Cardiogenic shock represents a high mortality, high-cost cohort of patients for which there are several clinical interventions available (i.e. mechanical circulatory support devices) with the potential to impact outcomes, however to date, have not translated in demonstrable benefits.
This is likely due to the recognition of cardiogenic shock at too late a stage, by which time the adverse pathophysiological cascade cannot be reversed. Furthermore, there is a lack of risk stratification tools for early detection of CS, and current risk scoring algorithms are designed to predict in hospital or 30-day mortality, not the development of CS.
By leveraging artificial intelligence (AI) techniques and routinely collected clinical data, algorithms which detect evolving cardiogenic shock at an earlier stage present a significant opportunity to support more rapid detection and clinical decision making, enabling earlier intervention and improved patient outcomes.
Such AI algorithms deployed across multiple centres would provide opportunity for consistent identification of cardiogenic shock, enabling the development of a multisite clinical trial platform to drive collaborative large-scale research efforts (e.g. registries and clinical trials) in cardiogenic shock, which to date has represented a significant challenge towards advancing the research evidence base. The key aims of our project are:
Develop digital phenotypes of acute heart failure and cardiogenic shock based on routinely collected clinical information from electronic health records, and use artificial intelligence techniques to develop algorithms that enable earlier recognition of cardiogenic shock.
Define the analytics-driven architecture to embed this digital phenotype within the Victorian Heart Hospital for clinical evaluation.
Last updated07 November 2025