Oriented graph of possible patient journeys. Example of a graph that models the data collected for a cluster of journeys, according to their specific initial features (condition on admission), I. The expected (or positive) result of this path is O, which can be reached with probability pO. The patient's risk is defined by the probability of reaching different final conditions given the same starting condition upon admission. The risk of obtaining a final condition different from O, starting from condition I on admission, is given by the list of probabilities [p (O1), p (O2), p (O3), p (O4)]. Each node of the graph represents a possible step (initial, intermediate, or final) of the journey (and thus a stage of the patient's condition between admission and discharge), and each edge (or line between nodes) represents a possible succession of steps/stages according to the probability of encountering each of them. Such journeys make up our knowledge base. In particular, we can calculate the probability of arriving at the final step/stage, given any initial node of the graph, be it initial or intermediate.This model defines a measure of health risk as the probability of reaching a different final step/stage, Oi, compared to the expected one O. At the time of the patient's admission (when we define the initial condition of his real journey), we identify the cluster to which the patient belongs. Then we interrogate the associated graph to determine what the expected risk is, and the probability that this patient will find himself at the end of his intra-hospital journey in a final condition different from the expected one upon admission. By using this information, physicians can make informed choices that help them implement clinical protocols and best practices according to real patients' situations., In particular, this model makes use of a data-driven approach to improve patient outcomes and organizational efficiency by putting the patient and their intra-hospital journey at the center., The transition to data-driven decision-making in healthcare is largely driven by the need to improve patient outcomes. Traditional methods, which relied heavily on physician experience and intuition, are now being augmented with data-backed insights that provide a more comprehensive understanding of patient health..