Deep learning with 4D spatio-temporal data representations for OCT-based force estimation

•We propose 4D deep learning with streams of volumes for OCT-based force estimation.•Efficient decoupling of spatial and temporal processing is effective in 4D models.•4D models outperform previous methods with 3D spatio-temporal data.•High-dimensional spatial image encoding improves deep learning performance.•A temporal history improves performance and allows for short-term force prediction.