Figure 3. Hierarchical clustering of physiological measurements across chronological age.

The trends of physiological measurements across chronological age for all the 385,918 adult patients analyzed were derived from applying the locally weighted scatterplot smoothing (LOWESS) algorithm. Prior to performing hierarchical clustering, the trends for all measurements were z-score transformed. Hierarchical clustering was performed using Euclidean distance and average as the linkage function. Physiological measurements with similar trends across age are clustered together. The age range of patients used to produce these figures is 18 to 100 years.