<p>Due to anthropogenic activities, the increase in air pollution and its varying spatio-temporal patterns have become one of the most severe atmospheric challenges for developing countries. Consequently, people face numerous health-related issues and environmental degradation. Therefore, accurate monitoring and forecasting of air pollution characteristics is essential to develop effective policies and mitigation strategies. For effective monitoring, it is crucial to characterize air pollution attributes using a probabilistic and standardized framework, allowing reliable regional and cross-country comparisons. However, many existing air quality indices lack these statistical aspects, which limits their ability to provide accurate and consistent pollution assessments. In this study, we introduce a novel air pollution index, the Standardized PM2.5 Concentration Index (SPM2.5I), developed by standardizing the cumulative distribution function of K-Component Gaussian Mixture models (KCGMD). The application of the proposed index is demonstrated across seven regions of Pakistan using classical forecasting methods, machine learning models, and steady-state probabilities of Markov chain. To validate the suitability of KCGMD, we compared its Bayesian Information Criterion (BIC) with 32 univariate probability models. The results show that across all stations, KCGMD consistently outperforms univariate distributions, making it the most appropriate choice for capturing the complex and multimodal nature of PM2.5 data. Furthermore, we evaluated the usability of both classical forecasting and machine learning models in predicting the time series data of the SPM2.5I. The results indicate that ARIMA consistently outperforms machine learning models, achieving the lowest RMSE of 0.52 and MAE of 0.40 in Lahore, which is the most polluted station among all regions. The model also maintains strong predictive accuracy across the remaining stations, confirming its reliability for air pollution forecasting. In addition, steady-state probabilities reveal that in Lahore, the probability of experiencing unhealthy and very unhealthy air quality conditions exceeds 68%, emphasizing the severity and persistence of pollution in the region. Similar trends are observed in other urban areas, reinforcing the importance of effective monitoring and forecasting. In general, the findings of this research demonstrate that the proposed SPM2.5I can be used effectively to monitor and forecast the characteristics of air pollution.</p>