Biostatistics facts for kids

Kids Encyclopedia Facts

"Biometry" redirects here. For the automated recognition of people based on intrinsic physical or behavioural traits, see Biometrics.

Biostatistics is a special part of statistics. It uses statistical methods to study many different things in biology. This includes planning experiments, collecting and analyzing data from them, and understanding what the results mean. It helps scientists make sense of numbers in the world of living things.

A Look Back: Biostatistics and Genetics

Biostatistics has been very important in understanding modern biological ideas. For example, genetics studies have always used statistics to understand what happens in experiments. Some scientists even helped create new statistical tools!

Gregor Mendel started genetics by studying how traits were passed down in pea plants. He used statistics to explain his findings. Later, in the early 1900s, scientists rediscovered Mendel's work. But there was a debate between those who followed Mendel's ideas (Mendelians) and those who supported different theories (biometricians).

One famous biometrician, Francis Galton, thought that traits came from many ancestors. But William Bateson and others, following Mendel, believed traits came only from parents. This led to big arguments! Eventually, Mendel's ideas won out.

By the 1930s, statistical models helped solve these disagreements. They led to the "modern evolutionary synthesis." This brought together genetics and evolution.

Key Scientists in Population Genetics

Three important scientists used statistics to create population genetics. This field studies how genes change in groups of living things over time.

Ronald Fisher: He developed many basic statistical methods. He studied crop experiments and wrote important books like Statistical Methods for Research Workers. He also gave us ideas like the ANOVA (a way to compare groups) and the p-value (which helps decide if results are meaningful).
Sewall G. Wright: He created F-statistics to understand how genes are related within a population.
J. B. S. Haldane: His book, The Causes of Evolution, showed how natural selection works using math and Mendel's genetics.

These scientists helped combine evolutionary biology and genetics. They made it possible to study these fields using numbers and data.

Planning Your Research

Any research in life sciences tries to answer a scientific question. To get good answers, we need accurate results. Planning carefully helps reduce mistakes. A good research plan includes:

The question you want to answer.
What you think will happen (your hypothesis).
How you will set up your experiment.
How you will collect data.
How you will analyze the data.
The costs involved.

It's important to follow three main rules for experiments: randomization (doing things randomly), replication (repeating the experiment), and local control (managing conditions).

What's Your Question?

Your research question sets the goal for your study. It needs to be clear and focused on something new and interesting. To find a good question, you might need to read a lot of other scientific papers. This helps make sure your research adds value to the scientific community.

Making a Hypothesis

Once you have a question, you can guess the possible answers. This guess is called a hypothesis.

The main guess is the null hypothesis (H₀). This usually says there is no difference or no effect. For example, if you test two diets for mice, H₀ would be: "There is no difference between the two diets on mouse metabolism."
The alternative hypothesis (H₁) is the opposite. It says there is a difference or an effect. For the mice, H₁ would be: "The diets have different effects on mouse metabolism."

You define your hypothesis based on what you want to find out. There can even be more than one alternative hypothesis!

Choosing a Sample

Scientists usually want to understand how something affects a whole population. In biology, a population can be all the individuals of a certain species in an area. But in biostatistics, it can also mean all of a specific part of an organism, like all the cells in a body.

It's usually impossible to measure every single thing in a population. So, we use sampling. This means picking a small, representative part of the population. This "sample" should show most of the differences found in the whole population. The size of your sample depends on your research goals and available resources.

Designing Your Experiment

Experimental designs are the blueprints for your study. They help you randomly assign different treatments. Common designs include:

completely randomized design
randomized block design
factorial designs

In agriculture, good design is key because the environment greatly affects plants and animals. In clinical studies, samples are often smaller. Scientists use randomized controlled clinical trials to compare results.

Collecting Your Data

How you collect data is important. It affects your sample size and experiment design.

For qualitative data (like descriptions or categories), you might use surveys or observations. For example, you could rate disease severity on a scale.
For quantitative data (numbers), you use instruments to measure things. In agriculture, you might measure crop yield. For diseases, you might use score scales.

Modern methods, like high-throughput platforms for studying genes, allow scientists to collect huge amounts of data quickly. All collected data must be stored in an organized way for later analysis.

Important Statistical Ideas

Power and Errors

When you test a hypothesis, there are two types of mistakes you can make:

Type I error (or false positive): This is when you incorrectly say there's a difference, but there isn't.
Type II error (or false negative): This is when you miss a real difference and say there isn't one.

The significance level (called α) is the chance of making a Type I error. You choose this before you start. The statistical power of your test (1 − β) is the chance of finding a real difference if one exists.

What is a p-value?

The p-value tells you how likely your results are if your null hypothesis (H₀) is true. A small p-value (usually less than 0.05) means your results are unlikely to happen by chance. If p is less than α, you reject the null hypothesis. This suggests there is a real effect.

Testing Many Things at Once

When you test many hypotheses at the same time, the chance of getting false positives goes up. Scientists use strategies to control this. One way is the Bonferroni correction, which makes it harder to say a result is significant. Another is controlling the false discovery rate (FDR), which is less strict but might lead to more false positives.

New Developments and Big Data

Recent changes have greatly impacted biostatistics. We can now collect huge amounts of data very quickly. Also, computers can do much more complex analysis. This comes from advances in areas like DNA sequencing, Bioinformatics, and Machine learning.

Using High-Throughput Data

New technologies, like microarrays and next-generation sequencers, create massive amounts of data. Biostatistical methods are needed to find the real signals in all this "noise." For example, a microarray can check thousands of genes at once to see which ones act differently in sick cells compared to healthy ones.

Sometimes, many predictors (like gene levels) are very similar. This is called Multicollinearity. Biostatisticians use methods like principal component analysis to simplify this data. Old statistical methods don't work well with huge datasets where there are more features than observations. In these cases, it's important to test your model on new, independent data.

It's also helpful to group information from many predictors. For example, Gene Set Enrichment Analysis (GSEA) looks at whole groups of related genes. This approach is stronger because it's less likely that a whole group of genes would appear changed by accident.

Using Computers for Analysis

Modern computers and cheap computing power have made complex biostatistical methods possible. These include bootstrapping and re-sampling methods, which involve re-using data to get more reliable results.

Recently, random forests have become popular. They create many "decision trees" to help classify data. Decision trees are easy to understand, even without a lot of math knowledge. This makes random forests useful for systems that help doctors make decisions.

How Biostatistics is Used

Public Health

Biostatistics is vital in Public health. This includes epidemiology (studying diseases), health services research, nutrition, and environmental health. It helps plan and analyze clinical trials. For example, it can assess how serious a patient's condition is or predict how a disease might progress.

With new technologies, biostatistics is also used in Systems medicine. This aims for more personalized medicine by combining patient data, genetic information, and other "omics" data.

Understanding Genetics

Biostatistics helps link differences in genes (genotype) to differences in traits (phenotype). Scientists want to find the genetic reasons for traits that vary a lot, like height or weight. These are called quantitative traits.

QTL mapping helps find regions in the genome responsible for these traits. It uses molecular markers and studies populations created from experimental crosses.
Genome-wide association studies (GWAS) identify these regions based on how traits are linked to markers in natural populations. This was made easier by new ways to quickly check many genetic variations.

In animal and plant breeding, markers help in selection. This is called marker-assisted selection. Genomic Selection (GS) uses all molecular markers to predict how well animals or plants will perform. This helps breeders choose the best ones for breeding.

Gene Expression Data

Studies that look at how much genes are "expressed" (turned on or off) use biostatistics. This includes data from RNA-Seq and microarrays. The goal is to find genes that change their activity under different conditions. Experiments are designed with repeats and randomization.

RNA-Seq data, which counts gene activity, is analyzed using special statistical distributions. This helps account for natural biological differences.

Other Areas

Biostatistics is also used in many other exciting fields:

Ecology and predicting environmental changes.
Analyzing biological sequences (like DNA).
Systems biology to understand gene networks.
Developing new medicines.
Studying population dynamics, especially in fishing.
Understanding Phylogenetics (how species are related) and evolution.
Studying how drugs affect the body (Pharmacodynamics and Pharmacokinetics).
Analyzing brain images (Neuroimaging).

Tools for Biostatistics

Many tools help with statistical analysis of biological data. Here are a few:

R: This is a free computer language and environment. It's great for statistics and making graphs. It has many "packages" (extra tools) made by scientists worldwide, especially for Bioinformatics.
SAS: A widely used software for data analysis in universities and businesses.
Orange: A visual tool for data processing, data mining, and showing data. It has tools for gene expression.
Weka: A Java software for machine learning and data mining. It includes tools for visualizing data, grouping data, and making predictions.
Python (programming language): A popular programming language used for image analysis, deep learning, and machine learning.

Other important tools include:

ASReml (for estimating variance)
CycDesigN (for creating experimental designs)
PLA 3.0 (for drug testing analysis)
SQL databases (for organizing data)
NumPy and SciPy (for numerical computing in Python)
MATLAB (for technical computing)
Apache Hadoop and Apache Spark (for big data processing)

Learning Biostatistics

Most biostatistics programs are for students who have already finished college. You can often find them in schools of public health, or connected to schools of medicine, forestry, or agriculture. Some universities have special biostatistics departments. Others have biostatistics experts working in statistics or epidemiology departments.

Biostatistics departments might focus on bioinformatics and computational biology. Older departments, often linked to public health schools, might focus more on disease studies and clinical trials.

The main difference between a general statistics program and a biostatistics program is that statistics programs might do more theoretical research. Also, statistics programs might cover business or economics, while biostatistics focuses on biological and medical uses.