In multi-hop wireless networking, a hop count is a total number of intermediate nodes, especially router devices, through which data travels from source to destination. Hop count prediction is crucial for resource-constrained exigent data delivery applications like healthcare services with limited network lifetime or high processing costs. Many schemes are proposed earlier to estimate hop count for the dense or sparse networks. This paper uses Support Vector Regression (SVR) to predict hop count in an intermittently connected networking environment where end-to-end connectivity is not inevitable. A novel data forwarding mechanism is proposed for IoH aware framework. Finally, the network performance of the proposed scheme is compared with two other popular data transmission mechanisms in the same scenario, i.e., PRoPHET and PRoPHET+, for evaluation purposes. Since the SVR mechanism is computationally cheaper, shows better generalization capability, and depends only on a subset of data points (support vectors) for decision function, it is believed that this method will perform well for estimating hop count analysis and selecting minimum hop count for data delivery. The multi-feature training with dataset and predicting hop count techniques in machine learning can significantly reduce the hop count from source to destination and perform well in terms of overhead ratio and delivery probability with acceptable latency.