The Full Factorial design is foundational to the theory of Design of Experiments (DOE). Although it is not the most efficient in terms of time or resources, understanding this design is essential before moving on to more advanced and economical techniques. A full factorial experiment tests every combination of factors at the same number of levels, which allows for all main effects and interactions to be measured directly. In practice, however, full factorial designs can quickly become costly and labor intensive when as the number of factors and levels increases. As a result, this design is only well suited for smaller studies.

In this tutorial, we will walk through a complete example of a full factorial DOE experiment. Using freely available Python libraries, we will generate the design matrix, explore the data, build a regression model, and interpret the results.

The late 1990s and early 2000s saw the widespread adoption of electronic health records (EHRs), which made the vast amounts of real-world patient data much more accessible to researchers. At the same time, the limitations of the classic Randomized Controlled Trials (RCT) were becoming more and more apparent: the highly controlled conditions of these trials did not always reflect the real-world usage of interventions, and their strict inclusion/exclusion criteria lead to study populations that did not represent the full diversity of patient groups who will ultimately use the drug. And with the rising healthcare costs, the need to demonstrate the value of new treatments in the real-world setting became increasingly important.

The emergence of Real-World Data (RWD) and Real-World Evidence (RWE) concepts has revolutionized how drug efficacy and safety are evaluated, how clinical trials are designed, and how healthcare decisions are made. In this note, we will cover the definitions, applications, and implications of RWD and RWE in pharmaceutical research.

Sixty years ago, in July of 1964, Abraham Savitzky and Marcel J. E. Golay published a seminal paper in Analytical Chemistry where they introduced a technique designed to smooth out noise in spectroscopic data and enhance the signal without significantly distorting the underlying trends. The Savitzky-Golay filter was one of the first signal processing techniques that I’ve learned in grad school, and the simplicity and elegance of this method has captured my heart ever since. As we approach the 60th anniversary of this influential technique, let’s explore its development, applications, and the enduring impact on science.

Oh, and we are also going to implement it in Excel VBA.

Anyone working in the pharmaceutical industry must have a good comprehension of the applicable regulations. While attending courses and workshops is very helpful in staying up to date, there is really no substitute to reading the source documents. Here I keep my currated “Reference Section” of regulatory documents and guidance that I find myself referring back to all the time. It may be useful to anyone working in clinical sample testing and/or IVD development.

Mistral is a series of Large Language Models (LLM) developed by a French company Mistral AI. These models are notable for their high performance and efficiency in generating text. Many of Mistral’s models are open source and can be used without restrictions under the Apache 2.0 license. Here is a quick troubleshooting walkthrough on how I set up dolphin-2.5-mixtral-8x7b to run locally with cuBLAS back-end for hardware acceleration. I experienced a few annoying hangups in the process, so hopefully this little note can help anyone experiencing the same.

Design of Experiments (DOE) is a systematic approach to planning, conducting, and analyzing efficient scientific experiments. It is an indispensable tool in optimization of complex processes, especially in engineering, applied research, and manufacturing. Unlike traditional one-variable-at-a-time methods, DOE involves simultaneously varying multiple factors to efficiently assess their individual and interactive effects on the outcome. The most obvious benefit of DOE is that it allows to dramatically reduce the number of experiments needed to characterize a system (well, depending on the design you choose). But more importantly, it allows us to uncover interactions between variables, something that one-variable-at-a-time testing simply cannot do.

This note provides a high-level overview of DOE principles; future notes in this series will explore specific design types and analysis techniques in more detail.

In the realm of clinical research and data management, adherence to standardized formats and structures is paramount. Two such standards, CDASH (Clinical Data Acquisition Standards Harmonization) and SDTM (Study Data Tabulation Model), play crucial roles in ensuring consistency and interoperability across clinical trial data. There is a lot of overlap between CDASH and SDTM, which often creates a false impression that they serve the same purpose, or even compete with each other. In reality, the two standards are designed to complement each other, while addressing similar but distinct needs. In this note, we delve into the key differences between these two standards to better understand their respective roles in the clinical research landscape.

Mermaid.js is one of the coolest and nerdiest projects I’ve seen in a while. I mean, you write code to generate diagrams!

And we all know that diagrams make everything better. Here, let me show you with a diagram!

Hugo is an open-source static site generator (SSG) designed for speed and efficiency. It offers sub-second rendering times, freedom to organize your content to your heart’s desire, and a good toolset for setting up advanced file processing pipelines. Having suffered through dealt with many SSGs before, I was pleasantly surprised by the simplicity × power factor that Hugo offers. Here I document my experimentation with this tool.

Here is a Markdown-formatted list about the things that I always forget about Markdown.