M.S. AAI Capstone Chronicles 2024

requiring rapid response, and influence future research on sepsis and other time-sensitive illnesses. Data Summary The dataset consisted of time-series samples of clinical and laboratory readings grouped to a unique patient identifier. There were 1,552,210 total samples related to 40,336 unique patients (Reyna et al., 2019). Each sample provided 40 explanatory variables, with 8 continuous (float) variables related to clinical readings of the patient’s vital signs at the time step, 26 continuous (float) variables corresponding to readings obtained from laboratory samples and 6 variables that capture the patient’s demographic characteristics, including age and gender. Each sample was captured at an hourly time step and included a target binary variable that indicates whether the patient developed sepsis at the time step or not. Figure 1 Sample patient time series readings of vital signs and some laboratory variables

266

Made with FlippingBook - professional solution for displaying marketing and sales documents online