Decoding a Hidden Energy State Based on Marked Point Process Cortisol Secretory Events During Cardiac Surgery

Cortisol is critical in regulating one's energy state in response to stressful events such as surgical procedures. Decoding a cortisol-related energy state during surgery can assist in managing one's overall health status under inflammation. In this study, we decode a hidden cortisol-related energy state from each patient's cortisol profile during coronary arterial bypass grafting surgery. In particular, we employ a Bayesian state estimation approach within an expectation-maximization framework and estimate the energy state from the observation vector, which consists of the inferred cortisol secretory events coupled with a reconstructed high frequency cortisol profile. This reconstructed cortisol profile has a one-minute resolution and is obtained by using the estimates from deconvolution of cortisol data sampled at every 10 minutes. We find a higher energy state within the post-surgery phase compared to the surgery phase for all the studied patients (10 patients), which may depict the decoder's reliability in manifesting clinically relevant information. Tracking the person-specific cortisol-related energy state during surgery could provide insights into intervention design procedures and treatment plans.