<p>Drought necessitates sector-specific assessment and response in meteorology, agriculture, and hydrology. However, considering the multi-sectoral complexity of prolonged drought impacts, an integrated drought response is also required. The integrated drought index (IDI) provides an effective approach to support such comprehensive drought management. This study proposes an AI-based IDI estimation model using a GA-LSTM framework, which combines a genetic algorithm (GA) and long short-term memory (LSTM). In this framework, the GA performs iterative evolutionary optimization, while the LSTM estimates the IDI by learning both nonlinear and temporal dependencies among three sector-specific drought indices: meteorological (MDI), agricultural (ADI), and hydrological (HDI). To validate the IDI, a combined drought warning (CDW) was constructed by integrating historical sector-specific drought warnings, and its performance was assessed using receiver operating characteristic (ROC) analysis. Regional accuracy varied according to dominant drought types, with hydrological-dominated areas showing the highest consistency, while meteorological and agricultural drought regions also exhibited improved alignment. These results highlight the spatial adaptability and operational potential of the IDI for regional drought monitoring. The proposed approach provides a foundational methodology for advancing drought early warning systems and supports coordinated inter-agency management strategies.</p>