US CSS Hands-on Workshops
Please note the following Hands-on Workshops are subject to change. Registration for the Hands-on Workshops will open in April.
Safety Analytics Workshop
Mary Nilsson, Eli Lilly
The Analysis and Displays White Papers project team has developed six white papers outlining recommendations for analyses and displays for clinical trial data common across therapeutic areas (e.g. adverse events, labs, vital signs, concomitant medications). These white papers cover recommendations for individual study and integrated analyses. While the white papers include a rationale for the recommended analyses and displays across various choices, some topics would benefit from additional cross-functional education on safety analytic principles and a more thorough rationale.
This Workshop will cover common pitfalls and questions when analyzing safety data from clinical trials. The material will be presented in a manner appropriate for a cross-functional audience (e.g. medical, medical writers, regulatory scientists, statisticians, statistical programmers). This Workshop will provide an opportunity for attendees to gain a greater understanding of the recommendations, improve their expertise in safety analytics, and debate alternatives. Attendees are encouraged to read the following four white papers from the PhUSE Deliverables Catalog as a pre-read:
Let's Make a Knowledge Graph! An Interactive, Hands-on Workshop
Tim Williams, UCB
Knowledge Graph is a term that is gaining popularity to describe multi-dimensional graph databases that use a reasoner to infer knowledge from data. It sounds complex, but at its core is a very simple way to join data together using meaningful relationships. F.A.I.R. data (https://www.go-fair.org/fair-principles/) is built on these Linked Data concepts and it can provide future-proof data for pharma, breaking down traditional data silos in a highly inter-connected, extensible way.
This is an updated version of previous workshops at the US CSS and EU Connect conferences. We invite attendees who have not participated in the previous workshops to experience Linked Data in this interactive session. You will use a web application to diagram relationships for clinical trial processes and data, then convert your whiteboard drawing to Resource Description Framework (RDF). You will then query the data, using an ontology to infer values and relationships not in your original content. As a last step, you will seamlessly merge your study with data from all the other attendees.
This introductory workshop provides the background you need to launch your own exploration of this technology or to participate in a PhUSE Working Groups project. We welcome attendees with no previous experience with Linked Data.
Pre-registration is required. You must bring a laptop with remote desktop capability and attend a preparatory webinar in the days preceding the workshop.
Machine Learning Programming Workshop
Sairam Gorthi, JNJ Kevin Lee, Clindata Insight
The Machine Learning Programming workshop is intended for statistical programmers and statisticians who want to learn how to conduct simple machine learning projects. The Machine Learning Programming workshop will go through the following simple steps:
- Identify the problems to solve.
- Collect the data.
- Understand the data by data visualization and metadata analysis.
- Prepare data – training and test data.
- Feature engineering.
- Select algorithm.
- Train algorithm.
- Validate the trained model.
- Predict with the trained model.
This workshop will use the most popular machine learning program – Python – and the most popular machine learning platform, Jupyter Notebook/Lab. During the workshop, programmers will see actual Python codes in Jupyter Notebook to run simple machine learning projects. Programmers will also be introduced to popular machine learning modules – pandas, numpy, scikit-learn, TensorFlow and Keras.