INTRODUCTION

In March 2020, the World Health Organization declared a pandemic in relation to COVID-19, bringing significant changes to people’s lives. During the COVID-19 emergency phase, people were advised to maintain social distancing and follow a series of health precautions, which led to a shift in people’s routines, beginning a period in which everyone had to adapt to the reality imposed by the disease.

With advances in knowledge about the disease, the World Health Organization announced in March 2023 that the COVID-19 emergency phase had ended. This does not mean that the pandemic is over. However, public health systems were being directed to begin transitioning from emergency mode to manage COVID-19 and other infectious diseases^1,2. However, the COVID-19 pandemic imposed drastic changes on the routine of university students around the world, resulting in lifestyle changes, including increased sedentary activities, especially in front of computer screens³.

These changes can impact the physical and mental health of these students and their academic performance, as a sedentary lifestyle is a health risk factor, increasing susceptibility to obesity, cardiovascular disease, and cognitive impairment⁴. Remote learning during the pandemic has forced students worldwide to spend more time sitting at screens. The increase in sitting time and sedentary lifestyle can be observed in Europe⁵, the USA⁶, Asia⁷, and South America⁸. According to Matos and collaborators⁹, COVID-19 has affected young university students in Brazil, decreasing physical activity and increasing time spent sitting, aggravating a sedentary lifestyle already present in this population.

With increased sitting time, especially in front of screens, students’ academic performance has declined. According to Frota and collaborators¹⁰, performance in educational activities has decreased significantly, despite increased sitting time. These data are supported by other authors who claim that there is a relationship between excessive sitting time and lower academic performance among students, and that university students are directly affected, with their academic performance decreasing, leading to unsatisfactory or low performance^{11, 12}.

With the real increase in sitting time, the rise in sedentary lifestyle, and the implications for students’ academic performance during the COVID-19 emergency phase, it is necessary to determine whether these variables are linked. Therefore, the present study aimed to investigate whether there is a link among sitting time, academic performance, and physical activity practice among university students after the COVID-19 emergency phase.

METHODS

RESULTS

Table 1 presents the frequency distributions for the categorical variables in the sample, including absolute, relative, and cumulative relative frequencies.

The behavior of the three main study variables was presented in Table 2. The mean and standard deviation, median, normality, p-value, effect size, and percentage variation of the means for the variables Sitting time, physical activity time, and academic performance were presented. The present study revealed a 98% decrease in sitting time after the emergency phase. Physical activity time decreased by 34% after the emergency phase. Academic performance increased by around 6%. Variations were found with a large effect size for the variable Sitting time (3.1^L) and a small effect size for Academic performance (0.3^S), demonstrating substantial variation in these two study variables between the beginning and the end of the critical period of Covid-19.

Figure 1 (A) shows the behavior of the study variables through Boxplot and their quartiles. Significant differences (P<0.05) were found for the variables Sitting time and Academic performance. The dendrogram allows visualizing the hierarchical structure of the clusters, helping to identify patterns and relationships between the data. The dendrogram presented in Figure 1 (B) shows a similarity relationship between the variables Physical activity time and Academic performance.

Figure 1 : (A) Boxplot with the behavior of the medians and their quartiles (Mann-Whitney test P-values); (B) Similarity dendrogram (Euclidean Index). 

For daily sitting time, data are presented as boxplots, with the center line indicating the median, the box edges representing the 25th and 75th quartiles, and whiskers extending to 1.5 times the interquartile range. Gray circles represent outliers. The Wilcoxon test for paired samples revealed a statistically significant reduction in sitting time (p < 0.001). This drastic reduction reflects the transition from remote to in-person learning, where students no longer remain seated for long, uninterrupted hours in front of screens. For time spent on physical activity, the Wilcoxon test for paired samples showed a downward trend, though not statistically significant (p = 0.069). The 34% decrease in physical activity suggests that, despite the return to in-person routines, students did not resume pre-pandemic levels of structured physical exercise, possibly due to changes in habits, stress, or altered priorities. Regarding comparisons of academic performance, given the reduced number of observations in the pre-COVID period, interpretation should be cautious. The Wilcoxon test for paired samples showed no statistically significant difference (p > 0.05). The apparent 6% improvement in academic performance may reflect selection bias or students adapting to the new educational context (Figure 1A).

Each point represents a participant. Pearson’s correlation coefficients (r) and p-values are presented in each panel. None of the correlations were statistically significant (all p > 0.15), indicating no strong linear association between the variables. The dispersion of the points reinforces the complexity and multifactorial nature of academic performance (Figure 2).

Figure 2 : Scatter plots showing the bivariate relationships between sitting time, physical activity time, and academic performance in the post-COVID-19 period (n=216). 

The dashed black line represents the perfect prediction (where predicted = actual). The large dispersion of points around the ideal line highlights the model’s low predictive capacity, suggesting that sedentary behavior and physical activity, in isolation, are insufficient to explain academic performance (Figure 3).

Figure 3 : Scatter plot comparing actual academic performance values with values predicted by the Linear Regression model (best model, R2 = 0.031). 

Negative or near-zero R² values indicate that the models failed to capture the data’s variability and thus failed to establish a robust predictive relationship. The Linear Regression model showed the best relative performance (R² = 0.031), but still explained less than 4% of the variance, highlighting that other unmeasured factors are determinants of academic performance (Figure 4A). Lower RMSE values indicate a better fit. Linear Regression showed the lowest error (RMSE = 1.497), followed by SVM and AdaBoost. More complex models, such as Gradient Boosting and Decision Tree, showed overfitting, with good performance in training but high error in testing (Figure 4B). Error bars represent the variability in performance between folds. Negative average R² values indicate that, on average, the models underperformed the simple average prediction, reinforcing the absence of a consistent predictive relationship between behavioral variables and academic performance (Figure 4C).

Figure 4 : Comparison of Machine Learning Models. (A) Comparison of the performance (R2 on the test set) of eight machine learning models trained to predict academic performance from sitting time and physical activity time (n=216). (B) Comparison of the root mean squared error (RMSE) in the test set of the eight machine learning models. (C) Comparison of the stability of Machine Learning models through 5-fold cross-validation (mean R2 ± standard deviation). 

The graph shows the relationship between the variables (represented by vectors) and the distribution of participants (points), colored according to three clusters identified by K-means. The first principal component (PC1) explains 36.7% of the variance. It is strongly associated with sitting time, while the second component (PC2) explains 35.7% and is related to physical activity and, inversely, to academic performance. The dispersion of the clusters indicates heterogeneity in behavioral patterns, but does not reveal clear groupings associated with academic performance (Figure 5).

Figure 5 : Biplot of Principal Component Analysis (PCA) of behavioral and performance variables in the post-COVID-19 period (n=216). 

DISCUSSION

During the critical period of Covid-19, changes in students’ routines meant they no longer needed to travel to the university or walk to the library or to a place to eat, since all of these activities were carried out at home^16,17. This sudden change in students’ routines led to an increase in sitting time in this population^16,18.

The present study revealed a 98% decrease in sitting time among university students after the critical period of COVID-19, which contradicts Romero-Blanco et al.¹⁶, who reported that university students’ sitting time increased by more than 2 hours per day. One of the possibilities for the decrease in sitting time observed in the present study may be a greater appreciation of activities prohibited during the lockdown, or less time spent effectively studying, often due to difficulty accessing the technology used for classes, or simply due to a lack of internet access¹⁸.

The Post-Pandemic Behavioral Paradox: The analysis confirmed the drastic reduction in sitting time (-98%) and the decrease in physical activity (-34%). The central insight is that these phenomena are not mutually exclusive. The reduction in sitting time is linked to structural changes (the end of remote classes). At the same time, the drop in physical activity reflects a behavioral change in leisure time, a “physical activity paradox” already documented in the literature in other contexts.

One of the biggest concerns among college students about the time spent sitting is the increase in sedentary lifestyles in this population. Increased sedentary lifestyles can lead to increased anxiety and depression in this population, in addition to being associated with increased mortality from any cause¹⁹. The time spent on physical activity during the pandemic’s critical period was affected by the ban on gatherings and outdoor activities. The present study revealed a 34% decrease in physical activity during the crucial period of the pandemic, which is in line with reports in the literature for the same population of college students^20,21. In contrast, Romero-Blanco et al.¹⁶observed an increase in physical activity among college students during the same period, which, according to the authors, is justified by the motivation to stay healthy, even in the face of difficulties encountered, since they were students in the area of Health Sciences. The present study shows a 6% increase in academic performance. The increase in academic performance has already been reported in the literature²², but the author states that much of it may be related to the use of technologies to solve problems and issues related to the course and the discipline studied. This discussion is fueled by studies showing a decrease in students’ academic performance, where the authors describe the impact of these technologies on sleep quality, which would negatively affect university students’ academic performance²³. We can observe the importance of technology and its dangers when we talk about teaching during the critical phase of COVID-19 and students¹⁸.

When we observed the three variables of the present study together, a similarity emerged between time spent practicing physical activities and academic performance. However, no similarity was observed between sitting time and any other variable. The association between physical activity and academic performance has been the subject of study for a long time. The literature states that physical activity leads to cognitive benefits that justify increased academic performance²⁴. These data underscore the importance of physical activity during the critical period of COVID-19, offering significant physical and mental benefits for maintaining health and academic performance^25,26. Another point that deserves attention is the lack of association between academic performance and sitting time, indicating that even though students spend more time sitting in front of screens, often after classes, this is not a predictor of academic performance, as reported in the literature^11,17.

Considering that exposure to adverse experiences in childhood is regarded as a significant predictive factor for overweight or obesity²⁷, especially after the COVID-19 pandemic, which had an important social impact on all age groups²⁸, behavioral and attitudinal studies involving post-traumatic stress are essential for science and public health²⁹.

This study has some limitations, such as a lower response rate in the second data collection. However, even with dropouts during the evaluation period, the study still had a significant number of participants, and it also raised an essential topic for discussion. The analysis revealed a flaw in the collection of “pre-COVID academic performance” data, with only 31 valid observations. This is a fundamental insight, as it justifies the impossibility of robust paired analyses for the primary outcome variable and underpins the need to focus association analyses on the post-COVID cross-section, treating this as an explicit and essential methodological limitation.

Absence of a Direct and Predictive Relationship: The most striking finding is the lack of a statistically significant association between sitting time, physical activity, and academic performance in the post-pandemic data. None of the correlation tests, regression models (R² < 2%), or the eight machine learning algorithms (maximum test R² of 0.031) established a robust predictive relationship. Insight: This strengthens the article by turning a weakness (the absence of tests) into a relevant finding. The conclusion is not that there is no relationship, but that, in this population and in this context, the relationship is weak, non-linear, or strongly moderated by other factors (mental health, study strategies, etc.) that were not the focus of the study, opening a clear path for future research.

CONCLUSION

The findings of the present study show a decrease in sitting time and an increase in academic performance between the beginning and the end of the critical phase of COVID-19, which contradicts conventional theory and previous studies that suggest an association between sitting time and academic performance. The present study also shows a similarity between the time spent practicing physical activity and academic performance.