Modelo de predicción de accidentes con alejamiento usando técnicas de aprendizaje automático
DOI:
https://doi.org/10.46421/sbtic.v5i00.7473Palabras clave:
Gestão de segurança, Indústria 4.0, Inteligência Artificial, AcidentesResumen
En la construcción civil, los accidentes con alejamiento implican un alto costo para la salud y la seguridad social, además de los costos indirectos para la empresa y los daños ocasionados al trabajador. En este contexto, el aprendizaje automático (AA) se destaca como una tecnología prometedora para agilizar el análisis de grandes volúmenes de datos y predecir eventos, lo que contribuye a la gestión preventiva de accidentes. Por lo tanto, este artículo tiene como objetivo desarrollar un modelo de predicción de accidentes con alejamiento utilizando algoritmos de AA y datos históricos de accidentes. El conjunto de datos comprende 28.100 accidentes (incidentes en la construcción civil – trabajadores regulares), ocurridos entre 2018 y 2023, distribuidos en 15 variables binarias y categóricas. Tras el preprocesamiento, las predicciones se realizaron mediante siete modelos con diferentes clasificadores. El modelo Random Forest presentó el mejor desempeño, alcanzando una precisión (accuracy) de 0,775, una precisión (precision) de 0,768 y una sensibilidad (recall) de 0,776. Como contribución, este estudio demuestra el potencial de la tecnología para el análisis y la predicción de accidentes con alejamiento, además de ofrecer información sobre los atributos que más influyen en la ocurrencia de estos eventos. No obstante, el estudio resalta la necesidad de incluir atributos relacionados con el contexto del accidente, con el fin de aumentar la precisión y la robustez del modelo.
Descargas
Citas
BAKER, Henrietta; HALLOWELL, Matthew R.; TIXIER, Antoine J.-P. AI-based prediction of independent construction safety outcomes from universal attributes. Automation in Construction, [s.l.], v. 118, 2020, p. 103146. ISSN 0926-5805. doi: https://doi.org/10.1016/j.autcon.2020.103146.
BORGES, Nathália de Faria; VILAÇA, Isabela Pessanha; LAURINDO, Quézia Manuela Gonçalves. Acidentes do trabalho e cultura de segurança no setor da construção civil. Revista Perspectivas Online: Exatas & Engenharia, v. 11, n. 33, p. 19-33, out. 2021. ISSN 2236-885X (Online). doi: https://doi.org/10.25242/885X113320212353.
BRASIL. Ministério da Previdência Social. Anuário Estatístico da Previdência Social – 2023. Brasília, DF, 2024. Disponível em: https://www.gov.br/previdencia/pt-br/noticias/2024/dezembro/anuario-estatistico-da-previdencia-social-2023-ja-esta-disponivel-para-consulta. Acesso em: 05 jan. 2025.
BRASIL. Ministério do Trabalho e Emprego. Relatório de Transparência Salarial de Mulheres e Homens. Brasília, DF, 2024. Disponível em: https://www.gov.br/trabalho-e-emprego/pt-br/noticias-e-conteudo/2024/Fevereiro/mte-e-mmulheres-tiram-duvidas-sobre-o-relatorio-de-transparencia-salarial/copy_of_RelatriodeIgualdadesalarialdeMulhereseHomens_09022024.pptx. Acesso em: 10 fev. 2025.
CHOI, Jongko; GU, Bonsung; CHIN, Sangyoon; LEE, Jong-Seok. Machine learning predictive model based on national data for fatal accidents of construction workers. Automation in Construction, [s.l.], v. 110, 2020. ISSN 0926-5805. doi: https://doi.org/10.1016/j.autcon.2019.102974.
EWALD, Fiona Katharina; BOTHMANN, Ludwig; WRIGHT, Marvin N.; BISCHL, Bernd; CASALICCHIO, Giuseppe; KÖNIG, Gunnar. A guide to feature importance methods for scientific inference. In: LONGO, L.; LAPUSCHKIN, S.; SEIFERT, C. (eds.). Explainable Artificial Intelligence. xAI 2024. Communications in Computer and Information Science, v. 2154. Cham: Springer, 2024. ISSN 1865-0929. doi: https://doi.org/10.48550/arXiv.2404.12862.
FONTANEDA, Ignacio; CAMINO LÓPEZ, Miguel A.; GONZÁLEZ ALCÁNTARA, Oscar J.; GREINER, Birgit A. The “Weekday Effect”: A Decrease in Occupational Accidents from Monday to Friday—An Extension of the “Monday Effect”. BioMed Research International, v. 2024, p. 1-12, 2024. ISSN 2314-6133. doi: https://doi.org/10.1155/2024/4792081.
GOMIDES, Luciana de Melo; ABREU, Mery Natali Silva; ASSUNÇÃO, Ada Ávila. Desigualdades ocupacionais e diferenças de gênero: acidentes de trabalho, Brasil, 2019. Revista de Saúde Pública, v. 58, p. 13, 2024. ISSN 1518-8787. doi: http://dx.doi.org/10.11606/s1518-8787.2024058005342.
JAFARI, Parinaz; MOHAMED, Emad; PEREIRA, Estacio; KANG, Shih-Chung; ABOURIZK, Simaan. Leading Safety Indicators: Application of Machine Learning for Safety Performance Measurement. ISARC - International Symposium on Automation and Robotics in Construction, 2019. ISSN 2413-5844. doi: https://doi.org/10.22260/ISARC2019/0067.
KANG, Kyung-Su; KOO, Choongwan; RYU, Han-Guk. An interpretable machine learning approach for evaluating the feature importance affecting lost workdays at construction sites. Journal of Building Engineering, [s.l.], v. 53, 2022, p. 104534. ISSN 2352-7102. doi: https://doi.org/10.1016/j.jobe.2022.104534.
KANG, Kyungsu; RYU, Hanguk. Predicting types of occupational accidents at construction sites in Korea using random forest model. Safety Science, [s.l.], v. 120, 2019, p. 226-236. ISSN 0925-7535. doi: https://doi.org/10.1016/j.ssci.2019.06.034.
KIM, Ji-Myong; ADHIKARI, Manik Das; BAE, Junseo; YUMB, Sang-Guk. A deep neural network algorithm-based approach for predicting recovery period of accidents according to construction scale. Heliyon, v. 10, n. 11, e32215, 2024. doi: https://doi.org/10.1016/j.heliyon.2024.e32215.
KOC, Kerim; EKMEKCİOĞLU, Ömer; GURGUN, Asli Pelin. Determining susceptible body parts of construction workers due to occupational injuries using inclusive modelling. Safety Science, v. 164, 2023. ISSN 0925-7535. doi: https://doi.org/10.1016/j.ssci.2023.106157.
KOC, Kerim; EKMEKCİOĞLU, Ömer; GURGUN, Asli Pelin. Developing a National Data-Driven Construction Safety Management Framework with Interpretable Fatal Accident Prediction. Journal of Construction Engineering and Management, v. 149, n. 4, 2023, p. 04023010. ISSN: 0733-9364. doi: https://doi.org/10.1061/JCEMD4.COENG-12848
KOC, Kerim; EKMEKCIOĞLU, Ömer; GURGUN, Asli Pelin. Integrating feature engineering, genetic algorithm and tree-based machine learning methods to predict the post-accident disability status oconstruction workers. Automation in Construction, [s.l.], v. 131, 2021, p. 103896. ISSN 0926-5805. doi: https://doi.org/10.1016/j.autcon.2021.103896.
KOC, Kerim; EKMEKCIOĞLU, Ömer; GURGUN, Asli Pelin. Prediction of construction accident outcomes based on an imbalanced dataset through integrated resampling techniques and machine learning methods. Engineering, Construction and Architectural Management, v. 30, n. 10, p. 3792–3820, 2023. ISSN 0969-9988. doi: https://doi.org/10.1108/ECAM-04-2022-0305.
LEE, Jae Yun; YOON, Young Geun; OH, Tae Keun; PARK, Seunghee; RYU, Sang Il. A Study on Data Pre-Processing and Accident Prediction Modelling for Occupational Accident Analysis in the Construction Industry. Applied Sciences, v. 10, n. 21, p. 7949, 2020. ISSN 2076-3417. doi: https://doi.org/10.3390/app10217949.
MANU, Patrick A.; ANKRAH, Nii A.; PROVERBS, David G.; SURESH, Subashini. Investigating the multi-causal and complex nature of the accident causal influence of construction project features. Accident Analysis and Prevention, v. 48, p. 126-133, jan. 2012. ISSN 0001-4575. doi: https://doi.org/10.1016/j.aap.2011.05.008.
POH, Clive Q.X.; UBEYNARAYANA, Chalani Udhyami; GOH, Yang Miang. Safety leading indicators for construction sites: A machine learning approach. Automation in Construction, v. 93, p. 375-386, 2018. ISSN 0926-5805. doi: https://doi.org/10.1016/j.autcon.2018.03.022.
SMARTLAB. Relatório de Análise de Acidentes de Trabalho. São Paulo, 2024. Disponível em: https://smartlabbr.org/sst/localidade/0?dimensao=covid. Acesso em: 10 fev. 2025.
TIXIER, Antoine J.-P.; HALLOWELL, Matthew R.; TIXIER, Henrietta Baker. Safer Together: Machine Learning Models Trained on Shared Accident Datasets Predict Construction Injuries Better than Company-Specific Models. arXiv preprint arXiv:2301.03567, 2023. doi: https://doi.org/10.48550/arXiv.2301.03567.
TOPTANCÍ, Şura; ERGINEL, Nihal; ACAR, Ilgın. Predicting the severity of occupational accidents in the construction industry using standard and regularized logistic regression models. Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi, v. 12, n. 3, p. 778-798, 2023. doi: https://doi.org/10.28948/ngumuh.1212385.
TÖZER, Kemal Dirgen; GÜRCANLI, Gürkan Emre; YARKINER, Zalihe. Analysis of workday losses due to falls from scaffoldings in the construction industry. Journal of Construction Engineering, Management & Innovation, [s.l.], v. 5, n. 1, p. 15-27, 2022. ISSN 2630-5771. doi: https://doi.org/10.31462/jcemi.2022.01015027.
ZHU, Rongchen; HU, Xiaofeng; HOU, Jiaqi; LI, Xin. Application of machine learning techniques for predicting the consequences of construction accidents in China. Process Safety and Environmental Protection, [s.l.], v. 145, 2021, p. 293-302. ISSN 0957-5820. doi: https://doi.org/10.1016/j.psep.2020.08.006
Descargas
Publicado
Cómo citar
Número
Sección
Licencia
Derechos de autor 2025 SIMPOSIO BRASILEÑO SOBRE TECNOLOGÍAS DE LA INFORMACIÓN Y LA COMUNICACIÓN EN LA CONSTRUCCIÓN
Esta obra está bajo una licencia internacional Creative Commons Atribución 4.0.
Adota-se Atribuição 4.0 Internacional (CC BY 4.0). Ver detalhamento em https://creativecommons.org/licenses/by/4.0/deed.pt_BR
