Exploration and Practice of the Relationship between College Students' Learning Adaptation and Mental Health under the Information-Based Teaching Environment of Potential Profile Analysis

In order to further investigate and understand the relationship between college students' learning adaptation process and mental health in the learning process under the information-based teaching environment, this paper makes a questionnaire survey on college students' learning adaptation and mental health and selects 408 college students as the research object. The results show that the potential profile analysis shows that with the development of network and informatization, the learning adaptability of college students can be divided into four types: troubled group (accounting for 7.598%), marginal adaptive group (accounting for 42.892%), maladaptive group (accounting for 4.167%), and good adaptive group (accounting for 45.343%). The mental health level of the latter two is better, and the mental health level of the first two is worse. Students who do not adapt to learning and students who adapt well show common characteristics. Most of them are only children, and their parents have a high level of education. This kind of family often has good material conditions and loose family atmosphere, which will also form a protective mechanism for students' mental health, making students have good self-acceptance ability and good mental health level.


Introduction
Te rapid development of network and information technology has promoted the development of many industries, especially the rapid development of information technology, which provides more technical choices and mode innovation drivers for the improvement of modern teaching methods and teaching quality. In the information-based teaching environment, students' learning process is no longer limited by the "time and space" of traditional education, and information technology can provide students with more personalized learning methods and learning contents. Compared with traditional face-to-face teaching and classroom teaching, the transformation of teaching methods will also require students to constantly adjust their learning methods. Teachers need to change, and students also need to change [1]. However, the level of students' learning adaptability directly afects the learning efect. Some students will have maladjustment in the face of the environmental changes brought by the information environment, while some students will have strong interest in learning and can adapt well due to the novel learning methods, and diferent adaptation results will even afect students' mental health and indirectly afect their learning status, which is also the focus of this study.
Its main factors include learning attitude, learning technology, learning environment, and physical and mental health. At the same time, some scholars have defned learning adaptation. For example, Kelzang and Lhendup believe that learning adaptability refers to the ability tendency of students to adjust themselves to the learning environment in the learning process [3]. Xiao believes that learning adaptability is "when the learning environment, learning objects, and contents around individuals change, individuals take the initiative to overcome difculties and change themselves in order to avoid the decline of learning efciency, so as to achieve good learning efciency" [4]. Kim et al. defne learning adaptability as the ability of individuals to actively adjust their learning motivation and behavior, improve their learning ability, coordinate their learning psychology and behavior with the changing learning conditions, and achieve good learning achievements according to the changes of internal and external learning conditions and their own learning needs [5]. Jeong believes that learning adaptability is the psychological ability of individuals to actively react to the surrounding environment with certain behaviors and actions in the process of interaction between individuals and learning environment [6]. Mahasneh believes that learning adaptation is a psychological and behavioral process in which the subject tries to adjust himself according to the environment and learning needs, so as to achieve a balance with the learning environment. Chinese scholars have put forward their own diferent views and views on the concept of learning adaptation from their respective perspectives [7].
In view of the unique characteristics of Chinese college students' learning adaptability, some scholars have developed a scale suitable for measuring Chinese college students' learning adaptability. For example, based on the "learning adaptability test (middle school version)," the "college students' learning adaptability questionnaire" is compiled according to the age and learning characteristics of college students. On the basis of previous studies, a set of learning adaptability scale for college students is compiled. Te questionnaire is divided into fve dimensions: learning motivation adaptation, learning ability adaptation, learning environment adaptation, education style adaptation, and physical and mental health adaptation. Tere are 50 items in this questionnaire. Te three-level scoring method is adopted. Te higher the score, the better the learning adaptability. Te content of the questionnaire includes two dimensions of learning motivation and learning adaptive learning behavior, as well as eight factors such as professional interest, autonomous learning, stress response, method application, helpseeking behavior, environmental choice, information utilization, and knowledge application, with a full score of fve points [8,9].
At present, the research on the learning adaptability of college students in China is still in the exploratory stage compared with the research on the learning adaptability of primary and secondary school students. Te research on college students mostly focuses on the investigation and research on the learning adaptability of college freshmen. Te existing literature shows that the research content of college students' learning adaptability in China also focuses on two aspects. One is the discussion of the factors afecting college students' learning adaptability. Te other is that Chinese scholars have compiled some learning adaptability questionnaires suitable for Chinese college students according to their own characteristics. Scholars in China have conducted some investigation and research on the factors afecting college students' learning adaptability, as shown in Figure 1, which shows the common confusion of college students [10]. Trough the investigation, it is found that college students' learning adaptability is mainly afected by many factors, such as students' own cognitive evaluation, personality characteristics, school environment, family education methods, and social support. In the aspect of cognitive evaluation, this paper studies the relationship between higher vocational college students' self-concept, coping style, and learning adaptability. Te conclusion is that the total score of self-concept is signifcantly negatively correlated with the overall learning adaptation score, and the mature coping style is highly signifcantly positively correlated with the overall learning adaptation. Immature coping style has a very signifcant negative correlation with overall learning adaptation [11]. Some scholars have concluded that there is a close relationship between self-harmony and learning adaptability of normal college students. Te research shows that there is a signifcant positive correlation between college students' learning adaptability and learning self-efcacy and the total score of career commitment [12].

Bayesian Analysis of Latent Variable
Model under Random Effect 3.1. Random Efect Model. Te random efect model is as follows: where and y i , x i are datasets, which are recorded as follows: Y � y 1 , y 2 , . . . , y n T , Generally, it follows the normal distribution with mean value of 0 and standard deviation of σ u . In mean regression, it is generally assumed that the error term ε i follows the error term with mean value of 0 and variance of σ 2 ε . Here, we consider quantile regression, so it is assumed that ε i follows the Laplace distribution with unknown parameter of 0 and scale parameter of σ, that is, Its density function is as follows: 2 Journal of Environmental and Public Health where the form of 0 < p < 1; ρ p (ε i ) is as follows: Ten, the likelihood function of the random efect model can be written as Tis method needs to extract samples from a posteriori distribution [θ|Y, X] and then obtain an empirical distribution close to the real distribution according to the samples, which can characterize the characteristics of a posteriori distribution.
Assume that the a priori distribution of β is as follows: where b is the column vector of q dimension; when the observation values Y and X are given, the diagonal matrix of order q + 1 of formula V is as follows: Te posterior distribution of β is obtained. If the prior and posterior distributions obey the same type of distribution, this prior distribution is called conjugate prior distribution. Assume a priori distribution of U: where the column vector of n dimension of formula b u and the unit diagonal matrix of order n of formula I are composed of conditional distribution: After obtaining the prior distribution of U, there is the following process: Suppose the conjugate a priori distribution of σ −2 u is as follows:  Journal of Environmental and Public Health where a 1 , r 1 are superparameters, which can be obtained from the conditional distribution: If the hyperparameters in the conjugate a priori distribution are unknown, they should be regarded as unknown parameters with a priori distribution. However, these prior distributions also have their own superparameters, so it will be difcult to extract samples. Terefore, for the convenience of research, the hyperparameters in the conjugate a priori distribution are set to known values.
Te simulation of the random efect model is as follows: Adiagonal matrix in which the diagonal elements of the formula V are 0.5 and q + 1 dimensions.
I is an n-order unit matrix, and σ u � 0.5. First, 50 sets of simulation datasets with a sample size of n � 100 are produced. In this example, we need to estimate β, σ, U, σ u , iterate 10000 times, and do 50 repeated calculations. Te EPSR values of the parameters are close to 1, which are 0.968, 0.984, 0.975, and 0.989, respectively, indicating that the simulation process converges. Terefore, the frst 9000 times are discarded and the last 1000 times are taken as the simulation results. Figures 2-5 represent the straight line diagram under diferent quantiles corresponding to diferent beta values.
Ten, 50 groups of simulation datasets with sample size of n = 30 and 50100 are generated, and 50 repeated calculations are made, respectively. Assuming q = 4, the calculation model is iterated 10000 times each time, the frst 9000 times are discarded, and the last 1000 times are taken as the simulation results. Considering the diference of samples and the deviation of calculated estimates, the root mean square between the actual value and the corresponding estimated value is defned as follows: Te following results are given. Tables 1-3 reveal the true value, estimated value, deviation, and root mean square of parameters under diferent samples.
From the results of deviation and root mean square, it can be seen that even in the case of small samples (n = 30), the deviation value is acceptable because more than 80% of the deviation value is less than 0.05, that is, the efective value is about 95%. In terms of root mean square, with the increase of sample size, the root mean square of σ u and β 0 decreases uniformly, indicating that the sample size has an impact on the estimation results. To sum up, all the results can show that Bayesian estimation is also close to the real value in the case of small samples, which is consistent with the theoretical results.

Relationship between Students' Learning Adaptation and Mental Health in Information-Based Teaching Environment.
Similar to the defnition of mental health, diferent researchers have diferent opinions on the defnition of mental health standards. Maslow believes that a person with the personality characteristics of self-expression is a person with mental health. Drawing on the achievements of researchers at home and abroad, this paper summarizes six standards of mental health: a correct understanding of reality; selfknowledge, self-esteem, and self-acceptance;self-regulation ability [13,14]; the ability to establish close relationships with people; stability and coordination of lattice structure; and life enthusiasm and work efciency [15].
Adaptability has always been considered to be closely related to the level of mental health. Kusuma et al. even believe that adaptation to life is one of the constituent elements of mental health [16]. Learning adaptation will not only have a direct impact on students' academic performance but also afect students' psychological development and mental health level. Research shows that there is a signifcant positive correlation between learning adaptability and mental health. Te mental health level of students with high learning adaptability is signifcantly better than that of students with low learning adaptability [17,18].

Research
Methods. 408 students and junior middle school students from the two places were randomly selected to investigate the students' learning situation under the information-based teaching environment (including the basic situation survey, the students' learning adaptability questionnaire, and the mental health questionnaire under the information-based teaching environment) through the network platform developed by the Department of Educational Technology in the University. All students surf the Internet through the computer room of their school and complete the questionnaire anonymously on the Internet. Tere are 133 primary school students (21 in grade 2, 59 in grade 3, and 53 in grade 4) and 275 junior middle school students (220 in grade 1 and 55 in grade 2), with an average age of 11.75 years; 214 boys and 194 girls; 210 in Beijing and 198 in Guangdong; 382 Han people and 26 ethnic minorities; 227 only children and 181 non-only children; father's education level: 4.9% in primary school, 50.5% in middle school, 32.6% in university, and 12% in master's degree or above; and mother's education level: 8.6% in primary school, 47.1% in middle school, 32.8% in university, and 11.5% in master's degree or above.

Research Tools.
Based on the self-regulated learning theory, we propose that students' learning adaptation in the information-based teaching environment includes fve dimensions: learning motivation, information acquisition methods, metacognitive strategies, knowledge acquisition, and knowledge expansion [19]. Learning motivation refers to the internal motivation that directly promotes students' learning; information acquisition means the ability to obtain    Journal of Environmental and Public Health learning resources through information technology; metacognitive strategy refers to the strategy of controlling the process of information and monitoring and guiding the cognitive process; knowledge acquisition and knowledge expansion focus on students' ability to understand and apply knowledge at diferent stages of the learning process [20]. Te questionnaire on students' learning adaptability in the information-based teaching environment (initial version) includes 40 questions, including 8 questions on learning motivation, 6 questions on information acquisition, 8 questions on metacognitive strategies, 12 questions on knowledge acquisition, and 6 questions on knowledge expansion. Using a 4-point score, the subjects are required to respond to the degree of conformity between the situation presented by the questionnaire and themselves according to their actual situation in the past week. Te degree of conformity is as follows: very inconsistent (1), relatively inconsistent (2), relatively consistent (3), and very consistent (4). Te Mplus robust maximum probability method (MLR) was used for confrmatory factor analysis. According to the correction index, delete the questions with higher correction index. Te subscales of the fnal questionnaire are 6 questions of learning motivation dimension, 4 questions of information acquisition method dimension, 5 questions of metacognitive strategy dimension, 6 questions of knowledge acquisition dimension, and 5 questions of knowledge expansion dimension. χ 2 value is 810.01, DF is 289, CFI is 0.94, TLI is 0.94, SRMR is 0.03, and RMSEA (90% CI) is 0.069 (0.063, 0.074). Te correlation between each topic and the total score is greater than 0.35, indicating that the topic has a good discrimination. Te homogeneity reliability (α coefcient) of the total scale is 0.80, and the homogeneity reliability (α coefcient) of each subscale is more than 0.82. In short, the self-made questionnaire on students' learning adaptability in the information-based teaching environment has good reliability and validity [21,22].

Self-Compiled Student Mental Health Questionnaire.
Trough literature review, expert discussion, and interviews with primary and secondary school students, this paper summarizes fve dimensions of primary and secondary school students' mental health: learning, self, society, emotion, and behavior. Learning refers to the ability of selfregulated learning, including the development of attention, critical thinking, and creative thinking; self includes selfconcept, self-evaluation, and self-regulation; society mainly investigates interpersonal relationships, including parentchild, teacher-student, and classmate relationships; emotion mainly refers to the ability of emotion regulation [23]; behavior refers to students' behavior problems such as aggression, hyperactivity, and violation of discipline. Accordingly, we have preliminarily formed a student mental health questionnaire under the information-based teaching environment, with a total of 119 questions, which adopts a 5point score (0-4). Subjects were asked to answer the severity of the questions in the questionnaire within one week according to their physical and psychological conditions. 0 means no, and 4 means serious. Te higher the total score of the questionnaire, the more unhealthy the psychology is [24]. Te results of exploratory and confrmatory factor analysis were not ideal. After expert discussion, the questionnaire was modifed and a 67-question questionnaire was obtained, including 15 questions on learning dimension, 10 questions on self-dimension, 14 questions on social dimension, 13 questions on emotional dimension, and 15 questions on behavioral dimension. 613 pupils and junior middle school students were tested for the second time, and Mplus7 robust maximum probability method (MLR) was used for confrmatory factor analysis. According to the correction index, delete the questions with higher correction index. Te subscales of the fnal questionnaire are 9 questions of learning dimension, 8 questions of self-dimension, 8 questions of social dimension, 7 questions of emotional dimension, and 15 questions of behavior dimension. χ 2 value is 1242.79, DF is 726, CFI is 0.91, TLI is 0.90, SRMR is 0.05, and RMSEA (90% CI) is 0.034 (0.031, 0.037). Te discrimination of the questions is good, and the correlation between each question and the total score is greater than 0.50. Te homogeneity reliability (α coefcient) of the total scale is 0.96, and the homogeneity reliability (α coefcient) of each subscale is more than 0.75. In short, the self-made student mental health questionnaire has good reliability and validity and meets the requirements of psychometrics. Considering the intersection between the learning dimension and learning adaptability in the mental health questionnaire, we removed the learning dimension when analyzing the relationship between learning adaptability and mental health [25].

Result Analysis.
Mplus7.0 software and SPSS16.0 software programs were used for data processing and statistical analysis. Using potential profle analysis, this paper discusses the potential categories of students' learning adaptation in the information-based teaching environment composed of fve dimensions. Analysis of variance was used to explore the diferences of students with diferent learning adaptation categories in each dimension and total score of mental health.

Descriptive Statistics of Students' Learning Adaptability and Mental
Health. Te descriptive statistical results are shown in Table 4. Te analysis shows that there is a positive correlation between the dimensions and total scores of the learning adaptability questionnaire and the dimensions and total scores of the mental health questionnaire. Tere is a negative correlation between the dimensions and total scores of the learning adaptability questionnaire and the dimensions and total scores of mental health, and ps < 0.01. In addition, the average scores of each dimension and total score of the learning adaptability questionnaire are greater than 3 points. Te highest and lowest dimensions are knowledge acquisition dimension (3.370) and knowledge expansion dimension (3.122), respectively. Te scores of each dimension and total score of mental health questionnaire are mainly below 0.5, and the highest and lowest dimensions are social dimension (0.435) and emotional dimension (0.359), respectively.

Potential Profle Analysis of Students' Learning
Adaptability. Taking the scores of students in the fve dimensions of learning adaptability questionnaire (learning motivation, information acquisition methods, metacognitive strategies, knowledge acquisition, and knowledge expansion) as explicit variables, the potential profle model is established. Te ftting indexes of potential profle models with diferent categories are shown in Table 5. It is found that the AIC, BIC, and AIBC indexes of the model decrease gradually with the increase of the number of categories. Te change range of the three indicators is divided by the number of categories "4." Te change range of the indicators of the frst three classifcation models is large, and the change range of the indicators of the latter three types of models tends to be fat, indicating that with the increase of the number of model classifcations, the optimization degree of the latter model gradually decreases compared with the former model. In addition, the entropy value of all category models is greater than 0.94, showing a good degree of model ftting, and the model with the number of categories from 2 to 4 is better (category 2: 0.989; category 3: 0.957; category 4: 0.979). From the perspective of LMRT, the LMRT of the models with four classifcation numbers from category 2 to category 5 reached a signifcant level (ps < 0.01). Considering the above ftting indexes and referring to the simplicity and actual situation of the model, four types of models are fnally selected as our potential profle analysis model. Table 6 shows the distribution of the number of people in the four potential categories measured by the learning adaptability questionnaire and the Z scores of the corresponding categories in each dimension and total score of the learning adaptability questionnaire.
Among all categories of people, category 1 has the least number, accounting for only 4.2% of the total number, followed by category 2 (7.6%), and categories 3 and 4 account for a large proportion, accounting for 42.9% and 45.3%, respectively. Figure 6 describes the average scores of the four categories of people in each dimension and total score of the learning adaptability questionnaire.
As shown in the fgure, the distribution of the four categories in each dimension and total score is relatively consistent, without too much fuctuation. Te frst group has the lowest score in all dimensions and total scores and is named "maladjustment group"; the second group was named "troubled group"; the score of the third year group is second only to the fourth group, which is in the middle of the population and is named "marginal adaptation group"; the score of group 4 is the highest in all dimensions and total scores, indicating the best learning adaptability. It is named "good adaptation group." Further analysis of variance on the scores and total scores of diferent categories of people in each dimension of the learning adaptability questionnaire found that there were signifcant diferences in the scores and total scores of each dimension among the four categories (ps < 0.01), which also verifed the efectiveness of the four-category model of potential profle analysis from another side. In order to explore the diferences in demographic information among the four groups, we counted the distribution proportion of the four groups in terms of "gender," "whether they are the only child," "ethnicity," "father's education level," and "mother's education level" (Table 7) and conducted a chi square test.
Te results showed that there were signifcant diferences in the proportion of categories in the variables of "whether they are the only child," "father's education level," and "mother's education level" (ps < 0.01), but there were no signifcant diferences in the proportion of categories in the variables of "gender" (P � 0.057) and "ethnicity" (P � 0.604). Further analysis of the variables with signifcant chi square test shows that in the category proportion composition of the variable "whether it is an only child," the proportion of "only child" and "not only child" in the frst group ("maladaptive group") and the fourth group ("good adaptation group") is quite diferent, and the proportion of "only child" is signifcantly higher than that of "not only child." In terms of the category proportion composition of the variable "father's education level," we found that the proportion of fathers with higher education (graduate and above) in category 1 ("maladjustment group") and category 4 ("good adaptation group") was signifcantly higher than that in category 2 ("troubled group") and category 3 ("marginal adaptation group"). But at the same time, it is interesting to note that there is an obvious "average" trend in the proportion of fathers' education level in the frst group ("maladjustment group"), and the proportion of fathers with "primary school" education level in this group is signifcantly higher than that in the last three groups. Similarly, in terms of the category proportion of the variable "mother's education level," the proportion of mothers with higher education (graduate students and above) in category 1 ("maladjustment group") and category 4 ("good adaptation group") is signifcantly higher than that in category 2 ("troubled group") and category 3 ("marginal adaptation group"). Generally speaking, the "maladjustment group" and "good adaptation group" are more likely to be the only child, and the educational level of their parents is mostly highly educated (graduate students and above), but at the same time, compared  Journal of Environmental and Public Health with the "good adaptation group," the proportion of their parents with "primary school" academic experience is also higher.

Diferences in Mental Health of Students with Diferent
Types of Learning Adaptation. In order to explore the diferences in the scores of various dimensions and total scores of mental health among diferent categories of students, one-way ANOVA was carried out. Te results showed that there were signifcant diferences in the scores of all dimensions and total scores of mental health among diferent categories of students (ps < 0.01). Te posttest found that there were signifcant diferences in the scores and total scores of each dimension between group 2 ("troubled group") and the group 4 ("good adaptation group") and between group 3 ("marginal adaptation group") and group 4 ("good adaptation group") (ps < 0.01). At the same time, except for the selfdimension, there were signifcant diferences in the scores and total scores of other dimensions between the frst group ("maladaptive group") and the second group ("troubled group") (P social, behavioral and total scores <0.05, P emotional <0.01). In addition to self and behavior dimensions, there were also signifcant diferences between group 2 ("troubled group") and group 3 ("marginal adaptation group") (P social <0.01, P emotional and total score <0.05 "(P social, behavioral and total scores <0.05, P emotional <0.01)" and "(P social <0.01, P emotional and total score <0.05)" for correctness." In general, the mental health level of group 2 ("troubled group") is the worst, and that of group 4 ("well adapted group") is the best. Comparing the scores of all dimensions and total scores of various groups in the learning adaptability questionnaire, it can be found that when the learning adaptability level is in the middle, the mental health status is poor, while when the learning adaptability is in the two poles (best and worst), the mental health level is better.

Discussion.
Tis study found that there were signifcant diferences in the scores of all dimensions and total scores of mental health among students of diferent potential categories. Te mental health level of the troubled group was the worst, and the mental health level of the well adapted group was the best. When students are at the intermediate level of learning adaptability, their mental health is poor, while when students' learning adaptability is at the two poles (best and worst), their mental health is better. Specifcally, when students encounter difculties in learning adaptation and marginal adaptation, they will feel the greatest learning pressure, which will afect students' self-identity, social communication, and emotional stability and increase the probability of students' problem behavior. When students are extremely unft for the information-based learning process, they will avoid and maintain themselves. On the contrary, they will have a better mental health level as well as become well adapted students. In the above, we found that the proportion of only child students who are not well educated is higher than that of their parents. Tis kind of family often has good material conditions and loose family atmosphere, which will also form a protective mechanism  for students' mental health, making students have good selfacceptance ability and good mental health level.

Conclusion and Discussion
Trough the potential profle analysis, we fnd that under the information-based teaching environment, primary and secondary school students can be divided into four subgroups according to learning adaptation: non-adaptation group, distress group, marginal adaptation group, and good adaptation group. Tere are signifcant diferences in the scores and total scores of all dimensions of learning adaptation (learning motivation, information acquisition methods, metacognitive strategies, knowledge acquisition, and knowledge expansion) among the students of the four groups, showing the gradual rise of adaptation level, which shows that this objective classifcation method is accurate and efective. Te study also found that the proportion of only children is higher among well adapted and maladaptive students, and the educational level of their parents is mostly highly educated (graduate students and above), revealing that only children and students with highly educated parents are prone to polarization in learning adaptability. Among the four types of students, 45.3% are in the good adaptation group and 42.9% are in the marginal adaptation group, which is consistent with our observation. It shows that most students can better deal with the informatization of teaching and learning tools and will try to interact with technology to complete the learning process. Tese students have high learning motivation, can efectively use various learning strategies, can obtain learning resources, and can understand and use knowledge, so as to achieve better learning results. Te study also found that 7.6% of students are in the troubled group and 4.2% are in the maladjustment group. Although the number of these two types of students is small, they are at a low level in all aspects of learning adaptation, which may lead to their poor learning. In practical work, teachers can use the potential profle analysis method to identify students with learning maladjustment and build a more adaptive learning environment to guide and adjust these students from the aspects of motivation, strategy, information acquisition mode, knowledge mastery, and application, so as to help them get rid of learning difculties.

Data Availability
Te labeled dataset used to support the fndings of this study is available from the corresponding author upon request.

Conflicts of Interest
Te author declares that there are no conficts of interest.