Enhanced CO₂ Forecasting in Indoor Environments Using Advanced LSTM Models
DOI:
https://doi.org/10.61467/2007.1558.2026.v17i3.897Keywords:
Time Series Prediction, Machine Learning, Sick buildings syndrome, Memoria a corto y largo plazo, predicción de series temporales, aprendizaje automático, síndrome del edificio enfermoAbstract
Accurate forecasting of CO₂ levels in indoor environments is essential for effective air quality management and public health protection. This study assesses the performance of four Long Short-Term Memory (LSTM)-based models—LSTM, Spatial LSTM (sLSTM), Memory-Augmented LSTM (mLSTM), and Extended LSTM (xLSTM)—for CO₂ prediction. The results demonstrate that xLSTM consistently outperforms the others models across multiple evaluation metrics, establishing it as a highly reliable option for air quality monitoring. The research underscores the relevance of precise CO₂ forecasting for policymakers and building managers in optimizing indoor environments conditions. Future work will focus in integrating xLSTM into a real-time monitoring system powered by Big Data and Internet of Things (IoT) technologies, incorporating multiple forecasting algorithms. This approach aims to enhance indoor air quality management and support respiratory diseases prevention through continuous monitoring and advanced predictive analysis.
Spanish-language metadata / Metadatos en español
Título en español:
Mejora de la predicción de CO₂ en ambientes interiores mediante modelos LSTM avanzados
Resumen:
La predicción precisa de los niveles de CO₂ en ambientes interiores es esencial para una gestión eficaz de la calidad del aire y la protección de la salud pública. Este estudio evalúa el desempeño de cuatro modelos basados en la red de memoria a corto y largo plazo (LSTM) —LSTM, LSTM espacial (sLSTM), LSTM con memoria aumentada (mLSTM) y LSTM extendida (xLSTM)— para la predicción de CO₂. Los resultados demuestran que xLSTM supera sistemáticamente a los demás modelos en múltiples métricas de evaluación, lo que lo consolida como una opción altamente confiable para el monitoreo de la calidad del aire. La investigación destaca la importancia de contar con pronósticos precisos de CO₂ para que los responsables políticos y los administradores de edificios puedan optimizar las condiciones ambientales en los interiores. Los trabajos futuros se centrarán en integrar xLSTM en un sistema de monitoreo en tiempo real basado en tecnologías de Big Data e Internet de las cosas (IoT), que incorpore múltiples algoritmos de predicción. Este enfoque tiene como objetivo mejorar la gestión de la calidad del aire interior y contribuir a la prevención de enfermedades respiratorias mediante un monitoreo continuo y un análisis predictivo avanzado.
Palabras Claves:
Memoria a corto y largo plazo, predicción de series temporales, aprendizaje automático, síndrome del edificio enfermo
Smart citations:
https://scite.ai/reports/10.61467/2007.1558.2026.v17i3.897
Dimensions.
Open Alex.
References
Akinshin, A. (2022). Finite-sample bias-correction factors for the median absolute deviation based on the Harrell-Davis quantile estimator and its trimmed modification. https://doi.org/10.48550/arXiv.2207.12005
Beck, M., Pöppel, K., Spanring, M., Auer, A., Prudnikova, O., Kopp, M., Klambauer, G., Brandstetter, J., & Sepp Hochreiter (2024). XLSTM: Extended long short-term memory. https://doi.org/10.48550/arXiv.2405.04517
Bliemel, F. (1973). Theil’s forecast accuracy coefficient: A clarification. Journal of Marketing Research, 10(4), 444–446. https://doi.org/10.1177/002224377301000413
Dai, Z., Yuan, Y., Zhu, X., & Zhao, L. (2024). A method for predicting indoor CO₂ concentration in university classrooms: An RF-TPE-LSTM approach. Applied Sciences, 14(14), 6188. https://doi.org/10.3390/app14146188
Databot. (2023). Databot™: Real data, real science, real fun. https://databot.us.com
Domínguez Portillo, J., Rodríguez Peralta, L. M., Sampaio, P. N. M., & Nunes, É. de O. (2023). Determining environmental indicators related to the propagation of contagious diseases and health issues: A systematic literature review. In 2023 18th Iberian Conference on Information Systems and Technologies (CISTI) (pp. 1–5). https://doi.org/10.23919/CISTI58278.2023.10212010
Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29(5), 1189–1232. https://doi.org/10.1214/aos/1013203451
Gabriel, M., & Auer, T. (2023). LSTM deep learning models for virtual sensing of indoor air pollutants: A feasible alternative to physical sensors. Buildings, 13(7), 1684. https://doi.org/10.3390/buildings13071684
Gabriel, M. F., Marques, G., Filipe, D., Felgueiras, F., Cardoso, J. P., Azeredo, J., Kazdaridis, G., Symeonidis, P., Keranidis, S., Conradie, P., Azevedo, I., & Anagnostopoulos, F. (2024). Implementation of an IoT architecture for promoting healthy air quality in 84 homes of families with children. Building and Environment, 266, 112040. https://doi.org/10.1016/j.buildenv.2024.112040
Gokcesu, K., & Gokcesu, H. (2021). Generalized Huber loss for robust learning and its efficient minimization for robust statistics. https://doi.org/10.48550/arXiv.2108.12627
Sepp Hochreiter, S., & Jürgen Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
Inkbird. (2023). Wi-Fi 8-in-1 air quality monitor IAQM-128W. https://inkbird.com/products/wi-fi-air-monitor-iaqm-128w
Kreinovich, V., Nguyen, H. T., & Ouncharoen, R. (2014). How to estimate forecasting quality: A system-motivated derivation of SMAPE. University of Texas at El Paso. https://scholarworks.utep.edu/cs_techrep/865
Linares Alzamora, R. G., Maia Sampaio, P. N., Rodríguez Peralta, L. M., Posada Barrera, A. I., & de Oliveira Nunes, É. (2025). Sick building syndrome and indoor air quality: Leveraging Kolmogorov-Arnold networks for predictive pollutant control. In Lecture Notes in Networks and Systems (pp. 412–421). Springer. https://doi.org/10.1007/978-3-031-93109-3_37
Liu, Z., Wang, Y., Vaidya, S., Ruehle, F., Halverson, J., Soljačić, M., Hou, T. Y., & Max Tegmark (2024). KAN: Kolmogorov-Arnold networks. https://doi.org/10.48550/arXiv.2404.19756
Mohammadshirazi, A., Nadafian, A., Monsefi, A. K., Rafiei, M. H., & Ramnath, R. (2023). Novel physics-based machine-learning models for indoor air quality approximations. https://doi.org/10.48550/arXiv.2308.01438
Nunes, É. de O., Posada Barrera, A. I., Rodríguez Peralta, L. M., Maia Sampaio, P. N., & Cuesta Astudillo, F. L. (2025). Fuzzy risk assessment system for indoor air quality and respiratory disease prevention. IAES International Journal of Artificial Intelligence, 14(5), 3693–3701. https://doi.org/10.11591/ijai.v14.i5.pp3693-3701
Park, Y. (2022). Concise logarithmic loss function for robust training of anomaly detection model. https://doi.org/10.48550/arXiv.2201.05748
Park, J., Seo, Y., & Cho, J. (2023). Unsupervised outlier detection for time-series data of indoor air quality using LSTM autoencoder with ensemble method. Journal of Big Data, 10(1). https://doi.org/10.1186/s40537-023-00746-z
Posada Barrera, A. I., Rodríguez Peralta, L. M., de Oliveira Nunes, É., Maia Sampaio, P. N., & Cuesta Astudillo, F. L. (2023). Improving public health policies with indoor air quality predictive models. In International Conference on Computational Science and Computational Intelligence (CSCI). IEEE. https://doi.org/10.1109/csci62032.2023.00040
Posada Barrera, A. I., Rodríguez Peralta, L. M., de Oliveira Nunes, É., Maia Sampaio, P. N., & Cuesta Astudillo, F. L. (2024a). Beyond one room: Comprehensive predictive analysis of CO₂ in indoor air quality. In IEEE Latin American Conference on Computational Intelligence (LA-CCI) (pp. 1–6). https://doi.org/10.1109/la-cci62337.2024.10814744
Posada Barrera, A. I., Rodríguez Peralta, L. M., de Oliveira Nunes, É., & Maia Sampaio, P. N. (2024b). Influence of indoor conditions on sick building syndrome: A data-driven investigation. In Information Technology and Systems (Lecture Notes in Networks and Systems, Vol. 932). Springer. https://doi.org/10.1007/978-3-031-54235-0_5
Romero-López, Y., Martínez-Cruz, A., González, R. Á., & Gálvez, A. M. S. (2025). Implementation of an IoT system with a security scheme to predict indoor CO₂ levels and mitigate COVID-19 using time series algorithms. Integration, 103, 102427. https://doi.org/10.1016/j.vlsi.2025.102427
Sotelo Gómez, F., Maia Sampaio, P. N., Rodríguez Peralta, L. M., Cuesta Astudillo, F. L., & Nunes, É. de O. (2025). Leveraging quantum machine learning for accurate indoor air quality forecasting and risk mitigation. In Lecture Notes in Networks and Systems (pp. 276–286). Springer. https://doi.org/10.1007/978-3-031-93106-2_24
Qi, J., Du, J., Siniscalchi, S. M., Ma, X., & Lee, C.-H. (2020). On mean absolute error for deep neural network-based regression. https://doi.org/10.48550/arXiv.2008.07281
Tofallis, C. (2015). A better measure of relative prediction accuracy for model selection and model estimation. Journal of the Operational Research Society, 66(8), 1352–1362. https://doi.org/10.1057/jors.2014.103
Wei, Y., Jang-Jaccard, J., Xu, W., Sabrina, F., Camtepe, S., & Boulic, M. (2023). LSTM-autoencoder-based anomaly detection for indoor air quality time-series data. IEEE Sensors Journal, 23(4), 3787–3800. https://doi.org/10.1109/jsen.2022.3230361
Willmott, C. J., & Matsuura, K. (2005). Advantages of the mean absolute error over the root mean square error in assessing model performance. Climate Research, 30, 79–82.
Yang, G., Yuan, E., & Wu, W. (2022). Predicting long-term CO₂ concentration in classrooms based on BO–EMD–LSTM model. Building and Environment, 224, 109568. https://doi.org/10.1016/j.buildenv.2022.109568
Zhu, Y., Al-Ahmed, S. A., Shakir, M. Z., & Olszewska, J. I. (2022). LSTM-based IoT-enabled CO₂ forecasting for indoor air quality monitoring. Electronics, 12(1), 107. https://doi.org/10.3390/electronics12010107
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 International Journal of Combinatorial Optimization Problems and Informatics
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
