A cast/brace is a tight garment that restricts the movement and provides support to an injured zone. Traditional casts/braces suffer from material wastage, discomfort, patient dissatisfaction, odor, unnecessary weight, and dangerous extraction procedures. These issues can be solved partially by constructing the casts/braces via 3D printing. Toward this end, we print the personalized metacarpal casts/braces (MCB) via fused deposition modeling (FDM), and investigate their mechanical properties to ensure the desired functionality. However, printing the full-size MCB is time-consuming (takes more than 11 hours in our design), making it hard to collect a sufficient data set for the mechanical properties investigation. Here, we explore the utilization of reduced-size MCB to facilitate the analysis of full-size MCB via transfer learning. In particular, three critical process variables (i.e., raster width, layer height, and extrusion temperature) were varied, and a universal testing machine was used to measure the total deformation of the MCB. We then perform the prediction of the deformation in full-size MCB with transfer learning of data from reduced-size MCB and limited data from full-size MCB. From the case study, the transfer learning approach can reduce the needs of data collection in the time-consuming full-size MCB by leveraging the information from reduced-size MCB.