Predicting Life Expectancy: A student-built app turns global health data into insight

Have you ever wondered how long you might live? What if there were a way to predict your life expectancy—not through magic or fortune-telling, but through data and calculation? Would you be curious to find out?

Explore the Life Expectancy Prediction application developed by Ei Mon Soe, a senior student majoring in Statistics and Data Science (SDS) at Parami University. Using machine learning techniques, the project analyzes health, immunization, and socio-economic indicators to estimate life expectancy. Rather than offering a mystical prediction, this data-driven approach considers various aspects of lifestyle and living conditions. Drawing on a dataset sourced from Kaggle—originally provided by the World Health Organization (WHO)—the project transforms complex global health data into an intuitive tool that offers immediate insights into a country’s overall health profile.

Reflecting on her inspiration, Ei Mon Soe shared,“I have always wondered what truly determines how long a person lives. Is it solely medicine, or do factors such as economics, education, and social conditions play an even greater role?” This curiosity led her to develop the project as part of her midterm assignment for the Introduction to Machine Learning course.

Students enrolled in the Introduction to Machine Learning course taught by Dr. Nwe Nwe Htay Win are given the opportunity to develop machine-learning–based applications focused on projects that are personally meaningful to them. This hands-on learning approach reflects the essence of Parami University’s liberal arts education, where interdisciplinary thinking, real-world problem-solving, and personal engagement are central to the learning experience.

Data Preprocessing & Cleaning 

To turn this curiosity into a functional and reliable application by 2018, the project required careful handling of real-world data. The app was built based on the dataset from 2000 to 2015, and one of the limitations is that it cannot tell the exact current life expectancy, having some variability. Building an accurate life expectancy prediction model goes beyond selecting an algorithm; it depends on how well the data is prepared, cleaned, and structured before training. This made data preprocessing and feature engineering a critical next step in transforming raw global health data into meaningful and dependable predictions.

“Originally, the dataset had 2938 rows and 22 columns. Data in the real world is rarely perfect. The WHO dataset I used (covering 193 countries over 15 years) was noisy and had missing information. To fix this, I applied some rigorous statistical techniques,” explained Ei Mon Soe.

Data preprocessing was conducted as the initial stage of the machine learning pipeline. Missing values were addressed using mean imputation for columns with minimal gaps, country-level grouped mean imputation for the Alcohol variable, and median imputation for skewed features such as GDP, Hepatitis B, and Population. Feature engineering combined related variables into composite indices, including an Immunization Index and a Thinness Index, reducing input complexity while preserving key information. A linear regression model trained on the processed data achieved an accuracy of 81%, with predictions within approximately three years of actual life expectancy values.

Key Findings: What Matters Most?

This app is built for students, health workers, and curious individuals who want to understand the “why” behind health statistics without digging through messy spreadsheets. It helps users see the immediate connection between everyday factors—such as a few more years of schooling or better access to vaccines—and a longer life. By making this data easy to explore, the project turns a complex global health report into a simple, eye-opening tool for anyone to use.

As the developer explains, “My goal is to bridge the gap between complex data science and public awareness, ensuring that even those without a technical background can understand the factors that shape our longevity.”

The model reveals that education is the most influential factor in extending life expectancy, with more years of schooling strongly linked to longer lives. Economic factors, including how efficiently a country uses its resources and its GDP, also play a significant role. Vaccinations such as Polio and Diphtheria contribute to increased longevity, while higher adult mortality is associated with reduced life expectancy. Together, these findings highlight how investments in education, economic efficiency, and preventive healthcare can meaningfully shape population health outcomes.

Web Application Features

The application is developed using Streamlit with an emphasis on user experience (UX) and includes data from 193 countries, including Myanmar. Check out the application to find out your life expectancy. 

To get started, users select a country from the dropdown menu. The application then automatically assigns the country’s development status (developed or developing), which is locked to ensure data accuracy.

Next, users adjust input values using sliders and numeric fields designed for straightforward data entry. As inputs are modified, a dynamic summary table updates in real time to reflect the selected parameters.

In the final step, users generate a life expectancy prediction. Based on the predicted value, the application classifies life expectancy into four categories: Critical (≤ 45 years), At Risk (45–55 years), Unhealthy (55–70 years), and Healthy (> 70 years). The results are presented alongside visual elements to support interpretation.