An empirical study on the improvement of students’ physical fitness and health by college physical education programs based on the background of big data

True health is not the absence of disease, but the maintenance of the best state in physical, mental and social adaptation [1-2]. Under the strategy of education and health, health education has been an important part of realizing “five education”, and it is highly similar to physical education curriculum in terms of objectives, contents, etc. The integration of health education into physical education classroom is not only conducive to the healthy growth of students and stimulate their interest in learning, but also provides a new development idea for the reform of physical education curriculum of colleges and universities [3-5]. The realization of health education into physical education classroom is not only conducive to students’ healthy growth and stimulate students’ interest in learning, but also provides a new idea for the reform of physical education curriculum in colleges and universities.

At present, physical health is an important indicator of talent selection for social employers, and higher education institutions are important channels for the output of social applied first-line talents, and the university physical education program needs our further thinking and speculation in terms of how to better cultivate talents’ healthy living habits as well as to improve the level of talents’ physical fitness and health [6-7]. University physical education should gradually adapt to the requirements of the times from educational concepts, educational ideas to teaching content, mode, means and methods [8].

Moreover, as the future pillars of the country, the health condition of college students has been deeply concerned by all walks of life. Due to academic, family, emotional and other aspects of the pressure can not be timely catharsis, students suffering from depression, autism, the situation is endless [9-10]. At the same time, the improvement of the quality of life makes students neglect exercise, and obesity, myopia, hypoglycemia and so on become common diseases, which hinder the healthy growth of students. Therefore, in order to reduce the occurrence of students’ risky behaviors and improve students’ physical and mental health, colleges and universities should be committed to infiltrating health education into physical education teaching, not only teaching theoretical knowledge, but also focusing on the combination of theory and practice in the physical education curriculum, to ensure the unity of students’ consciousness and behavior, and to promote the development of students’ healthy behavior habits [11-13].

Quality education emphasizes the cultivation of moral, intellectual, physical and aesthetic comprehensive development of students, advocating that teachers in the development of students’ intelligence at the same time should pay attention to students’ physical fitness, healthy physique is the prerequisite and foundation of everything. In such a background, college physical education teachers must timely adjust the traditional ideological concepts to enhance the physical health of young people oriented curriculum reform, give full play to the nurturing value of physical education courses [14-16]. However, in terms of the current implementation of college physical education curriculum teaching, it is still far from meeting the requirements of quality education, and plays a limited role in helping the development of adolescent physical health, so how to study the reform of physical education curriculum based on the physical health of adolescents is very necessary [17-19].

The participation of students in sports has a positive significance for students’ physical health, but it is necessary to conduct relevant research on how sports affect students physically and mentally and what kind of impact they have. Literature [20] reveals that the construction of school sports facilities significantly affects students’ future participation in sports, based on a follow-up survey of students’ entry into the school learning process. Literature [21] combined health questionnaires, quality of life scales, and functional activity summary scales to reveal the associations between the health status of American adolescents during the cessation of physical activity in the context of the epidemic and factors such as gender, grade, and type of physical activity participation, which provide important references for health professionals in their health policy development. Literature [22] empirically examined the impact of college students’ sports participation and found that college students’ lifelong participation in sports effectively promoted their physical and mental health, and increased their self-esteem and well-being levels. Literature [23] investigated the demographic characteristics of sport participants online and combined independent t-tests, one-way ANOVA, and multifactorial ANOVA methods in a comparative study, noting that recreational sport participants engaged in many indoor sports for health prevention during the outbreak, and concluded that during the outbreak, joint efforts of individuals and organizations were needed to promote the maintenance of exercise habits and avoid social isolation. Literature [24] longitudinally assessed how physical activity affects bone health using a dual-energy x-ray absorptiometry (DXA) tool and found that physical activity promoted significant increases in leg BMD and BMC. Literature [25] systematically reviewed studies related to high-intensity interval training (HIIT) to understand the impact and influence of HIIT on human physical health, and synthesized the results of the studies to learn that HIIT has a positive effect on cardiorespiratory, blood glucose, and vascular function in humans. Literature [26] searched journal papers related to physical activity and the New Crown epidemic for integrated analysis, in which more studies exploring the epidemic’s impact on physical activity levels, followed by the epidemic’s impact on mental health, and noted that physical activity levels in the context of the epidemic declined sharply, but that exercise was effective in mitigating mental health disorders in the context of the epidemic.

Physical education helps to improve students’ physical fitness and motor skills, so it is very meaningful to optimize the teaching around physical education, and the optimization of physical education involves teaching content, teaching methods and the introduction of intelligent information technology. Literature [27] conducted a controlled experimental class in physical education and found that after applying the team game tournament (tgt) strategy to a twelve-week basketball physical education class, students’ motivation to learn physical education increased significantly, but there was no significant improvement in the level of athletic skills. Literature [28] envisioned a motor posture recognition model with quaternion algorithm as the core logic to assist physical education teaching, and conducted simulations through experiments to verify that the proposed model can support classrooms to grasp the level of students’ motor skill specification and effectively ensure the effectiveness of teaching. Literature [29] describes the development of artificial intelligence and virtual reality technology to promote the construction of informationization and intelligence in education, and reviews the research literature on these technologies in the field of education to deepen the understanding of these information technologies in educational practice, especially in physical education subjects. Literature [30] used a hierarchical evaluation strategy to identify the elements of the indicators in the evaluation system and assessed the role of multimedia teaching platforms in physical education, and the results of the evaluation revealed that multimedia teaching platforms play an important role in promoting the effectiveness of physical education teaching in colleges and universities. Literature [31] introduced a virtual reality-assisted convolutional neural network (VRA-CNN) in the Mobile Cloud Computing (MCC) e-learning platform, which strengthened the intensity of the sensor communication, and at the same time provided a real-time interactive environment for sports training for people with disabilities, and finally verified it using experimental methods, which pointed out that the introduction of VRA-CNN method effectively improved the athletes’ sports knowledge and sports confidence. Literature [32] used the Physical Activity Self-Efficacy Scale to explore the factors affecting college students’ sports participation, and suggested that university administrators need to work together with health professionals to promote college students’ sports habits and sustainable participation in lifelong sports.

In this paper, a method for classifying college students’ physical fitness and health data based on unsupervised learning is proposed. The method uses college students’ test data of BMI, lung capacity, body forward flexion, standing long jump, 50-meter run, 1000-meter run (male) or 800-meter run (female), pull-ups (male) or sit-ups (female) as the input data, and through the K-medoids clustering algorithm, the students with similar physical fitness are classified together, and the teacher develops targeted and individualized Teaching content. An empirical study of the changes in students’ physical fitness and health was conducted through a before-and-after controlled experiment.

Due to the large number of college students and the lack of any regularity in the physical fitness test data, which makes it difficult to find out the pattern from it, it is difficult for the teacher to design a personalized exercise program for each student, so as to efficiently and scientifically improve the students’ physical fitness. In this paper, the K-medoids clustering algorithm is utilized to identify students with similar physical fitness and classify their physical fitness characteristics. Teachers based on these data can improve the overall teaching direction of the class and the individual student’s personalized teaching guidance content, consider in advance the teaching strategies and pathways suitable for students in the class, and adapt to the personalized differences that exist between students [33]. For example, badminton and volleyball, which are widely implemented in colleges and universities and are highly popular among students, have the advantages of not requiring high comprehensive personal qualities, little difficulty in starting, and no special restrictions on the selection of venues, which can effectively mobilize students’ interest in learning and participation. In carrying out the badminton course, teachers in teaching students, according to the specific situation of the students were arranged teaching tasks, such as better physical quality and has mastered the swing, grip, serve and other technology students, you can appropriately increase the practice of running and jumping, blocking the ball, swinging and other movements of the movement of a larger amount of action. If the physical quality of students is general or they have not mastered the badminton operation skills, you can practice the simple serve, turn the ball, catch the ball, and other actions.Adopting such differentiated teaching methods can efficiently and scientifically improve the physical quality of students.

2.2

Classification of students’ physical health based on K-medoids

Assuming that the set of data samples of students’ physical health is X = {X_i = (x_i1,x_i2,⋯,x_im),i = 1,2,⋯,N}, the flowchart of K-medoids synthesis algorithm is shown in Fig. 1.

Figure 1.

Flowchart of K-medoids synthesis algorithm

The realization steps are as follows:

1) Initialize the data samples related to students’ physical fitness and health, select m data sample from the data sample set and set the maximum number of iterations, m data samples are used as the initial centroids, and the m initial centroids can be expressed as (M₁,M₂,⋯,M_m).

2) X_i and X_j are two data samples in the data sample set, and assuming that d_ij is the weighted Euclidean distance between X_i and X_j, it can be expressed as: (12) dij=||p(Xi−Xj)||2=∑k=1mp(xik−xjk)2\[{{d}_{ij}}=||p({{X}_{i}}-{{X}_{j}})|{{|}^{2}}=\sqrt{\sum\limits_{k=1}^{m}{p}{{({{x}_{ik}}-{{x}_{jk}})}^{2}}}\]

Where p represents the weighting factor, the value of which is determined based on the different differences in the contribution of each component in the clustering.

The clustering radius can be expressed using r. Since there is a certain distance between X_i and X_j, the number of missing information PQ on the straight line distance from data sample X_i to data sample X_j at the moment of t can be expressed using τ_ij. When τ_ij(0) = 0, it means that the number of missing information PQ on each straight line path at the initial moment is the same and all are 0. From this, the formula for the number of missing information PQ on the straight line distance between X_i and X_j can be expressed as: (13) τij(t)={ 1,dij≤r0,dij>r\[{{\tau }_{ij}}(t)=\left\{ \begin{array}{*{35}{l}} 1, & {{d}_{ij}}\le r \\ 0, & {{d}_{ij}}>r \\ \end{array} \right.\]

And whether X_i is subsumed into the X_j field yields equation (14): (14) pij(t)=τijα(t)ηijβ(t)∑s∈Sτsjα(t)ηsjβ(t)\[{{p}_{ij}}(t)=\frac{\tau _{ij}^{\alpha }(t)\eta _{ij}^{\beta }(t)}{\sum\limits_{s\in S}{\tau _{sj}^{\alpha }}(t)\eta _{sj}^{\beta }(t)}\]

Where: p_i(t) represents the probability of X_i subsuming to X_j, and when p_i(t) ≥ p₀(0 ≤ p₀ ≤ 1), then X_i subsuming to X_j in the domain, and p₀ represents only a probability constant. Let S = {X,|d_ij ≤ r,s = 1,2,⋯,j,j+1,⋯,N}, η represent the local heuristic function, i.e., the expected degree of data sample X_i moving to data sample X_j. α,β represent the role of the characteristic information accumulated during the multi-factor move in the process of the sample move path, and the role of the heuristic function in the process of the sample move path, respectively.

In summary, the process of determining the optimal path and clustering center between multiple data samples is as follows: Firstly, the distance d_ij between each data sample is calculated using the weighted Euclidean method, and the probability p_ij(t) that a data sample is subsumed into another data sample and the number of missing information PQ is derived from equation (14), which achieves the determination of optimal paths and clustering centers between multiple data samples, which is the historical optimal position of the data samples.

3) Reclustering and analyzing the historical optimal positions of the data samples based on the K-medoids algorithm. Using the historical optimal position of the ACO algorithm as the representative object in the K-medoids algorithm O_j, determine the clusters in which each data sample is located as well as the centroids between the classes.

4) For the new data sample set formed follow the method of step 2), calculate the optimal solution represented by each data sample and update the historical optimal position and global optimal solution of the data sample set.

5) Recalculate the weighted Euclidean distance d_ij between any two data samples to determine the new clustering center and find the optimal path.

6) Set D_i to be the deviation error for the jth clustering and ε to represent the clustering, then analyze the overall error Eq: (15) Dj=∑l=1J∑k=1m(xlk−cjk)2\[{{D}_{j}}=\sum\limits_{l=1}^{J}{\sqrt{\sum\limits_{k=1}^{m}{{{({{x}_{lk}}-{{c}_{jk}})}^{2}}}}}\] (16) ε=∑j=1kDj\[\varepsilon =\sum\limits_{j=1}^{k}{{{D}_{j}}}\]

Where c_jk denotes the krd component of the jnd clustering center.

At this time to determine whether the clustering ε is within the specified range, such as within the specified range, the clustering stops, if not within the specified range, it is necessary to turn to step 3) to continue the iteration.

7) Termination conditions are reached, the clustering ends, and the optimal clustering center is obtained.

The study proposes a method for identifying college students’ physical health based on the K-medoids algorithm, which clusters college students’ physical health categories and then launches highly targeted personalized course teaching. The physical health data of 600 juniors in a college of University A in the last semester of 2023 were clustered, and male and female students were clustered into 10 different physical health categories, and different teaching contents were adopted for different categories of students. After the curriculum reform, male students had a highly significant difference in standing long jump (P<0.01) and significant differences in BMI, lung capacity, 50 meters, 1000 meters, and pull-ups (P<0.05). Female students had a decrease in BMI mean, significant difference (P<0.05) in lung capacity, standing long jump, sitting forward bend, 800 meters. The total physical fitness scores of male and female students increased by a total of 12.09% and 14.03% in the excellent rate and good rate, respectively, compared with the pre-curriculum reform. The results of the above empirical analyses indicate that students’ physical fitness can be significantly improved after reforming the college physical education curriculum using big data technology.