Unpublished Research Paper

Our project is the creation of a website which aims to help with stress and headaches by allowing participants to answer a questionnaire that asks information about the participants sleep quality, academic performance, study load, extracurricular activities and then gives a predicted value for the stress level and headache probability every week. To obtain the predictors, we used an online dataset and we used random forests to train the parameters to get the predictor based on the input data. In our website, there is a questionnaire that the participant fills out and, based on the trained model, we are able to provide the predicted value of stress level and the headache probability. Moreover, another objective of ours is to tailor the prediction for the individual users, and this is done by adding the inputs and true outputs, that the user would enter themselves, to the larger design matrix of the original online dataset and train on this combined dataset to provide a predictors of headache probability and stress that are more tailored for the individual. Together, this will enable the user to better manage the 4 explanatory variables, to optimize the response variables of headache probability and stress level. Even though we focus here on headache probability and stress level, in a more general sense, the algorithm can be used for any other set of response variables and explanatory variables, to get tailored results for the user that they can utilize to optimize different aspects of their mental health.