Twitter is a popular social network for people to share views or opinions on various topics. Many people search for health topics through Twitter; thus, obtaining a vast amount of social health data from Twitter is possible. Topic models are widely used for social health-care data clustering. These models require prior knowledge about the clustering tendency. Determining the number of clusters of given social health data is known as the health cluster tendency. Visual techniques, including visual assessment of the cluster tendency, cosine-based, and multiviewpoint-based cosine similarity features VAT (MVCS-VAT), are used to identify social health cluster tendencies. The recent MVCS-VAT technique is superior to others; however, it is the most expensive technique for big social health data cluster assessment. Thus, this paper aims to enhance the work of the MVCS-VAT using a sampling technique to address the big social health data assessment problem. Experimental is conducted on different health datasets for demonstrating an efficiency of proposed work. Accuracy of social health data clustering is improved at a rate of 5 to 10% in the proposed S-MVCS-VAT when compared to MVCS-VAT. From obtained results, it also proved that the proposed S-MVCS-VAT is a faster and memory efficient for discovering social health data clusters.