Quanticate CRO Blog

If you search online for “how to transpose using PROC SQL” you will be told – repeatedly – that it cannot be done. Out of curiosity I decided to ignore the advice, and my solution came to 345 lines, achieving what PROC TRANSPOSE can do in 6; proving that you can do it – you just shouldn’t. However, tackling the challenge requires us to look deeper at tree traversal, used for finding combinations of variables for the transposed columns. Within SAS, basic tree traversal methods include nested and macro do loops, and as we get more advanced we can create efficient algorithms dealing with file searching, data-structuring and much more. This blog will give an overview of my method for transposing and use it to show how tree traversal can be understood and implemented in SAS.

Easily finding the right information is key to smart and timely business decisions. In Clinical Trials, it can even mean the difference between life and death! There are huge numbers of visualization tools available, but are they best? When you understand your data and user needs fully, are you better off building something from scratch within SAS? Using dummy Clinical Trial data, we'll compare our in-house web-based server-less solution, which uses PROC JSON and the Open Source D3 library for graphing and drill-down. We will also review SAS techniques helpful in building in-house platform: PROC GKPI, PROC STREAM and custom html+JavaScript codes run through SAS. As an alternative we look at paid-for solutions like Microsoft Power BI Desktop. Finally, we‟ll consider if the breadth of options offered by tools like Power BI help or hinder, compared to targeted reports defined by experts.

This blog explores what Machine Learning (ML) is and it’s difference variations. We will cover the three types of ML and present real-life examples from the pharmaceutical industry of all three types. We will also cover the SAS Data Mapper Tool which is one of the ML algorithms. In addition to this, we will also touch base upon challenges of data science and the regulatory processes for approvals of AI/ML Products.

Randomized controlled trials (RCTs) are the approach most often used in phase II and phase III clinical trials. In RCTs the probability of being assigned the experimental drug and the control is fixed throughout the trial and normally 50%, so that each drug is given to a similar number of patients. This leads to a high chance of identifying if one treatment is significantly better.

PROC DS2 is a new SAS® proprietary programming language with full release in version 9.4. It has many features but this blog’s focus will be on Object Oriented Programming (OOP) and multithreading. Multithreading and greater efficiency in the use of your system can be an exciting prospect, but the daunting task of learning OOP can slow down or block attempts to fully learn and utilise this exciting new procedure. 

Increasing amounts of data to be processed and further use of computationally intensive statistical techniques such as Bayesian Analysis and Multiple Imputation (MI) in clinical trials has resulted in a large increase in computer processing times which presents challenges when analyzing and reporting clinical trial data. This increase in processing time can cause delays in timelines if this has not been fully accounted for. The execution of the quality control (QC) programs of such tasks may also have to be coordinated and performed in parallel to limit the total time spent processing on production and QC which can be difficult to coordinate. It may even be the case that results from the production program are unable to be fully produced when double programmed due to time constraints (e.g. obtaining fewer samples in a Bayesian analysis or performing fewer imputations) which can reduce the quality of the QC process.

Randomization is widely acknowledged to be one of the more, if not the most, important parts of a properly planned and designed clinical trial. Flaws at a randomization level might lead to systematic imbalances in the allocation of patients to treatment groups, ultimately resulting in a lack of control of the overall type I error (i.e.: the pre-specified α level used as a reference for hypothesis testing). Whilst generation of randomization lists based on ‘static’ algorithms (e.g. stratified algorithms) is a relatively easy and standard process that can be done using standard software, a new type of method, that we’ll refer to as ‘dynamic’ randomization, is also increasingly used. This requires more complex algorithms to be embedded in the Interactive Web Response System (IWRS) integrated with the study database. The reason for this is that whilst with common algorithms the list of treatment allocations is fully determined a priori, i.e.: before we know the characteristic of the subjects that will be randomized, dynamic methods generate the randomization list case-wise, that is only when a new patients come in, using minimization algorithms to make sure that the groups are balanced with respect to specific characteristics not only when all subjects have been recruited, but during the whole recruitment process.

Having seen an increasing number of gene therapy approvals, the FDA has issued draft guidance¹ to help the developers of human gene therapy (GT) products for the treatment of hemophilia A & B.  In this article we will be focusing our attention on what guidance has been provided about the design of human gene therapy clinical trials for hemophilia A & B, including what is needed to support an accelerated approval approach. 

This blog explores the statistical methods used in Risk Based Monitoring (RBM) and how the result of such statistical methods enables improved data integrity across a clinical trial.