Biomedical time-series data from biosensors, electronic noses, or other medical devices is often limited as collecting it is expensive and can cause patient stress.

While AI/ML models are well known for time-series data from other domains such as activity tracking, not all developments can directly be transferred to biomedical data.

Here we will develop novel, more data-efficient methods with stronger regularisation/priors that incorporate domain expert knowledge to develop new models which can derive accurate and reliable predictions in a small-data regime.