Use of Data Science for Population Health Policy Analysis– The Case of Opioid Crisis

Background: Health policy formulation in general has been based upon findings from the scientific literature, expert opinion and requests from the community. It is usually the case that patients, their relatives and caregivers directly know what is missing and what they actually may want in order to take care of the patients. It will be good for policy makers to include opinions and sentiments of community patients, caregivers, and family in the formulation of policy. However, opioid policies are developed using scientific research. The assumption is that policies that take into account the opinions of the patients are more likely to develop policies that improve the health of patients based on the convergence of sentiments of patients and care givers and evidence-based research are more likely to be adopted by patients and communities Introduction: This evidence based research aims at investigating whether there is any convergence based on scientific evidence from literature and opinions of the patient population for opioid crisis H_0 1: There is no convergence between policies formulated for opioid control based on scientific publications and expert opinions and the needs of the opioid community H_1 1: There is convergence between policies formulated for opioid control and the needs of the community. To conduct this research, I use data from several journals which are scientific literatures and social media data Methods: The data above is unstructured. Therefore, i used data science, machine learning, natural language processing to analyze the data. Results: Multinomial logistic regression model with ngrams (1,3) outperformed the other approaches while analyzing journal articles. Several features were obtained from journal articles . Similarly, several features were obtained from results of social media with some sentiment words. Features with the highest score (coefficient) were selected. The needs of the community are not fully met at this time. Some of the suggestions include new policies for /to cancer patients, other disease/disorders where the pain is high, school/college going kids, border security, pharmaceutical marketing, eliminate stigma, physician training