Logistic Regression for International Students in Health: From Simple Formula to Machine Learning Application

This lesson is designed for international students in health sector, medical researchers, and early-career analysts who want to understand logistic regression in a clear, practical, and applied way.

Logistic regression remains one of the most important tools in public health research because many health outcomes are binary: disease/no disease, depressed/not depressed, vaccinated/not vaccinated, smoker/non-smoker, hospitalized/not hospitalized, or high-risk/low-risk.

This resource explains the method using a natural teaching style, focusing on:

• Probability, odds, and log-odds
• The sigmoid function
• Interpretation of coefficients and odds ratios
• Logistic regression as a machine learning classifier
• Model performance and prediction
• Sample size and power considerations
• Practical public health datasets for student exercises

The goal is not only to teach a formula, but to help learners understand how logistic regression can be used to solve real public health problems.

I would recommended who can use: 
Public health students, medical students, MPH learners, junior researchers, and lecturers teaching applied biostatistics or health data science.

Some words from me: 

This learning material was prepared with a strong belief that public health students can learn machine learning and statistics in a clear, practical, and meaningful way. Logistic regression is one of the most important starting points for public health data analysis. It helps us understand and predict binary health outcomes, such as whether a person has a disease, whether someone is at high risk, or whether a public health intervention reaches the intended group. However, many students find logistic regression difficult at first because it is often introduced through formulas before intuition. In this material, I try to do the opposite: begin with real public health questions, then gradually move toward probability, odds, log-odds, model interpretation, and machine learning applications. This document is not written only to teach a statistical technique. It is written to encourage confidence, curiosity, and independent thinking among learners. I hope it becomes a helpful companion for students, lecturers, researchers, and anyone who wants to use data for better health decisions.

Please read at:

Link: Logistic Regression: Machine Learning Guide from Odds to Prediction

Note: This material is shared as a learning draft and may still contain some limitations or mistakes. If you notice anything unclear, inaccurate, or not fully appropriate, I would be grateful to hear your feedback. Your suggestions will help me improve the document and make it more helpful for public health students and young researchers.

Author:
Tho Nguyen, MD, MPH
International Research Institute for Artificial Intelligence and Digital Transformation, Dong A University, Danang, 50000, Vietnam
Email: thon@donga.edu.vn