Staying active is one of the most important things you can do for your health. Regular exercise will help you live longer, lower your risk of illness, improve your mood and increase your energy levels. However, in reality, only a small percentage of people meet exercise recommendations. So why does someone continue to commit to the movement?
A team of researchers set out to find the answer. At the University of Mississippi, scientists used machine learning to analyze national health data, and WHO found patterns with physical activity (PA) guidelines and reasons. This approach helps doctors and trainers better support your health by understanding the motivations of people like you to keep moving.
The study, published in the Journal, Scientific Reports, examined data collected from the National Health and Nutrition Survey, which were collected between 2009 and 2018. This is a large US survey that tracks health and dietary habits. The research team included doctoral students Seungbak Lee, Jupil Chou and Professor Minsu Kang. They used a tool called Machine Learning to organize over 30,000 survey responses.
Machine learning helps computers find patterns in large quantities of data. Unlike older statistical tools that expect clean and linear data, machine learning works well even when the data is messy or complex. Like people who stick to exercise routines, they can organize the most important information in predicting behavior.
Researchers filtered the data to include only people over the age of 18 who are free from cancer, diabetes, and arthritis, who can limit exercise. After deleting the missing entries, the final dataset included 11,638 participants.
Each person's responses were divided into three main areas: demographics (such as age, gender, race, income), physical measurements (such as body mass index and waist size), and lifestyle habits (such as alcohol use, smoking, sleep, and sedentary times). The goal was to build a model that could predict whether someone met weekly activity guidelines.
According to US health officials, adults should get at least 150 minutes of moderate exercise or 75 minutes of intense activity each week. Unfortunately, the average American gets around two hours of activity each week.
Using six different machine learning algorithms, researchers constructed 18 predictive models to test a variety of factors. These models were measured by balancing how accurate they are, how well they can find patterns, and how well they can predict.
The best performance model was the decision tree using all the available variables. The accuracy was about 70.5% and the F1 score (balance between accuracy and recall) was 0.819. In other words, in most cases, they correctly predicted who met the exercise guidelines.
But not just performance, but the team wanted to know which specific factors would be most useful for making predictions. Using a technique called the importance of permutation function (PFI), we found that sedentary behavior, age, gender, and educational status were the most important predictors. Some models gave slightly different answers, but these factors continued to appear again and again.
“We expected factors like gender, BMI, race and age to be important to our predictive models, but we were amazed at how important the educational situation is,” Ju-Pil Choe said.
The team noted that people who have been sitting for a long time, have a low level of education, or have a particular gender are less likely to meet activity guidelines. This helps explain who is likely to stick to physical activity and why. These insights can guide future programs aimed at helping people develop healthier habits.
Although the results are promising, researchers pointed to some limitations on their approach. One important issue is that the survey data depends on self-reported activity levels. Often people overestimate how much exercise they exercise when asked to remember it from their memory.
“One limitation of our study was the use of subjectively measured physical activity data,” Cho said. “More accurate and objective data will improve the reliability of the research.”
Future research can fix this by using wearable fitness trackers or apps that automatically record physical activity. Machine learning can use its objective data to find even more powerful and detailed patterns.
Despite this limitation, this study shows that machine learning has great promise for studying health behavior. Not only does it tell you what trends are, it also helps to clarify why these trends exist in the first place.
Why is this all important? Because understanding the reasons behind someone's exercise habits can help healthcare professionals create better, more personalized plans. Instead of giving the same advice to everyone, doctors can use data-driven models to understand their motivations.
For example, if someone has a sedentary job and a low level of education, they may need more support and different types of motivation to maintain their activity. Knowing these factors are important allows experts to build programs that work for each individual.
This is especially useful for trainers, coaches, and even health app developers. They can create exercise routines that feel more achievable and tailored to your lifestyle, age and daily habits. It's easier to stick to your workout plan and more realistic.
Professor Kang summarized the purpose of this study. “Adhering to physical activity guidelines is a public health concern due to the relationship between disease prevention and overall health patterns. We wanted to use advanced data analytics techniques, such as machine learning, to predict this behavior.”
Other research has already used machine learning in relevant ways. For example, some researchers have built models to classify children's physical activity using motion sensors. Others used neural networks to sort activity levels based on body movement. However, this study is one of the first to focus on predicting adherence to activity guidelines using only self-report data and a wide range of demographic, physical, and lifestyle factors.
The results show that machine learning can be a powerful tool for public health. It reveals patterns that may not be visible in traditional ways. And it gives researchers a new way to help people live healthier and longer lives.
