What is the probability that you have had a stroke, based on several of your demographic variables?
This Shiny dashboard was created as supplemental material for a paper that was written for a biostatistics class presentation. The paper was on the subject of the effects of various variables on the probability of having had a stroke. The major findings of the paper were that smoking daily, depression, and a higher number of days marked by anxiety were associated with increased odds of having had a stroke. Neither the number of hours slept per night, nor being a heavy drinker were associated with stroke odds. However, sleep and drinking were associated with stroke odds when included as interaction variables.
This Shiny dashboard takes input from the user: sex, depression status, smoking status, age group, race/ethnicity, sleep hours and number of days heavily affected by anxiety. When the user changes any of the input variables, a logistic regression model is calculated and graphed to display their model (User Model) against a null model. This demonstrates whether the results of the User model is significantly different from a logistic regression model with no predictor variables. The graph output includes a probability point and 95% confidence intervals (CI). If the CI overlap, there is no evidence that the users’ results are different from that of the null model. If the CI do not overlap, there is significant evidence that the user’s model is different from the model with no predictor variables.