The power of your study is the probability that you will find a statistically significant difference or relationship (an “effect”) if that difference or relationship (effect) truly exists in the population.

A study with too small of a sample size is under-powered. This means that even if the effect you are testing for truly exists, you won’t achieve statistical significance. You will waste time by collecting a sample that is too small to properly power a study. Why perform a research if you can’t see significance for your desired effect?

A study with too large of a sample is over-powered. This means that you’ve collected such a large sample that you will see significance even on very small effects. However, the costs of subject recruitment, data collection, and follow-up (if needed) are quite large. Recruiting more subjects than needed unnecessarily inflates the temporal and monetary costs.

Questions related to the feasibility of a study can be answered by power analysis:

In this webinar attendees will learn the statistical power analysis and techniques for determining sample size (a priori techniques) calculation. Also attendees will get work examples in the free to use G*Power software. Some code and demonstrations will be provided for powering studies and performing power analysis simulations in R software.

Predicting future outcomes, the next steps in a process, or the best choice(s) from an array of possibilities are all essential needs in many fields. The predictive model is used as a decision making tool in advertising and marketing, meteorology, economics, insurance, health care, engineering, and would probably be useful in your work too!

Join Elaine Eisenbeisz of Omega Statistics as she presents the rationale and risks of predictive modeling via supervised learning techniques. Elaine will also provide an overview of some of the many available modeling techniques including:

• Linear regression
• Logistic regression
• Linear discriminant analysis
• K-Nearest Neighbors
• Resampling methods (Cross-Validation, Bootstrap)
• Subset selection
• Shrinkage methods (Ridge regression, Lasso regression)
• Tree-Based methods (Decision trees, Bagging, Random Forests, Boosting)

The role of statistics in clinical trials incorporates the tools used to develop a robust study, minimize bias, and assess efficacy of new treatments as relates to comparison to competing therapies.

Randomized clinical trials remain the standard for clinical research. However, the cost of a traditional randomized controlled trial especially in large sample sizes and long study duration, are limiting factors of innovation in the pharmaceutical and medical device arenas. A trial with an adaptive design can often result in lower costs and more efficiency by making use of interim analyses with data accumulated during the course of a trial to modify a study, without compromising validity and integrity. Adaptive designs also allow for checking trial assumptions and progress before the conclusion of the study. Studies can be adapted for dose response, patient accrual, and early stopping for futility or patient safety concerns.

However, there are special considerations in the planning and execution of a adaptive design to control Type I error rates and ensure consistency of treatment. In additional to protocol considerations, the FDA and regulatory agencies also require particular assurances that an adaptive design will incorporate flexibility at the expense of sacrificing study validity and patient safety.

This webinar includes an overview of Adaptive Design, and an emphasis in group sequential design, sample size re-estimation, and Phase II/III Adaptive Seamless Designs.

The objective of the webinar is to provide information that can be used immediately by personnel involved in the design and analysis of clinical trials. The presentation involves use of statistical techniques and a basic understanding of statistical theory and the framework of randomized controlled trials is desired. However, presentation of statistical theory and application will be limited to only what is needed by the attendees to understand and implement adaptive trial design and analysis.

R statistical software will be used to demonstrate analyses and simulation studies within the adaptive design framework. So bring your laptop computer!

Clinical data management (CDM) is paramount for a successful research. After all, Garbage In, Garbage Out (GIGO).

CDM involves all aspects of collecting, processing, and interpreting information. There are many types of computer applications and database systems to support data collection and management. However, there are elements of CDM that apply across the board. Review and approval of drugs or devices by regulatory agencies requires the assumption that the data presented are valid and reliable. Integrity of the data is paramount to ensure confidence in the results and conclusions you will make.

This seminar is based on FDA E6 GCP Guidelines which are the basis of effective data quality management. Even if your research is not FDA regulated, the information you learn in this course will help to ensure a robust data collection and management plan. The information conveyed in this course will also assist investigators in setting up processes for smoother data monitoring and auditing.

Examples of CRF’s and required documentation will be presented. Data collection and preparation techniques will also be demonstrated, and will include the use of CDISC and SDTM standards. Additionally, this workshop will provide you with the knowledge and tools needed to assure a CDM plan that holds up when the inevitable deviations from protocol occur.