Intelligent learning methods for improving student performance through virtual reality technology

Abstract

With the rapid development of information technology, the field of education is undergoing a profound change, in which intelligent hybrid learning and virtual reality technology are increasingly valued. This study proposes an intelligent hybrid learning method based on virtual reality for student performance improvement. This paper reveals the limitations of traditional learning methods in meeting the needs of modern education, and expounds the theoretical basis of intelligent hybrid learning and virtual reality technology. This paper collects and processes a large amount of learning data, based on which a new model of student learning performance prediction is established. The verification results of the model show that the model in this study has excellent performance in predicting students’ learning performance. This paper gives some suggestions for future educational practice and research. In general, this study provides a new learning method for the field of education and has important reference value for education reform and teaching practice.

Keywords

Intelligent blended learning virtual reality technology student learning performance

1. Introduction

In modern society, the development of educational technology and the reform of teaching methods have aroused wide concern among educators and researchers. The traditional teaching mode often neglects the difference of individual students, and the efficiency of information transmission and skill training is not high. With the development of science and technology, especially the rise of artificial intelligence and virtual reality technology, more efficient and personalized teaching methods have gradually emerged.

Intelligent blended learning, as a new teaching mode, combines the flexibility of online teaching with the interaction of traditional face-to-face teaching. Through big data and artificial intelligence technology, real-time monitoring and personalized recommendation of students’ learning process can be realized, which greatly improves the teaching effect. Virtual reality technology, with its immersive experience and powerful simulation ability, provides a new learning environment. Students can participate in the learning process in a virtual environment, which greatly improves the fun and effectiveness of learning.

Although there have been some studies on the application of intelligent blended learning and virtual reality in education, but how to organically combine the two to play a greater teaching effect has not yet formed a systematic study. Therefore, this paper aims to explore intelligent blended learning methods based on virtual reality in order to improve students’ learning performance

The purpose of this study is to explore and design an intelligent hybrid learning method based on virtual reality, which aims to solve the problems existing in traditional learning methods and improve the learning effect of students. The research program combines the immersion and interactivity of virtual reality with the flexibility and personalization of intelligent blended learning to create a completely new and effective learning environment. In order to quantify the effect of this new approach, relevant data will be collected and analyzed to build a model that will provide a visual representation of student learning outcomes. At the same time, the model will be verified in detail to ensure that it can accurately reflect the influence of learning methods on students’ learning results. Based on the research results, suggestions for future educational practice and related research are proposed to enable educators and researchers to better understand and apply virtual reality and intelligent blended learning, so as to improve the quality and effectiveness of teaching.

From the theoretical point of view, this research will help to enrich and deepen the research content in the field of educational technology. Intelligent blended learning based on virtual reality is a relatively new research field, and its theoretical exploration and empirical analysis will help to broaden the research horizon in this field and promote the development of related theories.

For educational practice, the research results will provide educators with a new teaching method, which is expected to improve the learning effect of students. Moreover, due to the wide applicability of intelligent hybrid learning and virtual reality technology, this new teaching method can be used not only for traditional school education, but also for vocational training, distance education and other educational occasions.

The process of data collection, model building and validation in the research is also of great significance. Through these processes, the effect of intelligent hybrid learning method based on virtual reality can be evaluated quantitatively, and new ideas and tools can be provided for the evaluation of teaching effect.

The results of this study will play an active role in promoting the development of educational theory and the improvement of educational practice.

With the rapid development of educational technology, the application of virtual reality (VR) technology in intelligent learning environment has gradually become the focus of research. Several articles have explored the application of virtual reality technology in education and training in various disciplines. Nuguri et al. proposed a cloud-based system called “vSocial” that is designed and implemented for social virtual reality learning environment applications in special education. Studies have found that the learning needs of special education students can be more effectively met by combining cloud technology and virtual reality [1]. In their research, Izard et al. explored the potential of virtual reality as a tool for medical education and training, arguing that VR can provide an immersive learning experience, thereby enhancing the learning quality and skill training of medical students [2]. For how to improve students’ perceptual learning effect in virtual reality interaction, Lin et al. proposed a new approach aimed at enhancing learners’ participation and perception through specific interactive strategies [3]. Chen et al. focused on how to combine spherical video-based virtual reality technology with creative writing teaching and found that this teaching strategy can effectively stimulate students’ imagination and thus improve their creative ability [4]. In addition, Rzanova et al. found in their research on engineering students that the combination of virtual reality technology and case method can effectively improve students’ vocational ability [5]. Shimizu et al., on the other hand, pointed out that in Asian medical education, combining hybrid problem-based learning with virtual reality can promote the transformation of students from passive learners to active learners [6].

In general, the application of virtual reality technology in intelligent learning environments has shown great potential. Whether it’s medicine, engineering, or creative writing, combining VR and other educational approaches can provide students with a more immersive, personalized, and effective learning experience.

This research focuses on the development and validation of models that use intelligent blended learning and virtual reality technology to predict student learning performance. Through detailed data analysis and statistical test, the results show that the model has high prediction accuracy. The numerical results not only confirm the validity of the model, but also reveal the application potential of intelligent blended learning and virtual reality technology in the field of education. However, the study also points out that the application of virtual reality in education still faces a series of challenges and limitations, such as high equipment costs and technical barriers. Therefore, this study not only provides useful reference information for educators in adjusting teaching strategies, but also opens up new directions and methods for future research.

2. Technical background and learning environment limitations

2.1 Limitations of traditional learning methods

In the modern educational environment, although the traditional learning methods such as classroom teaching still play an irreplaceable role in some aspects, its limitations are increasingly obvious [7]. First, these traditional methods are often teacher-centered and adopt a lecture-style teaching model, resulting in students becoming passive knowledge recipients. This one-size-fits-all approach is difficult to take into account the individual differences of students, such as learning interest and speed, which may affect the learning effect and students’ learning enthusiasm. Educational psychology research has even shown that individualized teaching can improve learning outcomes by as much as 20–30%.

Secondly, the traditional teaching method is insufficient in cultivating students’ practical ability, critical thinking and innovative ability, and focuses too much on the inculcating and memorizing of knowledge [8, 9]. This is not desirable in the fast-moving and changing society of the 21st century. Some educational studies even suggest that an excessive focus on memorization and testing can impair students’ ability to solve practical problems.

Furthermore, traditional methods have problems with monitoring and feedback. Usually, it is difficult for teachers to fully and accurately grasp the learning status of each student, which is not conducive to timely finding and solving the problems encountered by students and further affecting the learning effect. Finally, these traditional models are limited in time and space and lack flexibility, which may affect the efficiency of learning [10].

Taking these factors together, the limitations of traditional learning methods highlight the urgent need for more efficient and personalized teaching methods in modern education. This is also one of the important drivers driving us to explore new teaching models such as virtual reality-based technology and intelligent blended learning, in order to achieve better education quality and results.

2.2 Introduction to intelligent blended learning

Intelligent Blended Learning is an innovative educational model that blends online learning (intelligent learning systems) with traditional face-to-face education methods. This hybrid model not only inherits the advantages of both, but also further realizes the personalized learning through artificial intelligence technology [11, 12]. Core technologies include machine learning, data mining and natural language processing.

The main theoretical basis of this model comes from many research fields such as educational psychology, computer science and human-computer interaction. By analyzing students’ learning behaviors and effects, the intelligent blended learning system can accurately understand students’ needs and difficulties, and then provide personalized learning advice and teaching support for students [13]. At the same time, the system provides a more natural and user-friendly user interface through natural language processing technology, which greatly enhances the learning experience of students.

The blended learning model has obvious advantages in practical application [14]. For those students who cannot participate in traditional face-to-face learning due to time and place, this model provides more learning resources, learning tools, and more flexible learning time and place. It not only solves the problem of lack of face-to-face interaction in pure online learning, but also retains the instant feedback and rich expression gesture language in traditional education, thus helping to improve the effectiveness and efficiency of learning.

To sum up, intelligent blended learning is a new education model driven by artificial intelligence and integrating the advantages of multiple learning styles. It not only provides a more efficient and attractive learning environment, but also greatly improves students’ learning results through personalized teaching programs.

2.3 Introduction to virtual reality technology

Virtual reality (VR) is an advanced computer technology capable of generating three-dimensional virtual worlds that provide users with sensory experiences similar to or completely different from the real world. In this three-dimensional space, users can freely observe and operate things, realizing multi-angle perception under the human perspective and scale [15, 16]. The application of virtual reality in the field of education shows innovative learning possibilities, such as students can observe complex mechanical structures in a simulated three-dimensional environment, or conduct experiments in a simulated chemistry laboratory, which are difficult to achieve with traditional teaching methods.

When virtual reality technology is combined with intelligent blended learning, the learning experience and effect are further improved [17]. This combination will not only provide more vivid and intuitive learning content, but also provide personalized learning advice and teaching support by analyzing students’ behavior and reactions in the virtual environment through AI technology. This comprehensive application is realized on the basis of the intersection of several disciplines such as educational psychology, computer graphics and artificial intelligence.

This kind of combination of mutual promotion has broad application prospects [18]. In addition to natural science experiments and engineering, virtual reality can be used in many fields such as history, culture, and even medical education. On the whole, the combination of virtual reality technology and intelligent blended learning opens up new research and application directions in the field of education, which is expected to further enrich and optimize the learning experience of students and improve the learning effect.

3. Data collection and processing

3.1 Data collection

The data collection process is an important part of this study, which includes detailed records of individual information and learning behaviors of students. The collection of this information and behavioral data needs to be carefully designed and implemented to ensure the accuracy and reliability of the results.

The individual information of students is the basis of the research. This information includes the student’s gender, age, major and other basic information, as well as their learning background, such as learning experience, learning interests, learning motivation and so on. This information is obtained in two main ways. One way is through the questionnaire survey, the design of detailed and comprehensive questionnaires, covering students’ personal basic information and all aspects of learning background, to ensure the comprehensiveness and detail of the data. Another way is to obtain data such as students’ academic performance through the school’s educational administration system to assess students’ learning ability. The combination of these two approaches makes data collection comprehensive and accurate. Data collection sample information, as shown in Table 1.

Table 1
Sample information

Student ID	Sex	Age	Profession	Learning experience (years)	Learning interest	Learning motivation	Academic performance
001	Male	20	Computer	2	High	Middle	85
002	Female	21	Maths	3	Middle	High	88
003	Male	22	Physics	1	Low	Middle	81
004	Female	20	Physics	2	Middle	High	79
005	Male	21	Maths	3	High	low	90

Students’ learning behavior data is collected through the logging function built into the intelligent blended learning system. The system will automatically record the students’ behavior in the virtual reality learning environment, including the pages they browse, the links they click, the keywords they query, the assignments they submit, the time they read, the number of times they take notes, the number of times they participate in discussions, and so on. At the same time, students’ comments, suggestions and questions about the curriculum and teaching system will be recorded systematically. The sample learning behavior information is shown in Table 2.

Table 2

Sample learning behaviors

Student ID	Browsing time	Clicks	Query times	Number of submissions	Reading time	Number of notes	Number of discussions	Test score	Evaluate
001	58	32	16	8	120	10	6	84	Good
002	72	27	14	9	105	12	5	89	Optimal
003	68	30	12	7	98	9	4	80	Good
004	75	35	18	8	95	11	7	82	Middle
005	66	28	13	6	102	13	8	87	Fine

It is worth noting that in order to ensure the validity of the data, the study will adopt a variety of validation methods. This includes cross-validation with other studies and databases, and the use of statistical methods to assess the reliability and consistency of the data. Through such a rigorous data collection and processing process, the research aims to provide a highly scientific and reliable research basis.

In summary, through this series of detailed data collection and verification steps, it is expected that not only can students’ learning behaviors and preferences be comprehensively understood, but also various teaching phenomena in virtual reality and intelligent hybrid learning environments can be deeply analyzed and explained, so as to provide more reliable and accurate research results.

3.2 Data preprocessing

Missing value processing: During the data collection process, there may be situations where certain information or behavioral data for some students is missing. A common way to deal with these missing values is to fill them by interpolation. For example, you can use mean interpolation to fill in missing values with the average of the same feature. The following Eq. (1) is shown:

$\displaystyle x_{\textit{missing}}=\frac{1}{N}\sum\limits_{i=1}^{N}{x_{i}}$ (1)

Where $x_{\textit{missing}}$ is the missing value, $x_{i}$ is the I-th observation of the feature, and $N$ is the number of observations of the feature.

Outlier processing: During the data collection process, there may be some outliers that are significantly different from other data and may affect subsequent data analysis and modeling. For outliers, one approach is to use quartile rules for identification and processing. Specifically, the Interquartile Range (IQR) is a statistic used to measure the extent of a data distribution, calculated by. The following Eq. (2) is shown:

$\displaystyle\textit{IQR}=\textit{Q3}-\textit{Q1}$ (2)

Among them, the $Q_{1}$ is the lower quartile, $Q_{3}$ is the upper quartile.

The specific outlier judgment formula is as follows. The following Eq. (3) is shown:

$\displaystyle Q_{1}-1.5\textit{IQR}>x\ \textit{or}\ Q_{3}+1.5\textit{IQR}<x$ (3)

Here, $x$ represents an observation in the data.

Data standardization: In order to eliminate the impact of dimension and numerical size between different features, it is often necessary to standardize the data. A commonly used method is Z-score standardization, the converted data conforms to the standard normal distribution. The following Eq. (4) is shown:

$\displaystyle z=\frac{x-\mu}{\sigma}$ (4)

Where $x$ is the original observation, $\mu$ is the mean of the feature, and $\sigma$ is the standard deviation of the feature.

The specific data processing flow chart is shown in Fig. 1.

Table 3

Descriptive statistical analysis

Trait	Mean value	Standard deviation	Minimum value	Maximum value	Median
Age	20.8	0.84	20	22	21
Learning experience	2.2	0.84	1	3	2
Academic performance	84.6	3.78	79	90	85

Figure 1.

Data preprocessing process.

3.3 Descriptive analysis of data

The descriptive analysis of data is an important step in the research. Through the preliminary analysis of the collected data, the basic situation of students and the general situation of learning behavior can be obtained. It is helpful to further understand the learning behavior of students in the intelligent blended learning environment and provide a basis for the subsequent model building. Descriptive statistical analysis of the data is shown in Table 3.

Analyze the basic information of students. The data shows that the number of male and female students in the collected data is roughly equal, and the age distribution is mainly 20–22 years old, which is in line with the general age distribution of college students. Students’ specialties, including computer science, mathematics, physics, English and biology, reflect the diversity of the university. Students’ learning experience is between 1 and 3 years, most of the learning interest and motivation are medium or high, and the academic performance is mainly between 80 and 90 points.

The data of students’ learning behavior are analyzed. The data show that the activities of students in the virtual reality learning environment are quite active, and the browsing time, the number of clicks, the number of inquiries, the number of submissions, the time of reading, the number of taking notes, the number of participating in discussions and so on have a high amount of activity. At the same time, students’ evaluation of virtual reality learning environment is mainly good or excellent, reflecting the positive impact of virtual reality technology on improving learning results. It should be emphasized that these descriptive analysis results are indispensable for subsequent causality analysis and model construction. They not only help us to understand the behavior pattern of students in intelligent blended learning environment more deeply, but also provide detailed and concrete data support for further research. Analysis of students’ learning behavior, as shown in Fig. 2.

Figure 2.

Descriptive statistical analysis.

4. Model building

4.1 Theoretical basis of model establishment

The goals of the model need to be established. The goal is to understand and quantify the learning behaviors of students in virtual reality environments and how these behaviors affect their learning performance. To achieve this goal, linear regression models can be used. A linear regression model is a commonly used model in statistics that can be used to describe the relationship between two or more variables. In this model, students’ learning behavior is the input feature, which is the independent variable of the model, while learning performance is the target of prediction, which is the dependent variable. Suppose there are n independent variables ( $x_{1}$ , $x_{2}$ , $\ldots$ , $x_{n}$ ), then the learning performance y can be expressed as:

$\displaystyle y=\beta_{0}+\beta_{1}x_{1}+\beta_{2}x_{2}+\ldots+\beta_{n}x_{n}+?$ (5)

Where $\beta_{0}$ , $\beta_{1}$ , $\ldots$ , $\beta_{n}$ is the model parameter and $\epsilon$ is the error term of the model. By finding the best model parameters, the gap between the learning performance predicted by the model and the actual learning performance can be minimized.

Having identified the model between learning behavior and learning performance, the next challenge is how to optimize the virtual reality learning environment to enhance students’ learning performance. This brings us to optimization theory, a method for finding the best solution to a function. For example, gradient descent is a common optimization method [19]. Assuming that some parameters in the virtual reality environment are $\theta$ , then the optimization problem can be expressed, as shown in the following Eq. (6):

$\displaystyle\textit{minimize}:f(\theta)$ (6)

Here, $f(\theta)$ represents some loss function associated with learning performance (which is usually a negative value). It is hoped that by adjusting $\theta$ to minimize this loss function, the learning performance can be maximized.

Because we want to minimize negative learning performance, which means maximizing learning performance. Issues that may need to be considered in this process include how to select initial parameters, how to adjust the learning rate, and how to deal with possible constraints.

4.2 Model design and establishment

Independent variables in linear regression models should be important factors related to learning performance. In this study, the selected independent variables might include learning time ( $x_{1}$ ), engagement ( $x_{2}$ ), interaction times ( $x_{3}$ ), fitness of virtual environment ( $x_{4}$ ), etc. Models can be written as. The following Eq. (7) is shown:

$\displaystyle y=\beta_{0}+\beta_{1}x_{1}+\beta_{2}x_{2}+\beta_{3}x_{3}+\beta_{% 4}x_{4}+\epsilon$ (7)

Among them, the $y$ said learning performance, $\beta_{0}$ , $\beta_{1}$ , $\beta_{2}$ , $\beta_{3}$ and $\beta_{4}$ is the model parameters, epsilon is model error term.

After the model design is completed, the next step is the model building process, that is, parameter estimation. Parameter estimation should be based on actual collected data. For the choice of techniques and tools, the study uses Python’s Statsmodels and Scikit-learn libraries for data processing and model construction. The least squares method is implemented through functions built into these libraries to minimize the squared error between prediction and actual learning performance. In linear regression models, the least squares method is usually used to estimate the parameters, that is, to find a set of parameters that minimizes the sum of the square errors between the model’s predicted learning performance and the actual learning performance. Specifically, this process can be done with the following matrix operations. The following Eq. (8) is shown:

$\displaystyle\beta=(X^{T}X)^{-1}X^{T}y$ (8)

Where $X$ is the matrix containing all the independent variable data, $y$ is the vector containing the learning performance data, and $\beta$ is the model parameter vector to be estimated.

After this step is completed, the specific parameters of the model are obtained, so that a model predicting the student’s learning performance in the virtual reality environment is built. What’s more, by analyzing these parameters, research can gain insight into how various factors (such as learning time, engagement, etc.) affect learning performance. This not only provides a theoretical basis for the future learning environment optimization, but also gives concrete practical suggestions.

5. Verification of the model

5.1 Verification method and process

The data to be used in this study is collected from a virtual reality environment of intelligent blended learning. The environment recorded a large number of students in the process of virtual reality learning data, such as learning time, engagement, interaction times and virtual environment fitness. The data collection process is systematic and rigorous, ensuring the accuracy and integrity of the data.

In this study, we collected data from 500 students. Of these, 400 students’ data was used as the training set, while the remaining 100 students’ data was used as the test set. The division of the training set and the test set follows the principle of random sampling to ensure the similarity of the two statistical properties.

The data collected mainly came from the learning behavior of 100 students in a virtual reality environment. Each student’s data is recorded individually and given a unique ID. For example, here are some of the data for five of the students, as shown in Fig. 3.

Figure 3.

Data validation analysis.

Figure 4.

Prediction and actual analysis.

Through the interaction record of students in the virtual reality environment, the learning time and interaction times of each student are obtained. These data can be obtained directly from the learning management system. At the same time, the data of engagement was calculated by the ratio of the number of tasks completed by the students to the total number of tasks. The fitness data of the virtual environment was obtained through a questionnaire of students’ self-evaluation, which included questions related to the fitness of the virtual environment.

The verification of the model is mainly to test the prediction ability of the model on the unknown data, that is, the generalization ability of the model. Here, the verification methods used are the mean square error (MSE) and the coefficient of determination ( $R^{2}$ ).

Mean square error (MSE) is the average of the square of the difference between the predicted value and the actual value. The following Eq. (9) is shown:

$\displaystyle\textit{MSE}=\frac{1}{n}\sum\limits_{i=1}^{n}{(y_{i}-\hat{y}_{i})% ^{2}}$ (9)

In the above formula, $n$ is the number of samples in the test set, $y_{i}$ is the actual learning performance of the $i$ sample, and $\hat{y}_{i}$ is the predictive learning performance of the $i$ sample. The smaller the value of MSE, the better the prediction ability of the model.

The coefficient of determination ( $R^{2}$ ) is an indicator that reflects the degree of correlation between the predicted value of the model and the actual value. The closer the value is to 1, the better the predictive ability of the model. The following Eq. (10) is shown:

$\displaystyle R^{2}=1-\frac{\sum\limits_{i=1}^{n}{(y_{i}-\hat{y}i)^{2}}}{\sum{% i=1^{n}}(y_{i}-\bar{y})^{2}}$ (10)

Where, $y_{i}$ is the actual learning achievement observed in $i$ , $\hat{y}_{i}$ is the $i$ -predicted learning achievement predicted by the model, and $\bar{y}$ is the average of all observed actual learning achievements.

Prediction and actual learning behavior analysis, as shown in Fig. 4.

Cross-validation is also planned to evaluate the robustness of the model. In cross-validation, the data set is divided into a training set and a test set, and this is done several times to obtain a more comprehensive evaluation.

5.2 Verification results and analysis

5.2.1 Verification result

Through the verification of the model, it can be seen from the results of the test set that the prediction effect of the model has reached the expected goal.

The obtained mean square error (MSE) is 0.5 and the coefficient of determination ( $R^{2})$ is 0.96, indicating that the model has high prediction accuracy and good prediction ability, and can be used to predict the actual learning performance. The verification results are analyzed as shown in Table 4.

Table 4
Analysis of verification results

Student ID	The difference is $(y_{i}-\hat{y}_{i})$	The difference is squared $(y_{i}-\hat{y}_{i})^{2}$
071	0.8	0.64
072	0.3	0.09
073	0.1	0.01
074	0.7	0.49
075	0.4	0.16
MSE, $R^{2}$	MSE $=$ 0.5	$R^{2}=$ 0.96

The results of model verification prove the validity and accuracy of this study. First, it confirms the positive role of intelligent blended learning methods in improving learning outcomes. Secondly, through rigorous verification of the model, it provides a reliable method to predict students’ learning performance, which has important practical value for teachers to adjust teaching strategies and improve teaching results.

5.2.2 Result analysis

The prediction accuracy of the model is excellent. The mean square error (MSE) is 0.5 and the coefficient of determination ( $R^{2}$ ) is 0.96. This means that the average deviation between the learning performance predicted by the model and the actual learning performance is very small, and 96% of the data variance can be explained by the model. In the predictive model, the closer MSE is to 0 and $R^{2}$ is to 1, the better the predictive effect of the model is. In this case, the prediction effect of the model reaches a very high level, indicating that the construction of the model is successful and can be used to predict the actual learning performance.

The model of this study is based on the application of intelligent blended learning and virtual reality, and the accuracy of its prediction results validates the effectiveness of these two technologies in the field of education. It not only reveals the positive role of intelligent blended learning methods in improving learning outcomes, but also provides a reliable method to predict students’ learning performance through rigorous validation of the model.

From the perspective of educational practice, the model provided in this study has important reference value for educators. It provides a new possibility for teachers to adjust teaching strategy and improve teaching effect. For example, after students’ learning performance is predicted by the model, teachers can adjust their teaching methods according to the predicted results, so as to better meet students’ learning needs and further improve their learning effect.

The results of this study also provide new ideas and methods for future research. The application of virtual reality technology in the field of education is becoming a new research hotspot. This study successfully applied this emerging technology to intelligent blended learning, and proved its effectiveness through model verification. This not only opens up a new research path for hybrid learning methods, but also provides new research ideas and methods for future researchers.

5.2.3 Challenges and limitations of virtual reality in education

After analyzing the model’s outstanding performance and application potential, we must also examine some of the real-world challenges and limitations of virtual reality (VR) technology in educational practice. First, the relatively high cost of equipment and maintenance for virtual reality may hinder its widespread deployment in educational institutions. Second, the technical threshold may still be relatively high for many teachers and students, requiring time and professional training to adapt.

What’s more, while our model demonstrates that VR and intelligent blended learning can effectively improve learning outcomes, over-reliance on virtual environments may have its own risks, such as causing students to disconnect from the real world. Such limitations may not only affect the overall development of students, but also reduce the effectiveness and feasibility of the model in practical application.

Therefore, while the models in this study achieve encouraging results, the potential challenges and limitations of these technologies need to be fully and carefully considered before they can be considered for widespread application to educational practice. This will not only help us more accurately assess the practical value of these emerging technologies, but will also provide educators with a more holistic perspective to better adapt and apply these advanced tools.

6. Summary and suggestions

6.1 Article summary

This study successfully constructs and validates an intelligent hybrid learning model based on virtual reality, which aims to improve students’ academic performance and meet the diverse needs of modern education. The model performed well, thanks to several key factors. First, we achieved highly accurate predictions by intelligently selecting key independent variables that affect academic performance, such as learning time, engagement, number of interactions, and virtual environment fitness. Second, models also benefit from high-quality data sets that are not only accurate but also comprehensive, reflecting a variety of different learning situations. Furthermore, the linear regression model adopted by us achieves a good balance between complexity and generalization ability while avoiding overfitting and underfitting. Finally, by applying optimization algorithms such as gradient descent, we find a set of model parameters that perform well on both the training set and the test set.

In terms of data validation, the mean square error (MSE) of the model is 0.5 and the coefficient of determination ( $R^{2}$ ) is 0.96, which means that the model can accurately predict 96% of the variance of academic achievement. These results not only demonstrate the validity of the model, but also highlight the great potential of virtual reality and intelligent blended learning in modern education reform.

Overall, this study not only reveals the limitations of traditional teaching methods in adapting to the current educational environment, but also highlights intelligent blended learning as a powerful teaching reform strategy. The model has been rigorously verified by data-driven methods, but there are some limitations, such as the sample size is mainly concentrated in the higher education stage, and the data volume is relatively limited. Therefore, the focus of future research will be to expand the sample scope and further optimize the data processing methods.

6.2 Suggestions

This study makes a series of concrete and actionable recommendations for educational reform and practice. First, considering the excellent performance of intelligent blended learning and virtual reality in improving students’ learning results, education departments and schools should adopt this new teaching model more widely. Especially in higher education and K-12, the model can be effectively applied to online and hybrid course design, especially in highly interactive and hands-on subjects. For vocational training, virtual reality can be used to simulate operations or recreate specific scenarios to further improve the quality of training.

Second, this study highlights the great potential of data-driven research methods in education research. Therefore, education researchers should actively explore and apply data-driven research methods and consider how other advanced technologies, such as AI and big data, can be combined with virtual reality and intelligent blended learning. The data quality and processing method will directly affect the prediction effect of the model, so more in-depth research is needed to improve the prediction accuracy of the model.

Third, how to integrate virtual reality technology more closely with intelligent blended learning is still a question worthy of further exploration. Teachers can adjust and improve teaching strategies according to the predicted results of the model in time to better meet the individual learning needs of students. At the same time, it is also necessary to continue to pay attention to and optimize the teaching strategy, and explore how to use virtual reality technology to improve the teaching effect under the framework of intelligent blended learning.

Overall, these comprehensive and specific recommendations aim to integrate theoretical research more closely with practical applications to meet the complex and diverse needs of the modern educational environment.

References

Nuguri

Calyam

Oruche

Gulhane

Valluripally

Stichter

. vSocial: a cloud-based system for social virtual reality learning environment applications in special education. Multimed Tools Appl. 2021; 80(11): 16827-16856.

Izard

Juanes

García Peñalvo

Estella

JMG

Ledesma

MJS

Ruisoto

. Virtual reality as an educational and training tool for medicine. J Med Syst, 2018; 42(3): 50.

Lin

Lan

Wang

. A novel method for improving the perceptual learning effect in virtual reality interaction. Multimed Tools Appl. 2022; 81(15): 21385-21416.

Chen

Cukurova

. Unleashing imagination: an effective pedagogical approach to integrate into spherical video-based virtual reality to improve students’ creative writing. Educ Inf Technol. 2023. doi: 10.1007/s10639-023-12115-7.

Rzanova

Yushchik

Markova

Sergeeva

. Impact of virtual reality technologies in the context of the case method on engineering students’ competencies. Educ Inf Technol. 2023. doi: 10.1007/s10639-023-12123-7.

Shimizu

Nakazawa

Sato

Wolfhagen

IHAP

Könings

. Does blended problem-based learning make Asian medical students active learners:? a prospective comparative study. BMC Med Educ. 2019; 19: 147.

Murphy

Stein

Pahwa

McGuire

Kumra

. Improvement of Medical Student Performance in Telemedicine Standardized Patient Encounters Following an Educational Intervention. Fam Med. 2023; 55(6): 400-404.

Larabi-Marie-Sainte

Jan

Al-Matouq

Alabduhadi

. The impact of timetable on student’s absences and performance. Plos One. 2021; 16(6): e0253256.

Dean

Halpern

McAllister

Lazenby

. Nursing education, virtual reality and empathy? Nurs Open. 2020; 7(6): 2056-2059.

10.

Wang

Meng

Xie

Jiang

Ding

Shao

. Reference training system for intelligent manufacturing talent education: platform construction and curriculum development. J Intell Manuf. 2023; 34(3): 1125-1164.

11.

Sadik

, Abdulmonem

. Improvement in student performance and perceptions through a flipped anatomy classroom: shifting from passive traditional to active blended learning. Anat Sci Educ. 2021; 14(4): 482-490.

12.

Liu

Chen

Zhang

Rao

. Student engagement study based on multi-cue detection and recognition in an intelligent learning environment. Multimed Tools Appl. 2018; 77(21): 28749-28775.

13.

Zhang

. Design of blended learning mode and practice community using intelligent cloud teaching. Educ Inf Technol. 2023; 28(8): 10593-10615.

14.

Binh

Trung

Duy

. Responsive student model in an intelligent tutoring system and its evaluation. Educ Inf Technol. 2021; 26(4): 4969-4991.

15.

Injadat

Moubayed

Nassif

. Machine learning towards intelligent systems: applications, challenges, and opportunities. Artif Intell Rev. 2021; 54(5): 3299-3348.

16.

Bakare

Orji

. Effects of reciprocal peer tutoring and direct learning environment on sophomores’ academic achievement in electronic and computer fundamentals. Educ Inf Technol. 2019; 24(2): 1035-1055.

17.

Erümit

Çetin

. Design framework of adaptive intelligent tutoring systems. Educ Inf Technol. 2020; 25(5): 4477-4500.

18.

Yeung

Carpenter

Corral

. A comprehensive review of educational technology on objective learning outcomes in academic contexts. Educ Psychol Rev. 2021; 33(4): 1583-1630.

19.

Aljemely

Xuan

Jawad

FKJ

. A novel unsupervised learning method for intelligent fault diagnosis of rolling element bearings based on deep functional auto-encoder. J Mech Sci Technol. 2020; 34(11): 4367-4381.