LARRUZFounder | Author & Creator of BYOAI

AI For Prognosis (AI For Medicine)

9 Apr 2020 · 1 min

Watch on YouTube

1:10

Short excerpt from a YouTube lecture introducing a multi-week course on medical prognosis. Weeks cover: (1) definition of prognosis and routine risk calculations, (2) machine learning with decision trees for modeling nonlinear relationships and predicting mortality, (3) survival models for time-to-event outcomes (death, recurrence), and (4) strategies to build and evaluate survival models for comparing individual patient risks. The speakers emphasize using large medical datasets and machine learning to estimate individual risks (e.g., 10-year risk of heart attack).

Chapters

In this first week you learn what is prognosis and will see multiple examples of prognostic tasks including a few examples where prognosis using risk calculations is part of routine clinical practice. In the second week.
You will build machine learning models with decision trees. You will use trees to model nonlinear relationships which are commonly observed in medical data and apply them to the prognostic task of predicting mortality.
You learn about survival models. It allows you to model the time to an event, in this case the patient's possible death. Survival models are also used to model the time from treatment to recurrence. In.
Week four you will learn about strategies to build and evaluate survival models that allow you to compare the risks of individual patients.
As we collect larger and larger medical data sets, machine learning will become an invaluable tool to learn the complex relationships in medical data and to help us answer questions. Questions like why some people survive longer than others or what is the patient's 10 year risk of heart attack?
Let's dive.
In.