126 Big data on early identification of patients with palliative care needs: barriers and opportunities

Introduction The aim of this study was to explore the views of experts on the use of big data (BD) advanced analytics (i.e: machine learning, deep learning or artificial intelligence techniques) on the identification of frail older patients with non-malignant diseases who could benefit from early palliative care (PC). Methods This descriptive study corresponds to the first round of a Delphi study currently under performance. Participants were asked through a questionnaire survey about the level of agreement regarding i) the use of advanced analytics on specific clinical scenarios and ii) the critical aspects when using these applications on the frame of PC. The sample included professionals from different countries and disciplines with expertise on chronicity, geriatrics and/or PC. Consensus was defined as >70% of subjects agreement. Results At this time, fifteen experts have answered the questionnaire (37.5% of whom invited to participate). 78.6% were women, 71.4% were working in a clinical setting >15 years as physicians (57.1%), nurses (35.7%) and psychologists (7.1%) 80% of them considered that the strategy based on an automatic tool (BD based) combined with front-line healthcare staff is the best way for identification of patients who could benefit from early PC approach. They assessed as ‘very’ and ‘extremely useful’ the use of BD models on research applications, both as a population health management tool (clinical clustering) and as a tool to improve the prediction of an outcome risk. However, the need for clinical validation and the dearth of evidence of practical benefits are the main critical aspects on the implementation of these tools. Conclusions The view of experts can contribute to guide BD applications on advanced stages of illnesses. The consensus about the opportunities and gaps on the implementation of these tools will support clinicians in decision making processes.