A practical and example filled tour of simple and multiple regression techniques (linear, logistic, and Cox PH) for estimation, adjustment and prediction.

Loading...

來自 Johns Hopkins University 的課程

Statistical Reasoning for Public Health 2: Regression Methods

81 個評分

A practical and example filled tour of simple and multiple regression techniques (linear, logistic, and Cox PH) for estimation, adjustment and prediction.

從本節課中

Module 3B: More Multiple Regression Methods

This set of lectures extends the techniques debuted in lecture set 3 to allow for multiple predictors of a time-to-event outcome using a single, multivariable regression model.

- John McGready, PhD, MSAssociate Scientist, Biostatistics

Bloomberg School of Public Health

Greetings. John here again, and

again here to talk about multiple logistic regression action.

This time, when we're actually regressing time to event data in the presence of

the censoring on predictors of interest.

So, we're going to use a multiple Cox regression approach,.

And, this section will show that the ideas of in terms of

interpreting the results from Cox regression.

As adjusted measures of association are very parallel to what we did with

logistic regression for odd ratios and linear regression for mean differences.

We'll also see that we can compare the results form unadjusted Cox regressions,

or simple Cox regressions, to the results from,

and that, a model includes multiple predictors.

To get a sense of whether relationships of interest were confounded by other factors.

We'll also see in this sections,

while it's not something we could easily do by hand.

We could certainly invoke a computer and

sometimes you'll see the results of such analyses done in papers where they use

the results from multiple Cox regression models.

To actually estimate survival curves of an event for

different subgroups of the population based on their X values.

So, for example separate curves for

males of a certain age with a certain condition compared to separate curves for

females of the same ages with a different condition, et cetera.

And, they can show the estimated survival curves, translate these

adjusted hazard ratios into what they mean in terms of the percent surviving or

remaining event free across the follow up period.

So, at the end of this free lecture trifecta here on using multiple regression

for an estimation adjustment, we've now have some sense of how to get adjusted

associations using multiple regressions and compare them to their unadjusted

associations from simple regressions to look at the degree of confounding.

And, also use multiple regression as a tool to be able to

better predict our outcome by using more than one predictor at a time.

In the next lecture section, lecture nine, we'll briefly talk about how each of

these regression approaches can be altered to allow for the investigation.

Both of effect modification and also how to estimate,

deviations from the linearity assumption without making our cat,

continuous predictor categorical by default.