ICU Mortality/Event Visualization
This visualization tracks ICU patients’ predicted mortality risk over the course of their stay, with each dot representing a measurement taken at a given relative time. The y-axis shows the model’s predicted probability of death, and the x-axis shows relative time from admission. Each dot is colored by procedure category, and hovering over a dot reveals a tooltip with the patient ID, procedure type, relative time, and absolute time in hours. The chart includes a legend for procedure categories and uses an SVG line-and-dot layout with D3 v4. Animation is not present in the rendering despite the metadata listing it.
AI-generated descriptionUsing data from the MIMIC-III data set (data on ICU patients), I developed a mortality scoring model. Given the current measurements of a patient 24 hours after admission, the model will predict the chance of a patient living or dieing. When applied to the measurements taken over an entire ICU stay, a doctor or hospital administrator can review how close a patient is to dieing over the course of his or her ICU stay. In addition to this, I also overlaid the hospital procedures each patient has undergone, so doctors and administrators could see any possible effects that these procedures had as well. Finally, there's a mouseover tooltip that shows further details for each procedure the patient underwent.