Introduction to Data Science Biostatistics

Introduction to Data Science

• Textbook for Biostatistics
• Author: Agouzoul Hibatallah
• Supervisor: Pr. Issam Bennis
• Email: [email protected]
• University: Université Mohammed VI des Sciences et de la Santé (UM6SS)

Learning Objectives

• Understand and classify variable types (categorical vs. quantitative) using Jamovi.
• Organize and manage data efficiently in Excel for statistical analysis.
• Transform continuous variables into categorical ones in Jamovi or Excel.
• Create effective data visualizations (histograms, bar charts, scatter plots) in Jamovi and Excel.
• Understand cross-sectional study design and its application in biostatistics and medical research.
• Conduct basic statistical tests (Chi², t-test, correlation) to analyze relationships within data.

Table of Contents

• Introduction (page 4)
• Types of variables (pages 2-9)
• Data organisation (pages 10-11)
• Transforming quantitative to categorical variable (pages12-13)
• Visual presentation of data (pages 14-20)
• Cross-sectional studies (pages 21-22)
• Simple data analysis (pages 22-26)

Key Concepts in Biostatistics and Data Analysis

• Data analysis in biostatistics begins with organizing, cleaning data.
• Proper variable classification is essential for choosing statistical methods.
• Data transformation and visualization help uncover patterns.
• Understanding study designs (e.g., cross sectional) helps interpret data contextually.
• Statistical tests (Chi-square, t-test, correlation) reveal relationships.

Definition and Types of Variables

• Variables are characteristics, numbers, or quantities measurable across individuals or items.
• Quantitative Variables: describe quantities and can be measured along a scale
  • Continuous: Numeric values with infinitely many possible values within a range (e.g. height, blood pressure)
  • Discrete: whole numbers only (e.g., hospital visits, family members)
• Qualitative Variables: describe characteristics or qualities, not measurable on numerical scale
  • Nominal: categories with no inherent order (e.g., gender, blood type)
  • Ordinal: categories with a meaningful order (e.g., education level, disease stage)

Qualitative Variables

• Categorizing variables based on characteristics for data analysis.
• Nominal: Different categories without any inherent order; (e.g., eye color, marital status, blood type).
• Ordinal: Ordered categories; (e.g., pain severity, socioeconomic status, stages of addiction).

Quantitative Variables

• Numerical variables representing measurable quantities.
• Discrete: Whole numbers without intermediate values (e.g., hospital visits, surgeries).
• Continuous: Infinite number of values within a range (e.g., weight, temperature, cholesterol levels).

Classifying Variables in Jamovi

• Jamovi automatically detects variable types (nominal, ordinal, continuous, discrete) upon importing data.
• Users can manually adjust these classifications in the variable settings.
• Variables can be assigned specific roles (e.g., dependent or independent) for statistical analyses.

Applying Data Variables in Healthcare

• Both quantitative and qualitative variables are essential to understand disease patterns and outcomes.
• Quantitative variables (e.g. blood pressure, cholesterol) can be analyzed using descriptive statistics or regression models
• Qualitative variables (e.g. gender, disease status) can be analyzed with frequencies, or chi-square tests.
• Combining both types of variables leads to more comprehensive insights.

Excel for Data Organisation

• Effective data organization is crucial for biostatistical analysis.
• Use a structured layout with one column for each variable; rows represent unique observations.
• Label columns clearly and consistently using underscores or camel case for clarity (e.g., "Date_of_Birth", "BloodPressure").
• Use appropriate placeholders for missing data ("N/A" or "#N/A").

Labeling and Arranging Columns in Excel

• Use clear, consistent labels; prioritize identifying variables at the start.
• Group related variables together for organized data exploration.
• Maintain consistent data formats within each column to enable accurate analysis (e.g., numerical, categorical).

Importance of Unique Identifiers in Data

• Track individual data accurately.
• Facilitate data merging and linking from multiple sources (e.g surveys, clinical records).
• Minimize errors and duplicate data.
• Ensure effective data integration and privacy.

Grouping Quantitative Variables

• Converting continuous to categorical variables (binning) groups data into predefined intervals.
• Helps simplify analysis, identify trends, and calculate prevalence of conditions within defined ranges.

How Categorizing Continuous Variables Improves Data Interpretation

• Simplifies complex data and helps detect trends.
• Makes comparison of groups easier.
• Allows more suitable analysis (e.g. using chi-square tests).
• Enables better decision-making.

Visual Presentation of Data

• Effective visual representation of data is vital for clear communication and understanding.
• Quantitative Data: Use scatter plots for relationships between continuous variables, boxplots to summarize the distribution and identify outliers and histograms to display data distribution within specific intervals.
• Qualitative Data: Use Bar charts to show frequency or proportion of each category and pie charts to show the relative proportion of each category.

Key Differences Between Histograms and Bar Charts

• Histograms: display distribution of continuous data
• Bar Charts: display frequency or proportion of categories

Using Jamovi Software for Data Visualization

• Jamovi is a user-friendly software for creating graphs (e.g., histograms, box plots, bar charts).

Jamovi: A Powerful Tool for Data Visualization and Statistical Analysis

• Jamovi offers a wide range of tools for data visualization (histograms, box plots, bar charts, scatter plots).
• Integrates statistical analysis tools for generating descriptive statistics, tests, or regression analyses.

Descriptive Statistics

• Descriptive statistics (e.g. mean, median, standard deviation) provide summaries of data.
• Jamovi calculates relevant statistics and displays results alongside visualisations (e.g. histograms, bar charts).

Regression Analysis

• Jamovi allows for linear and logistic regression analyses.
• Enables visualization of results (scatter plots, regression lines).

ANOVA (Analysis of Variance)

• Jamovi facilitates one-way and two-way ANOVA to compare group differences.
• Jamovi provides visualizations like box plots, bar charts, or plots to show group differences and interactions.

Factor Analysis and Principal Component Analysis (PCA)

• Jamovi helps perform factor analysis and PCA, visualising results using biplots, scree plots.

Non-parametric Tests

• Jamovi handles various non-parametric tests including Mann-Whitney U test, Kruskal-Wallis test, and Friedman test, displaying visual results like box plots.

Customizable Plots

• Jamovi simplifies customization of plots (axis labels, color schemes, formatting) for more intuitive presentation.

Data Import and Export

• Jamovi handles various data formats (Excel, CSV, SPSS).
• Exports generated outputs.

Reliability Analysis (Cronbach's Alpha)

• Jamovi computes reliability coefficients such as Cronbach's alpha.
• Visualizes results for ease of access.

Data Transformation

• Jamovi allows for transforming data (e.g., creating new variables, recoding, normalizing data).

Study Designs in Medical and Epidemiological Research

• Cross-Sectional Studies: collect data from a population or representative subset at a single time point; not suited for exploring cause and effect, but valuable for prevalence estimates
• Longitudinal Studies: collect data from the same subjects over a prolonged period; useful for studying trends and cause-and-effect relationships.
• Interventional Studies: involve actively influencing a subject with a treatment or intervention; establishes causality more effectively than observational designs.

Analyzing Data from Cross-Sectional Studies

• Descriptive statistics: summarize data to describe characteristics (e.g. mean, median, frequency, percentage) of the population.
• Categorization: classify variables into groups (e.g. age groups, diseases) to facilitate comparisons.
• Statistical Tests: (e.g., chi-square, t-test, regression) to explore relationships between variables (e.g., smoking and lung disease), and estimate prevalence of conditions.