Gaps in time series data pose significant challenges for deep learning research, impacting the performance of predictive models. To address these challenges, this chapter proposes a novel hybrid model that combines Conv1D layers, inspired by the inception architecture, with long short-term memory (LSTM) units. This hybrid model was rigorously tested on datasets with extensive gaps and dense data points related to air quality at Tien Giang University. The results demonstrate the model’s superior ability to process low-quality time series data without the need for manual adjustments specific to each dataset. Additionally, the authors developed a comprehensive solution for air quality data management and predictive analysis. This solution includes a data management module for the efficient storage, processing, and organization of air quality data, as well as a deep-learning-based time series predictive model to estimate future air quality parameters using historical data. The performance of the hybrid model markedly exceeds that of conventional LSTM models, delivering higher accuracy with fewer parameters (Hybrid: 85,024; LSTM: 5,257,728).