Posts

Net Benefit and Decision Curve Analysis in R - Some helpful functions
Background One of the challenges of developing clinical prediction models is deciding which one is best. Sometimes traditional performance measures such as the Brier score or C-statistic might be used to quantify to what extent one model performs better than another.
Last updated on Nov 30, 2022 9 min read r, gmish
Using MetaMap with Python to access the UMLS MetaThesaurus - A Quick Start Guide
The National Library of Medicine (NLM) created a tremendous resource called the Unified Medical Language Systems (UMLS). The UMLS is a suite of data and software resources that offers several tools for processing, representing, and analyzing biomedical text data.
Last updated on Feb 25, 2022 4 min read python, nlp
Comparison of bootstrapping approaches for identifying variance of predictive performance
Background Bootstrapping is a helpful technique for identifying the variance of an estimate in a given sample when no other data are available. In the case of the evaluation of clinical prediction models, bootstrapping can be used to estimate confidence intervals around performance metrics such as the Brier score, the c-statistic, and others.
Apr 25, 2019 4 min read blog, r, bootstrapping, predictive modeling
Evaluating the equivalence of different formulations of the scaled Brier score
Background The Brier Score is a composite measure of discrimination and calibration for a prediction model. The Brier Score is defined as \[ BS = \frac{1}{N} \sum (y_i - \hat y_i)^2, \]
Apr 23, 2019 4 min read brier score, predictive modeling
Building a recurrent neural network to predict time-series data with Keras in Python
Recurrent neural networks and their variants are helpful for extracting information from time series. Here’s an example using sample data to get up and running. I found the following other websites helpful in reading up on code examples:
Feb 15, 2018 2 min read keras, rnn, python
Making choropleth maps by zip code
Every now and then it is useful to make a map. In times of political uncertainty, data can light a path forward. Your local elected officals may be interested in data, too, and how they impact policy (and re-election, of course).
Jul 17, 2017 2 min read r, maps
Grey’s Anatomy Network of Sexual Relations
Here’s the blog post originally posted on babelgraph.org on March 25, 2011. The relevant hyperlinks and igraph code have been updated to reflect the current state of things. The edgelist, despite many more intervening seasons, and thus many more romances, has not been updated.
Jun 15, 2017 4 min read babelgraph, blog, social network analysis
The structure of twitter participant relationships in conversations around #Libya, #Bieber, and #Rstats
Here’s the blog post originally posted on babelgraph.org on April 4, 2011. I suspect the twitter API and much of the code below are not up to date. I am a recent comer to twitter, and it took me a few weeks to figure out what this was all about.
Jun 15, 2017 5 min read twitter, r, babelgraph, blog
Transportation data in R: Why yes, yes you can
Not infrequently I am in a research meeting and someone says, “It would be really cool to get data on travel times for people. But I don’t know where to find that.
Jun 15, 2017 3 min read r, maps
Welcome to the blog!
Welcome to this blog, which is the new home for thoughts and technical things. As babelgraph.org has been retired, you can still get the code at: BabelGraph Alpha Over time I will try to migrate the old blog posts (mostly dealing with network analyses and C++ using R) to here along with the very informative user comments, and updates.
Jun 15, 2017 1 min read babelgraph, blog, meta