<p>Accurate prediction of traffic flow is fundamental for effective urban traffic guidance and control, playing a vital role in intelligent traffic management. However, achieving precise traffic flow prediction remains a challenging task due to the intricate spatiotemporal dependencies involved. To address this issue and capture the dynamic spatiotemporal characteristics of traffic flow concurrently, this paper introduces a pioneering approach for traffic flow prediction: the spatiotemporal interactive adaptive hybrid graph convolutional networks (STIAHGCN). First, an interactive learning structure is devised to dynamically aggregate the spatiotemporal features of hidden nodes within the traffic network. Second, a gated temporal convolutional network is constructed, employing dilated causal convolutional networks at multiple granularity levels to capture the temporal dependencies present in traffic flow. Moreover, the adaptive hybrid graph convolution module, comprising static adaptive graph learning, dynamic graph learning, and a spatial gate fusion mechanism, operates synchronously to effectively capture the dynamic spatiotemporal features within historical traffic flow data. The experimental results substantiate that the proposed STIAHGCN successfully extracts dynamic spatiotemporal features of traffic flow, delivering superior prediction performance compared to prevalent baseline methodologies.</p>