Latent similarity is a machine learning tool used to make predictions from inter-subject similarities. It is extremely quick to train and performs better than other models at small sample sizes.
The paper has been accepted to IEEE TBME! Here is the final pre-print version.
I made a presentation describing LatSim and another near-term research project.
Our pre-print has been posted on TechRxiv.
The LatSim code is now available on GitHub; we are working on a PIP package.
We have completed part of the tool as the ImageNomeR project at the UAB Multiomics Hackathon