David Clifton and co-workers have recently published a paper that trained a novel machine learning model (DeepAMR) on a subset of the dataset collected by the CRyPTIC consortium.
DeepAMR for predicting co-occurrent resistance of Mycobacterium tuberculosis
Bioinformatics, 35(18), 2019, 3240–3249
The paper, which is open access and therefore free to read, can be found here.