The volume of medical data is expanding rapidly along with its variety, which gives rise to the Internet of Medical Things (IoMT) concept. It is well known that clustering is an important procedure to mine the vast data in the medical industry. Many swarm intelligence clustering algorithms that can be parallelized to the benefit of mass data computation have emerged recently. However, few studies focus on the systematic analysis of the time complexities. For example, it is important to analyze the effect of instances (data size), attributes (dimensionality), number of clusters, and agents of these algorithms. In this paper, we do a performance comparison for the PSO, Firefly, Cuckoo and Bat algorithms based on both synthetic and real medical data sets. Finally, we conclude which algorithms are effective for the medical data mining. In addition, we gave recommendation for the more suitable algorithms that can be better applied to medical data for achieving optimal clustering. Our work has a significant contribution to implementing accurate and efficient analysis of big data based on the IoMT platform in the healthcare industry.