Tabriz University of Medical Sciences Depiction of Health 2008-9058 17 3 2026 09 30 Providing an Artificial Intelligence-Based Model for Predicting the Quantity of Municipal Solid Waste in Tabriz City 243 259 10.34172/doh.2026.20 EN Masoumeh Piri https://orcid.org/0009-0007-3368-2369 Saeed Piri https://orcid.org/0009-0001-8800-8191 Hassan Taghipour https://orcid.org/0000-0003-1557-5563 Mohammad Hassan Sarbazan https://orcid.org/0009-0008-3272-830X Mohammad Shakerkhatibi https://orcid.org/0000-0001-7767-1425 Journal Article 10.34172/doh.2026.20 2025 12 08 2026 09 20 Background. Waste management is essential, for improving urban services. Predicting the quantity of waste generated can play an important role in designing and optimizing the collection, transportation, and disposal processes. In this regard, machine learning techniques have been introduced as powerful tools for predicting the quantities of generated waste. Therefore, this study was conducted with the aim of predicting the quantity of municipal solid waste (MSW) using neural networks in Tabriz city, Iran. Methods. To predict the quantity of MSW, daily waste generation data from the Tabriz Municipality Waste Management Organization were used over a six-year period (2019–2024). The neural network employed consisted of an 8-layer LSTM network utilizing the Adam optimizer. The available data were divided into training and testing sets at an 85:15 ratio, and after the model training process, the predicted results were compared with the actual data to assess model accuracy. Root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) were used to validate the proposed model. Results. The results indicated good accuracy of the designed model in predicting the trend and quantity of waste generation; specifically, the RMSE, MAE, and MAPE values for the training and testing data were calculated as 50.62 and 70.47 tons/day, 32.61 and 53.02 tons/day, and 3.52% and 5.86%, respectively. Conclusion. The analysis of the findings indicated the model's ability to reduce prediction deviation and increase reliability in estimating actual values. The low MAE values indicate that, on average, the model's prediction error compared to actual data was acceptable, and the model performed well in predicting the overall trend of waste generation. On the other hand, RMSE, which assigns greater weight to large errors, was also within a reasonable range, indicating that the model demonstrated good robustness against sudden fluctuations and outlier data. These results suggest that the trained model is capable of identifying hidden patterns in the time-series data and can be used as a suitable tool for data-driven prediction and management of municipal solid waste. Research Insights · Accurate prediction of municipal solid waste quantity, considering temporal fluctuations in waste generation, is essential for planning collection, processing, recycling, and disposal capacities and, ultimately, for improving sustainable waste management. · Conventional and traditional prediction methods may have limitations in accurately representing the complex patterns, temporal fluctuations, and nonlinear relationships involved in waste generation, which may reduce their ability to accurately predict waste quantities. This limitation highlights the need for more advanced, AI-based approaches for forecasting waste generation time series. However, the use of neural networks for predicting waste quantity involves challenges, including the model’s dependence on the accuracy and validity of the data, limited access to sufficient data, and difficulty in determining an optimal network architecture. Therefore, further evaluation of neural network–based models for predicting waste quantity using real-world data is needed. · In this study, an LSTM neural network was used to predict municipal solid waste quantity. Using daily municipal solid waste generation data from Tabriz, Iran, for the period 2019–2024, the LSTM model was trained to learn patterns in the data and predict waste quantity, and its performance was evaluated using RMSE, MAE, and MAPE. · The findings showed that the trained model was capable of identifying complex and hidden patterns in waste data and could be used as a data-driven decision-support tool for estimating future waste quantities. These findings add to the existing evidence on the application of AI-based approaches for predicting municipal solid waste quantity, particularly using real-world data from Tabriz, and may provide a basis for future studies using longer time periods and larger datasets.