THE USE OF COMPLEX CLINICAL DATA AND TOPOLOGICAL DATA ANALYSIS FOR PERSONALIZED MEDICINE

摘要

Methodologies that could identify subgroups of patients that may or may not respond to a given treatment could be a revolutionary tool in personalized medicine, a new concept for treating a specific patient based on their particular health or physiology. The association between obesity and several of its comorbidities, including diabetes, hypertension, dyslipidemia, stroke, and cardiovascular disease, is well established. However, variability from patient to patient complicates the translation of these risk factors to the clinic to give actionable information about a patient's optimal treatment. Based on 28 physiological variables, we analyzed a cohort of 2700 patients from the Genetic Epidemiology of Network of Arteriopathy (GENOA) Study using topological data analysis (TDA), a new clustering algorithm tool. Variables used for the analysis included blood pressure, BMI, age, renal function, and metabolic markers. TDA clustered and separated out 6 distinct subgroups of obese patients with similar BMI but differed in over 100 variables including renal disease, serum inflammatory biomarkers, and prevalence of stroke, diabetes, and hypertension. This suggests that the association between obesity and its comorbid conditions is not always clear. These methodologies could potentially be used to discover patterns in a patient's physiology and advance personalized medicine.