Statistic in Life Science
All the course will be held online.
Grundlagenkurs/ basiccourse
basiccourse (STAT) - englisch
Date: Fri, November 8 2024
time: 08:45 - 12:15 UTC+1
In this part, the basics of "classical" statistics are laid out. The focus here is on the principles of hypothesis testing, which include the famous T-test, but also other widely used tests. Related to this are terms such as significance level, confidence, P values, standard errors, which are also discussed here, and their relationships, as well as advantages and disadvantages, are explained. We also learn about statistical power as a tool for assessing the "test strength" of a test, how important this is in terms of reproducibility and interpretation of P values, and also as a means of finding optimal sample sizes. In addition to statistical tests on outliers, normal distribution or counts, we will also look into the complex "ANOVA".
Advanced Courses
Advanced Course 1
Design of Experiment (DOE) und Statistische Prozesskontrolle (SPC) - English (see description below)
Date: Tue/Wed, March 18/19 2025
time: 08:45 - 16:45 UTC +1
Dieser Kurs ist die ideale Ergänzung und Fortführung der statistischen Themen zum GMP-Grundkurs. Während im GMP-Grundkurs die Grundlagen der statistischen Prozesskontrolle behandelt wurden, können wir in diesem Kurs die Anwendungsmöglichkeiten der SPC weiter vertiefen. Wir lernen kennen, wie SPC, und Capability Analysis genutzt werden können, um Prozesse zu charakterisieren, wir klären bestimmte „Mythen" rund um SPC und stellen die Verbindung der SPC zum Thema Qualität her – denn letztendlich soll jedes Produkt die bestmögliche Qualität aufweisen – von jeder Faltschachtel zu jedem Bauteil in einem modernen Fahrzeug. Wie man SPC nutzen kann, um Verbesserung in Prozessen in mögliche finanzielle Gewinne durch Prozessoptimierung zu übersetzen, wird ebenfalls Teil des Kurses sein.
Richtige statistische Versuchsplanung ist ein essentieller Teil, um systematisch und effektiv zu untersuchen, welche Prozessparameter einen Einfluss auf die Qualität eines Produktes haben. Die enge Verknüpfung von Prozessoptimierung (SPC) zu richtiger Versuchsplanung (DOE) Daher werden wir auch die Grundlagen des DOE und wichtige Anwendungsmöglichkeiten kennenlernen, die direkte Anwendung in industriellem Umfang finden, sowohl mit DoE, als auch SPC-Methoden ausgewertet werden kann. Da es sowohl im DoE als auch bei der SPC es u.a. um das Aufdecken verborgener cause-and-effect-Beziehungen ging, besteht eine tiefe Verbindung in diesen beiden Themen, und nicht zuletzt wird auch in den entsprechenden USP chaptern, oder der ICH Guideline oder EMA immer wieder auf genau diese beiden Tools verwiesen, um Prozesse nachhaltig und langfristig zu verbessern.
This course is the ideal complement and continuation of the statistical topics in the GMP basic course. While the GMP basic course covered the basics of statistical process control, in this course we can delve deeper into the possible applications of SPC. We will learn how SPC and Capability Analysis can be used to characterize processes, we will clarify certain "myths" surrounding SPC and establish the connection between SPC and quality - because ultimately every product should have the best possible quality - from every folding box to every component in a modern vehicle. How SPC can be used to translate improvements in processes into possible financial gains through process optimization will also be part of the course.
Correct statistical experimental design is an essential part of systematically and effectively investigating which process parameters have an influence on the quality of a product. The close link between process optimization (SPC) and correct experimental design (DOE) We will therefore also learn the basics of DOE and important application possibilities that can be used directly on an industrial scale and evaluated using both DoE and SPC methods. Since both the DoE and the SPC were concerned, among other things, with uncovering hidden cause-and-effect relationships, there is a deep connection between these two topics, and last but not least, the corresponding USP chapters, the ICH Guideline or EMA repeatedly refer to precisely these two tools in order to improve processes sustainably and in the long term.
Advanced Course 2
Regressionsmethoden und generalized linear model (REG) - Englisch
date: Tue, May 06 2025
time: 18:00 - 20:00 UTC+2
Linear mixed effects (hierarchical) models (lineare gemischte Modelle (LMM) - Englisch
date: Tue, May 13 2025
time: 18:00 - 20:00 UTC+2
Data Visualization & Storytelling (VIS) - Englisch
date: Tue, May 20 2025
time: 18:00 - 20:00 UTC+2
In this course, participants will receive a comprehensive introduction to the basics of linear and extended regression models, with a special focus on practical applications and data-based decision-making in the pharmaceutical industry.
We will begin with simple linear regression, one of the most basic statistical models for predicting a continuous target variable based on one or more independent variables. A typical example from the pharmaceutical industry would be investigating the relationship between the dosage of a drug and the resulting reduction in blood pressure. Here, participants will learn how to set up, interpret and evaluate the quality of such models.
In the next step, we will look at logistic regression, which is used to model binary target variables. An example application in this area could be predicting whether a patient will respond to a certain drug (responder/non-responder). Here, the concept of log odds or odds ratio is introduced to interpret probabilities in the context of the logistic function. Particular attention is paid to how this model is used in real-world application scenarios, such as predicting treatment success.
We then expand our view to Generalized Linear Models (GLMs), which are a generalization of linear regression. In the pharmaceutical industry, for example, GLMs could be used to model the number of side effects in different patient groups, which can be implemented using Poisson regressions.
We then look at Linear Mixed Effects Models, which make it possible to integrate both fixed and random effects in one model. This is particularly useful when data is available from groups or hierarchies, such as in clinical studies in which repeated measurements are taken of patient groups. Here, it is possible to investigate how individual patient characteristics influence treatment success.
The course concludes with a practical part on data visualization, in which the most important techniques and tools for presenting and communicating analysis results are presented. For example, the effect of a drug in different patient groups could be visualized to support decision-making processes in clinical development. Participants will learn how to present their models in a visually understandable and convincing way.
This course is aimed at participants with basic knowledge of statistics and data analysis who want to deepen their skills in applying advanced regression models.
Advanced Course 3
available in German