Identification of Specific Human Behaviours from Cortisol profiles using Bagged and Boosted Decision Trees

Cortisol, a glucocorticoid hormone, plays an essential role in regulating many functions in the body, like blood pressure, stress levels, immunity, metabolism, etc. Although serum concentrations of some hormones like cortisol, are impacted by daily activities, the correlation has not been widely studied or documented. Understanding this critical correlation will help design better and more customized healthcare plans tuned to a patient’s unique physiological needs. In this preliminary study, we used decision tree models that can analyze a 48-hour cortisol concentration profile and identify if the patient engaged in certain activities in the same period. The goal is to enable patients to track the impact of their lifestyle/habits/activities on cortisol concentrations, and regulate it by modulating or varying these activities. We were able to analyze the cortisol levels and successfully identify if the patient had consumed caffeine, nicotine, anti-depressants or engaged in exercise with a best accuracy of 73%, 67%, 77% and 74%, respectively. The results lay the foundation of establishing predictive correlations between cortisol and daily activities, and this information would assist in self-regulating cortisol concentrations and restrict it within a healthy range. This preliminary study is intended to encourage more in-depth assessments of cortisol and other hormones, leading to thorough health monitoring, early diagnosis or predictions of diseases and preventative healthcare.