Population physiology: leveraging electronic health record data to understand human endocrine dynamics

Abstract

Studying physiology and pathophysiology over a broad population for long periods of time is difficult primarily because collecting human physiologic data can be intrusive, dangerous, and expensive. One solution is to use data that have been collected for a different purpose. Electronic health record (EHR) data promise to support the development and testing of mechanistic physiologic models on diverse populations and allow correlation with clinical outcomes, but limitations in the data have thus far thwarted such use. For example, using uncontrolled population-scale EHR data to verify the outcome of time dependent behavior of mechanistic, constructive models can be difficult because: (i) aggregation of the population can obscure or generate a signal, (ii) there is often no control population with a well understood health state, and (iii) diversity in how the population is measured can make the data difficult to fit into conventional analysis techniques. This paper shows that it is possible to use EHR data to test a physiological model for a population and over long time scales. Specifically, a methodology is developed and demonstrated for testing a mechanistic, time-dependent, physiological model of serum glucose dynamics with uncontrolled, population-scale, physiological patient data extracted from an EHR repository. It is shown that there is no observable daily variation the normalized mean glucose for any EHR subpopulations. In contrast, a derived value, daily variation in nonlinear correlation quantified by the time-delayed mutual information (TDMI), did reveal the intuitively expected diurnal variation in glucose levels amongst a random population of humans. Moreover, in a population of continuously (tube) fed patients, there was no observable TDMI-based diurnal signal. These TDMI-based signals, via a glucose insulin model, were then connected with human feeding patterns. In particular, a constructive physiological model was shown to correctly predict the difference between the general uncontrolled population and a subpopulation whose feeding was controlled.

Figure 1. Depicted above are (a) the glucose for the standard glucose-insulin model with continuous feeding; and (b) the glucose for the standard glucose-insulin model with realistic meal structure.

(a) Glucose-insulin model with continuous feeding and glucose response. (b) Glucose-insulin model with three meals and glucose response.

Figure 2. Depicted above are (a) the mean and standard deviation in glucose, by hour, for patients whom have been normalized to mean zero and variance one, with at least two glucose measurements from the CUMC EHR; (b) the two individual patients mean and standard deviation in glucose measurements by hour, note the variability in patient for which there are far fewer measurements than for patient ; (c) the mean and standard deviation in glucose and enteral (i.e., tube) feeding rates, by hour, for normalized patients in the neural ICU; (d) glucose, by hour, for various different model feeding patterns.

(a) Normalized population glucose by hour. (b) Single patient normalized glucose by hour. (c) Normalized NICU population glucose and feeding by hour. (d) Normalized model glucose by hour.

Figure 3. Depicted above are (a) the TDMI curves for all EHR-data based populations and model output for all feeding patterns resolved to one hour intervals for time delays of up to one week, note the sharp decay in TDMI in all cases, and the diurnal peaks in all periodically fed populations or models — note this plot is split into dynamical regimes in Figs. 2 and 2; (b) the TDMI curves for all populations and models over time-delays of to hours; and (c) the TDMI curves for all populations and models from to hours, notice the diurnal peaks in all periodically fed populations or models. Recall that the model feeding patterns are given by: — continuously fed population; — continuously fed population with random hour gaps; — periodically fed individual; — noisy-periodically fed individual; and — a randomly fed individual.

(a) All data sets and models — a global view of the TDMI. (b) All data sets and models — feeding scale TDMI for of to hours. (c) All data sets and models — diurnal scale TDMI for of to hours.