With human-computer collaboration on the rise, there is a steady need for artificial intelligence (AI) models to assist humans in decision-making. The majority of decision support systems (DSS) account for the technical aspect of decision-making, while few include system and environmental features. The inclusion of user (human) and environmental contexts into the decision-making process aims to enhance the overall recommendations of decision-support models by incorporating perceptional states that represent the perspectives of the decision-maker. Yet, the inclusion of excess information during modelling may give rise to ambiguity in recommendations. This work presents an experiment that explores the impact of incorporating internal and external contexts in the decision-making process to analyse the distinction of overall decision policy recommendations. Using clinical decision-support systems’ data (CDSS) of patients’ risk of acquiring type 2 diabetes mellitus (T2DM) evaluated by medical professionals, we derive and model the relative perceptional states of medical professionals during decision evaluation via inverse reinforcement learning (IRL). Furthermore, we investigate the individual impact of having internal and external context (i.e., doctors’ characteristics and patients’ risk of developing T2DM, respectively) in three strategic scenarios, revealing the changes in reward functions and differences in final outcomes relative to human decision-makers’ characteristics. The results present a step toward optimizing human-AI interaction to improve decision-making collaboration.