Learning Analytics-Driven Personalized Exercise Prescriptions in Secondary School Physical Education: A Quasi-Experimental Evaluation of an AI-Supported Formative Assessment System
This quasi-experimental study demonstrates that an AI-supported, learning analytics-driven system significantly improves secondary school physical education outcomes by delivering personalized exercise prescriptions and actionable feedback, with the greatest benefits observed among students with lower baseline fitness levels.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are walking into a gym where everyone is trying to get stronger, faster, or better at a sport. In a perfect world, every single person would have a personal coach standing right next to them, watching their every move, knowing exactly which muscles are weak, and handing them a custom workout plan that changes every day based on how they performed yesterday. But in real life, especially in a crowded school gym with thirty kids at once, that's impossible. Teachers have to teach the whole class the same way, and students often just get a final score at the end of the term with no idea how to actually fix their mistakes. This is where a field called "Learning Analytics" comes in. Think of it as a super-smart detective that gathers all the tiny clues from a student's performance—like how fast they ran or how far they jumped—and turns those raw numbers into a clear map for improvement. When you mix that detective work with Artificial Intelligence (AI), you get a system that can act like that missing personal coach, even in a huge classroom. The big question researchers have been asking is: Can this digital magic actually help real kids get better at sports, or is it just a fancy gadget that doesn't change anything?
This paper tells the story of a team of researchers who decided to test this idea in a real middle school in Chengdu, China. They set up a 12-week experiment involving 133 seventh-grade students. They split the students into two groups: a "Control Group" that got the usual, standard physical education classes, and an "Intervention Group" that got the special AI treatment. For the students in the special group, the process worked like a high-tech feedback loop. First, they took a test to see their starting fitness level. Then, an AI system looked at their results and diagnosed exactly what they were bad at—maybe they were great at throwing a ball but terrible at running long distances. Based on this, the AI generated a personalized "exercise prescription" for each student, which was basically a custom workout plan with just two or three specific tasks to do every week.
But the AI didn't just talk to the students; it also talked to the teachers. The teachers had a special dashboard on their computers that acted like a weather map for the whole class. It showed them, for example, that "Hey, 80% of the kids in Class 1 are struggling with endurance running," or "Student X hasn't finished their workout plan this week." This allowed the teachers to adjust their lessons on the fly, focusing on the specific problems the AI had spotted. The students also got regular reports that told them how they were improving compared to their own past self, rather than just comparing them to everyone else.
The results of this experiment were pretty exciting. After the 12 weeks, the students who used the AI system improved their total physical education scores by an average of 3.70 points, while the students in the regular classes only improved by 1.60 points. That's a difference of 2.10 points, which is a significant jump in the world of school sports. The AI group got better at almost everything they were tested on, including endurance running, strength exercises, standing long jumps, and ball-handling skills.
One of the most interesting discoveries was who benefited the most. The students who started out with the lowest fitness levels saw the biggest gains. It's like if you were learning to ride a bike: the person who was wobbling the most at the start got the most help from the training wheels and the coach, and they improved faster than the person who was already riding pretty well. The data showed that students with lower baseline scores improved by nearly 5 points on average, while the high-performing students only improved by about 2 points. This suggests the system is really good at helping those who are struggling to catch up, which is a big deal for fairness in school.
The researchers also looked at how the students used the system to see what made the difference. They found a clear link: the more students actually finished their assigned workouts and the more often they looked at their AI progress reports, the more their scores went up. It wasn't just about having the technology; it was about using it. The teachers, too, were actively using the data. In the classes using the AI, teachers looked at the dashboards an average of 5 to 7 times a week and made specific changes to their lessons based on what they saw, whereas the control group teachers didn't have this data to guide them.
However, the paper is careful not to call this a magic cure-all. The researchers point out that this was a specific experiment in one school with a specific type of test (preparing for high school entrance exams), so we can't be 100% sure it would work exactly the same way everywhere else. They also note that the study lasted only 12 weeks, so they don't know if the kids would keep getting better or if they would forget what they learned after a few months. But the evidence they have is strong: the AI system didn't replace the teachers; it gave them a superpower to see what was happening in the classroom and helped students get personalized guidance that was previously impossible to give to everyone at once. It turns a generic "run faster" command into a specific "run at this speed for this long, then rest, then run again" plan that fits the individual kid.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.