Adaptive Modulation and Coding (AMC) technology plays a critical role in maintaining the transmission quality in 4G LTE (Long Term Evolution) and 5G NR (New Radio) wireless communication systems, to overcome the dynamic fluctuations of radio channel. The performance of AMC may deteriorate significantly if the outdated radio channel information is used for selecting the most suitable modulation and coding scheme. In this paper, an intelligent AMC method based on resource block (RB) group level uplink interference prediction is proposed. By taking the correlation relationships of the RBs into consideration, the AI/ML prediction model are derived in RB group level. It assures good prediction accuracy and also helps to shorten the inference latency. The prediction accuracy and network performance of the proposed solution are verified under a variety of scenarios with varied traffic patterns, network loads and mobility settings of interference users, demonstrating its superiority over existing solutions and applicable scenarios.