Predict Likelihood of Completion for Future Lifestyle Medicine Program
Project Description :
The United States is in a dubious position: it spends more on per capita health care than any other country, has the highest burden of adults with chronic disease, and ranks first in avoidable deaths. Ninety percent of its annual healthcare dollars are spent in treatment of chronic diseases and their complications. Proactive disease prevention and wellness is essential to reversing these healthcare trends and to improving the health outcomes of our communities. This project will be based on the future Lifestyle Medicine Program at one of the largest integrated health systems in the Midwest. Lifestyle Medicine is a multidisciplinary approach to Health and Wellness that enables individuals to cardinally change their lifestyle to improve or even eliminate a chronic condition and enhance their wellbeing long-term.
Lifestyle Medicine targets specific cohorts of patient population: individuals with the current history of diabetes mellitus, hyperlipidemia, hypertension, BMI of 25-40, and prediabetes. In addition to the data on medical conditions, this project will use simulated data of psycho-demographic variables known to play a role in a) likelihood for self-selection into comparable programs; b) likelihood for completion of the program in its entirety. The project will rely on the ample published literature and structured interviews with subject matter experts such as physicians certified in Lifestyle Medicine.
Project Goals • Gain understanding of Lifestyle Medicine and how it works to prevent or reverse target conditions. • Study relevant literature to create a set of psycho-demographic variables that can serve as reliable predictors for self-selection for and completion of the Program. • Conduct interviews with SMEs to validate simulated data. • Create simulated data on the basis of the literature review and SME interviews. • Perform data cleaning, wrangling, and feature engineering on the dataset. • Complete exploratory data analysis and data visualization on the dataset . • Implement unsupervised? machine learning models to predict self-selection and completion. • Evaluate and compare the machine learning models using quantitative measures: accuracy, precision, recall, area under the curve (AUC). • Make conclusions about how the model can be used in the future Lifestyle Medicine Program to target individuals for enrollment.