Introduction to Biostatistics

Biostatistics Introduction

• Biostatistics is a branch of applied mathematics focused on collecting, analyzing, and interpreting data from biomedical research.
• Medical knowledge generation is an empirical process, relying on observations, experimentation, and inductive reasoning.
• Clinical research validates new medical knowledge in real people.
• Examples include testing vaccine effectiveness through randomized experiments where participants receive a vaccine or placebo.
• A sample is a subset of a larger population (all possible individuals)
• Effective vaccines reduce the risk of disease in the population represented by the sample.

Sample vs Population

• The effectiveness of a sample outcome (e.g., rate of disease in the vaccine group vs. placebo group) is compared to the potential effectiveness in a larger population.
• Two possible scenarios exist for analyzing samples relative to populations: Either a vaccine is effective in reducing risk or it is not effective, and the observed difference is due to chance.
• A larger sample size and more noticeable difference increase the likelihood of a true effect rather than a chance occurrence.
• Statistics provides tools to determine the likelihood of observed differences.

How Medical Knowledge is Generated

• Science relies on empirical data.
• Basic research informs clinical research, with a feedback loop between the two.
• Research verifies new knowledge in real individuals.

Clinical Research

• Clinical research utilizes samples to infer results about a larger population of interest.
• Defining the population of interest is crucial, different from the research sample.
• Inferences about a population are not based on samples, to account for population differences.
• Accurate data measurement and appropriate sample selection are essential for avoiding bias.
• Statistics is used to analyze data, making connections between samples and populations.
• Calculating the probability of differences to be due to chance is a crucial statistical task.
• Data is collected, processed, and analyzed within a rigorous clinical research design.

Data Types

• Data in research is typically rectangular or tabular format.
• A unit of observation describes the subject of the data (often a patient).
• Variables or fields are the columns of the table.
• Categorical variables include Nominal, Ordinal, and Dichotomous types.
  • Nominal variables are categorical data without inherent order (e.g., blood type, sex).
  • Ordinal variables have an inherent order (e.g., educational level, satisfaction level).
  • Dichotomous variables have only two categories (e.g., healthy/diseased, yes/no).
• Numerical Variables include Continuous and Discrete types.
  • Continuous variables represent measurements on a continuous scale (e.g., weight, blood pressure).
  • Discrete variables represent countable values with no or whole units of measure (e.g., number of children, number of deaths).

Creating and Using Statistical Plans

• Refer to the study protocol when a new dataset is received.
• Identify units of observation and variables.
• Process data by verifying variable types and handling missing values. Applying logical and consistency checks.
• Examine the distribution of variable values, including outliers.
• Conduct descriptive analyses to understand the dataset and simple univariate analyses
• Use complex multivariate analyses based on insights from simple univariate analyses; highly varies depending on the type of variables.
• Output results in tables, charts, and graphs.

Missing Values in Data

• Assess variables for missing values.
• Missing values require appropriate handling as they may influence results. Software tools help manage these situations.
• Typically some values will be excluded. Statistical methods usually ignore missing value scenarios.
• Using multiple imputation techniques may compensate for some missing data scenarios.

R Environment and Usage

• R is a statistical computing environment beyond a statistical package.
  • Flexible and extensible.
  • Utilizes real programming language.
  • Facilitates reproducible research in statistics, avoiding "black box" results.
• R features free, open-source software for collaborating research teams.
  • Cross-platform compatibility (Linux, Mac, Windows)
  • Comprehensive documents and extensive online resources.
• Install the base R environment, choose a suitable editor or IDE (e.g., RStudio) (for ease of use, highlighting, and syntax) and then load relevant packages based on research needs.

R Concepts:

• R offers interactions with the console.
• User-created scripts to perform complex statistical tasks, executable at any time.
• R utilizes objects to store data, multiple objects are stored in the workspace, each with a unique class (type).
• R functions: accept inputs, perform actions, yield output, either as objects or function calls.
• R functions can be programmed: examined for use, through online documentation.
• Arguments: some are required; others are optional (default values).
• Data structures: Vector, List, data.frame, Factor.
  • Vectors collect values; lists hold other objects; data.frames hold vectors of equal length; factors reflect categories.
• Values: assign data values to named objects.
• Indexing: selecting data elements from vectors.
• Basic functions: extract data.
• NA is a special value to signify missing data.

Working with Datasets in R

• R can import datasets from various formats, including from Excel spreadsheets using libraries in R.

Next Steps in Course

• Focus on descriptive statistics.
• Introduce the concept of statistical distributions in R.
• Complete and understand any associated lab instructions.

Additional References

• Provides sources for deeper exploration of R and statistical methodologies.

This quiz covers the essentials of biostatistics, focusing on its role in biomedical research and clinical studies. It addresses key concepts like samples vs. populations and the analysis of vaccine effectiveness through empirical processes. Test your understanding of how data is collected and interpreted in medical contexts.