Resources for learning Machine Learning to create clinical predictive models.

Question

[egascon]

Hello

I am writing to ask if you know of any resources and tutorials to learn how to create predictive models to validate genes as potential biomarkers and make survival predictions.

I have been reading several papers and many of them apply techniques such as the Lasso algorithm, SMV, Random Forest, ROC curves... etc. I understand the theory but I am struggling to find a step-by-step guide to perform these analyses.

An example workflow that I would really like to learn is the following: enter link description here

What learning resources do you recommend? (Besides the Biostars book, which I want to buy as well)

Thank you very much for your help.

Answer 1

[Jeremy]

There are a couple of statistics books that use R that might help you. "Statistics: An Introduction Using R" by Crawley has a short chapter on death and failure data. "An Introduction to Statistical Learning with Applications in R" by James et al. has a more detailed chapter on survival analysis and censored data. (It looks like there's also a Python version of this book available now.) The downside is that each book costs about $30 (at least on Amazon).

You could also look for notebooks on biomarkers and/or survival analysis on Kaggle. I would pay special attention to notebooks made by masters and grandmasters (with a red or yellow circle around their profile picture). For example, here's a Kaggle Python notebook on pancreatic cancer that you might find useful:

Predicting Pancreatic Cancer

You can also check out Kevin Blighe's Biostars tutorial here:

Survival Analysis