With the rapid popularization of mobile smart devices in the IoT and the 5G environment, nodes’ requirements for network response speed are constantly increasing. Edge computing uses edge servers to perform simple processing when data is transmitted, increasing the response speed of devices and reducing the pressure on network traffic. However, the random movement of many nodes in an opportunistic social network easily leads to dynamic changes in the network structure and unstable transmission links. Therefore, this research proposes a data transmission strategy based on node motion prediction in opportunistic social networks (MPDTS). Any node will be assigned to a different cluster depending on how likely it is to meet other nodes. Messages are forwarded within and between clusters using different judgment indicators. This method effectively combines the connection between nodes, the node’s activity level, and sports characteristics successfully reduces the waste of resources for invalid message delivery. Simultaneously, it improves the possibility of message forwarding to the target node. Comparative experiments with several methods show that the MPDTS method has more outstanding performance.