Classifying malaria using multidimensional space maps (P3108)

Abstract

If the path to sickness differs from the path to recovery, we should treat each route differently. Current treatments that inhibit pathogen growth might not be useful when pathogens are already being cleared. To determine the trajectory of both disease and recovery we created a multidimensional map of a malaria model infection in mice, tracing the route from initial infection through recovery. We did this using physiological parameters like red blood cell counts, temperature, weight and blood sugar and included molecular markers from blood transcriptome that was serially collected daily. We then plotted transcript levels versus parasite load and anemia to create a map for the full course of the infection. We can now use this map, constructed from longitudinal data, to scaffold data obtained in human cross sectional studies where we do not know the point of the patient in the infection timeline. This reveals previously unseen structure to these data and should provoke a re-examination of past conclusions. As we add data from non-recovering mice we will build maps that predict outcomes. Our goal is to not simply use these maps as a way of identifying biomarkers but to measure how treatments warp this multidimensional disease space.