High utility itemset mining (HUIM) is a powerful data mining technique that extracts information from transactional datasets. This paper proposes a new approach to mining high utility itemsets (HUIs), based on Ant Colony Optimization (ACO), which uses Roaring Bitmaps as the core data structure for managing transaction sets and Remaining Transaction Weighted Utility (RTWU) for negative utility support. The algorithm works with the help of iterative ants, which construct HUIs by selecting items probabilistically. The ants are guided by pheromones and heuristic information, while leveraging EUCS for effective pruning and Roaring Bitmaps for fast utility calculations. The overall architecture of this algorithm preserves the constructive nature of ACO but significantly enhances the internal mechanics of each phase. In each iteration, multiple ants select items randomly guided by probabilistic information and then construct a high utility itemset by selecting those itemsets that are likely to lead to the highest utility itemset. In addition, the algorithm employs the Estimated Utility Co-occurrence Pruning strategy (EUCS) to prune all paths that couldn’t potentially lead to HUIs. The experimental results are compared with five recent state-of-the-art algorithms.