Despite the use of current tumour-based biomarkers it remains challenging to predict whether a given individual will have sustained response to ICI treatment. The patient’s gut microbiome profile and many nutrition-related features provide additional prognostic information, but it is not clear how best to incorporate these data to improve ICI treatment planning. A combination of statistical modelling and machine-learning techniques, such as archetypal analysis, is proposed to identify patterns or combinations of features associated with better outcomes after ICI treatment. This pattern-recognition approach may also identify mechanistically important co-dependencies between nutritional and gut microbiome data and lay the groundwork for future intervention studies to improve outcomes.