water Demand Forecasting to Increase Trust in Predictions," Water Research, vol. 268, p. 122, 2024.H. Tang and Et. al, "Analysis and evaluation of explainable artificial intelligence on suicide risk assessment," Scientific Reports, vol. 14, no. 1, 2024.dinisurunisal, "Suicide-Risk-Prediction-Project," 04 Mayo 2020. [Online]. Available: https://github.com/dinisurunisal/Suicide-Risk-Prediction-Project.P. N. Snirivasu, U. Sirisha, K. Sandeep, S. P. Praveen, L. P. Maguluri and T. Bikku, "An Interpretable Approach with Explainable AI for Heart Stroke Prediction," Diagnosis, vol. 14, no. 2, p. 128, 2024.Á. Torres Martos and et al, "Multiomics and eXplainable artificial intelligence for decision support in insulin resistance early diagnosis: A pediatric population-based longitudinal study," Artificial Intelligence In Medicine, vol. 156, 2024.B. Davazdahemami, H. M. Zolbanin and D. Delen, "An explanatory analytics framework for early detection of chronic risk factors in pandemics," Healthcare Analytics, vol. 2, 2022.D. Bernard and et al, "Explainable machine learning framework to predict personalized physiological aging," Aging Cell, no. 8, p. 22, 2023.H. Kang, "The prevention and handling of the missing data," Korean J. Anesthesiol, vol. 64, no. 5, pp. 402-406, 2013.S. A. A. Kharis and A. H. A. Zili, "Predicting life expectancy of lung cancer patients after thoracic surgery using SMOTE and machine learning approaches," J. Nat, vol. 23, no. 3, pp. 152-161, 2023.W. Zhu, R. Qiu and Y. Fu, "Comparative Study on the Performance of Categorical Variable Encoders in Classification and Regression Tasks," arXiv, 2024.S. Kumari and B. Jayaram, "Measuring Concentration of Distances - An Effective and Efficient Empirical Index," IEEE Transactions on Knowledge and Data Engineering, vol. 29, no. 2, pp. 373-386, 2017.Scikit-learn, "StandardScaler," Scikit-learn, [Online]. Available: https://sckit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html.Scikit-learn, "MinMaxScaler," Scikit-learn, [Online]. Available: https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html.Scikit-learn, "1.10. Decision Trees," Scikit-learn, [Online]. Available: https://scikit-learn.org/stable/modules/tree.html.O. Rainio, J. Teuho and R. Klén, "Evaluation metrics and statistical test for machine learning," Scientific Rep, vol. 14, no. 1, 2024.S. Lumberg and S. I. Lee, "A unified approach to interpreting model predictions," arXiv, 2017.PublicationORIGINALBanquezAndres_2025_AnalisisCodigosFuente.pdfBanquezAndres_2025_AnalisisCodigosFuente.pdfTrabajo de grado de pregradoapplication/pdf691244https://bibliotecadigital.udea.edu.co/bitstreams/07542e42-1db2-496e-a67b-564196ef3b98/download8c42cc3d6cd517eacfb91446dfa35c7dMD56falseAnonymousREADPoster.pdfPoster.pdfAnexoapplication/pdf765521https://bibliotecadigital.udea.edu.co/bitstreams/49fb266f-a729-4b0a-a5bf-650fd8bbd40a/downloade10838b9808c406b8d57c9dc42b3394bMD54trueAnonymousREADLICENSElicense.txtlicense.txttext/plain; charset=utf-814837https://bibliotecadigital.udea.edu.co/bitstreams/b7d99f7b-d60e-4bf9-b729-92b62704d92a/downloadb76e7a76e24cf2f94b3ce0ae5ed275d0MD52falseAnonymousREADCC-LICENSElicense_rdflicense_rdfapplication/rdf+xml; charset=utf-81160https://bibliotecadigital.udea.edu.co/bitstreams/26ca0d99-ecf6-4a06-a919-2dc8faf9490d/download5643bfd9bcf29d560eeec56d584edaa9MD55falseAnonymousREADTEXTBanquezAndres_2025_AnalisisCodigosFuente.pdf.txtBanquezAndres_2025_AnalisisCodigosFuente.pdf.txtExtracted texttext/plain67586https://bibliotecadigital.udea.edu.co/bitstreams/523652ed-f8d3-47f1-8d5f-fe7158b879b8/download02d7c9aafe7d53a7468e7026f2baeee8MD57falseAnonymousREADPoster.pdf.txtPoster.pdf.txtExtracted texttext/plain2391https://bibliotecadigital.udea.edu.co/bitstreams/6cc48cdb-a2aa-482c-a661-4e61d6c464c2/downloadfdf250596dd73ac5645f04b55eb5c3f7MD59falseAnonymousREADTHUMBNAILBanquezAndres_2025_AnalisisCodigosFuente.pdf.jpgBanquezAndres_2025_AnalisisCodigosFuente.pdf.jpgGenerated Thumbnailimage/jpeg5953https://bibliotecadigital.udea.edu.co/bitstreams/104c32d7-654b-423c-b654-5032fdd79bdf/download6a6ca98e692d3476e2d28f72814e97eeMD58falseAnonymousREADPoster.pdf.jpgPoster.pdf.jpgGenerated Thumbnailimage/jpeg20029https://bibliotecadigital.udea.edu.co/bitstreams/3f893c67-dd75-4add-a032-e2e3c24e0568/downloadb65aff6670eaa5a15936afd8a9a825cbMD510falseAnonymousREAD10495/48393oai:bibliotecadigital.udea.edu.co:10495/483932025-11-26 04:06:31.666http://creativecommons.org/licenses/by-nc-sa/4.0/Attribution-NonCommercial-ShareAlike 4.0 Internationalopen.accesshttps://bibliotecadigital.udea.edu.coRepositorio Institucional de la Universidad de Antioquiaaplicacionbibliotecadigitalbiblioteca@udea.edu.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 |