Clinical decision support for chronic cardiovascular diseases using high resolution health records

Abstract

Precision medicine involves the delivery of personalized care. In the cardiovascular area,the characterization of the heart to distinguish abnormal from normal behaviour is of deepimportance, especially when dealing with chronic patients.The aim of this thesis is to contribute to the solution of two problems in the medicalcommunity.The first project, the PLAR Study, aims to validate and apply a new visualization toolfor electrocardiography. First, the tool has been validated on a population of patientswith Familial Hypertrophic Cardiomyophathy from the Brigham and Women’s Hospital inBoston. Half of the patients present a mutation in the sarcomere and the other half donot. Through visualization and data analysis, the tool has been proven to show patientevolution over time. Moreover, changes from one year to another have been identified andrelated to physiological changes, seen in echocardiographic data. Differences between typesof patients have not been found in the analysis, which might imply that the mutation inthe sarcomere does not have a direct implication in ECGs.The second project, the MAP Study, focuses on detecting patient-specific heart ratethresholds using a lumped-parameter heart model, CVSim, developed at MIT. Eightmeasurable physiologic parameters have been defined to cover a wide range of cardiovascularconditions. The results identify a relationship between heart rate and mean arterial pressurefor specific patients, which can be used to predict cliffs of risk.The final results of the thesis are a first step towards a clinical decision support tool,thus open a window of opportunity for future projects