Nonorthogonal multiple access (NOMA)-based multiple-input-multiple-output (MIMO), which has the potential to provide both massive connectivity and high spectrum efficiency, is considered as one of the efficient techniques for sixth-generation (6G) wireless systems. In massive Internet of Things (IoT) networks, the user-set selection is crucial for enhancing the overall performance of NOMA-based systems when compared with orthogonal multiple access (OMA) techniques. In this article, we propose a user-set selection algorithm for IoT uplink transmission to improve the sum data rate of the NOMA-based MIMO systems. In order to exchange data between the selected IoT pairs, we propose to employ wireless physical-layer network coding (PNC) to further improve the spectral efficiency and reduce the delay to fulfill the requirements of future IoT applications. Performance evaluations are provided based on both the sum data rate and bit error rate for the proposed NOMA-based MIMO with PNC in the considered massive IoT scenarios.