Bulletin of the Technical Committee on Learning Technology (ISSN: 2306-0212)
A NetLogo-based simulation model is proposed in this study to help return informatics and computer education to offline teaching while reducing the transmission rate of SARS-CoV-2. The model can simulate the number of high-risk students brought in by superspreaders under different course schedules. The model can also reset the transmission radius and transmissibility of variant and sub-mutant strains according to the evolution of the SARS-CoV-2 virus. In response to the simulation results, this study proposes the use of nonlinear optimization to rearrange the schedule, which reduces the number of high-risk students by approximately 10% in the optimized schedule for the same simulation conditions. Although the model has some imperfections, combined with the optimization method, it will be used in some universities’ autumn semester informatics course schedules. The model and methodology proposed in this study will play a crucial role in the movement to return to offline education in the post-COVID-19 era.
Keywords: Informatics courses, NetLogo-based simulation, Offline education, Post-COVID-19, Transmission model.
Information and computer education is fundamental to modern education. In most higher education systems worldwide, informatics education is a combined lecture and practical training course . In particular, students’ computer skills are irreplaceable in terms of knowledge acquisition and skill improvement. However, this educational model is being destroyed by an epidemic, COVID-19, which has been accompanied by urban lockdowns during which students have not been able to attend school. The COVID-19 pandemic is a huge challenge to education systems, especially for courses that rely on practice, such as informatics education . Meanwhile, offline education has begun to return to normal in various countries, along with increased COVID-19 vaccination rates. However, offline education is still not fully back to normal, as face-to-face education during pandemics continues to pose two risks. The first is that a significant percentage of students will not be vaccinated for a variety of medical reasons, while their exposure to the virus remains a threat. The second is that vaccines’ protective capacities decline as the virus mutates . This means that masks and alcohol disinfection remain the first choices when universities conduct face-to-face courses.
Despite the dire conditions, universities in Japan still place great emphasis on informatics education courses, while the Japanese government has even made informatics a subject for university admission tests in 2025 . To ensure the quality of informatics education, the education community is looking for ways to return informatics education courses to the face-to-face mode. However, to completely return to offline informatics education is difficult because the government occasionally declares a state of emergency or a quasi-state of emergency. Therefore, in Tokyo, the most densely populated city in Japan, most universities use a hybrid model for informatics teaching. In hybrid mode teaching, lecturers deliver the core course content to students through real-time online teaching and uploaded videos . Offline, lecturers will offer hands-on coaching to students on computer skills. The combined online and offline teaching model reduces the risk of infection on the commute or on campus. However, there is a risk of a disconnection between online and offline teaching because lecturers do not receive timely feedback from students. In an effort to maximize the quality of informatics education, some universities in Tokyo decided to meet the challenge of returning to offline informatics education in the new academic year.
To help bring informatics education fully back to offline teaching, this study presents a model based on NetLogo that predicts the number of high-risk students in the presence of superspreaders. In addition, the schedule is optimized for the simulation results to reduce the risk of further transmission. The most significant advantage of the model is that it is composed according to the real layout of the informatics classroom and can be dynamically adjusted depending on the strength of viral transmissibility. This study hopes that the rise in vaccination rates, together with a well-organized schedule, will return students to offline, high-quality informatics and computer education and training.
II. PRELIMINARY WORK
The SARS-CoV-2 virus that caused the COVID-19 pandemic is transmitted by droplets and microscopic particles; thus, the layout of an informatics education classroom is essential. Some studies have demonstrated that appropriately setting partition walls in a classroom can inhibit the spread of the coronavirus to a certain extent , . Therefore, universities have placed a transparent material wall between computers in classrooms, as shown in (d) of Fig. 1.
As the partition walls do not completely block virus-carrying bioaerosols indoors, the doors and windows of the room are always open both during and after classes , . To increase the airflow volume in the classroom, a ventilation fan is switched on 24 hours, and an air freshener with a filter is located in the area of (c) of Fig. 1. However, because there is little evidence that the SARS-CoV-2 virus is transmitted through the surfaces of objects, an alcohol sanitizer and wet wipes are still placed at the entrance. Students and teachers are asked to clean their hands with the alcohol sanitizer and clean mice and keyboards with the wipes before class , , as shown in (b) of Fig. 1. Finally, everyone who enters the campus is required to wear a mask and have their body temperature tested, and will be denied access to the campus when their temperature exceeds 37°C.
Fig. 1 A layout map and photo of a classroom for the informatics course.
Although the informatics education environment has been prepared with utmost care, a need remains to consider the course scheduling. Some studies have shown that if a social distance is not maintained, there is a risk of infection, even if a mask is worn . Although the layout of the classroom can control students’ social distance during class, it cannot be guaranteed after class. Scheduling courses to maximize the use of campus space during the pandemic requires simulation. In this study, a multi-agent-based model is proposed to simulate the risk of infection among students during informatics education classes.
A. Simulation Environment
In this study, NetLogo (version 6.2.1) is used as the simulation environment. NetLogo is a programming language and integrated development environment (IDE) for agent-based modeling . In addition, this study uses the PyNetLogo package (version 0.3) as a bridge between Python and NetLogo . The simulation environment is Ryzen 1920X 12-core, 24-thread CPU with 128G of RAM using the Windows 11 operating system.
The real layout and schedule at the Information and Computer Education Center at Mejiro University were used to make the model realistic, as shown in Fig. 2. The right side of Fig. 2 shows the real-world layout and schedule, which is reconstructed in the model on the left side of Fig. 2. In the model, students can enter the classrooms through the corridors, and areas in black that are off-limits to students, such as offices, are ignored.
Fig. 2. Layout and schedule in the model.
Students are allowed to enter the general classrooms and free space after class, and the informatics classrooms can only be entered for classes. Therefore, each classroom’s doors will have two states: locked and unlocked. The locked state is red, while the unlocked state is green. Students are not allowed to stay in the corridor for a long time; thus, each student has an aiming area in each short term. There are five informatics classrooms (IC), four general classrooms (GC), and five entrances/ exits (EE) on the floor of the Information Education Centre. Each student will randomly select an aim at the time of a term change in three subsets, where
The capital abbreviations are for classroom sets and the lowercase abbreviations are for individual classrooms.
Each aim set, in turn, contains at least one subset, and students can choose only one element of the subsets as an aim in a term. In the program, the aim choosing process is sampling with replacement. If the premise is that all aim areas are accessible, the minimum probability is 7.14%. However, each classroom will become inaccessible with the schedule, and the maximum number of students per classroom is no more than 40; thus, the probability is not fixed. The informatics classrooms are open according to the class schedule by the center, while the general classrooms and free spaces are not under the control of the center. Based on the above logic, the interface of the model is shown in Fig. 3.
Fig. 3. The interface of the simulation model.
In Fig. 3, the “a-1” area can be used to control the schedule of informatics courses by selecting whether each classroom is “on” or “off” during five terms in one school day. In addition, different weekdays can be selected to emulate different course schedules through the button at bottom. Since there is a fixed number of students per classroom for informatics courses, an “attendance” button is set up in the “a-2” area to adjust the number of students attending an informatics course. In addition, the “num” button represents the number of students who desire to use general classrooms and free space on a weekday. The number is divided by five and is distributed equally among all terms. All buttons in the “a” area are parameters. The “b” area is a display frame of the simulation, where the observer can see each agent’s activity trajectory. There are two buttons in the “c” area, “setup” and “go.” The function of “setup” is to import parameters into the program to initialize the agent and the timeline. The function of “go” is to run the program. The “d” area is for a report, and the program automatically counts the number of students stranded in the corridor. The above rules are converted into the programming language; however, the virus diffusion algorithm is not yet incorporated into this program. Prior to that, the model must be validated and optimized.
C. Model Validation and Optimization
The validation of the model is a very important process, and the reasonableness of the model leads directly to the extent to which the results are correctly reported . However, for NetLogo, as a multi-agent simulation model for complex systems, validation is difficult. Having attempted several validation methods, the validation of the model in this study was performed by inputting real parameters . In this study, the informatics class schedule for the fall semester of 2021 was entered into the model and a tally counter was used to compare the students’ actions in the corridor against the simulation results.
Following validation, the results of the simulation did not truly reflect the trajectory of the students’ actions in the Information and Computer Education Center. An analysis revealed three reasons why the model did not correctly reflect the real situation. The first reason was that the lecturers would transfer the informatics class from offline to online, as appropriate, based on the situation report on COVID-19. Second, the informatics classroom would be used to instruct students in computer skills when no classes were scheduled. Third, the university policies tend to reduce the amount of time students are retained on campus and the classes in the general classrooms have still not reverted to the offline mode. Therefore, the model must be modified and improved by setting the real situation as a mathematical variable.
In addition, some details may be refined. For example, healthy students must be distinguished from superspreaders by color, while superspreaders and students must be assigned more parameters as constraints. Therefore, an improved model is proposed.
Fig. 4. The improved model.
In the improved model, the shortcomings of the previous model have been corrected; in addition, the SARS-CoV-2 virus spread model has been added. As shown in Fig. 4, the correction points are described below. First, the simulation period is changed from a school day to a school week. A longer simulation period will afford agents more freedom and achieve better simulation results. Second, unnecessary layouts have been removed and the dimensions of the classrooms, corridors, and agents in the model re-scaled. Although the authenticity of the model is sacrificed, its complexity is reduced. Third, the students’ behavior has been adjusted according to the new university policy.
Regarding the SARS-CoV-2 virus spread model, it is not significantly different from that of influenza in how the virus spreads, i.e., contact (direct and through fomites), large droplets, and aerosols , . Therefore, the improved model refers to some classical NetLogo-based virus spreading simulation models , . In the current model, superspreaders are released from movement restrictions after class. There are two parameters regarding superspreaders: the number of symptomatic positives and whether the symptomatic positives are superspreaders. If the superspreaders’ switch is fixed at “True,” then all symptomatic positives are superspreaders. Conversely, the superspreaders are picked according to symptomatic positives and a random integer value from -5 to 5, because asymptomatic students can also become superspreaders, as shown in Equation (1).
k (natural number) denotes the number of superspreaders, while p denotes the number of symptomatic positives.
The continuous mutation of the SARS-CoV-2 into new variants is the reason for the delayed end of the pandemic. Therefore, the aerosols formed by different variants have different transmission radii and levels (Simplify with Basic-reproduction-number). For this study, a simplified version of Bazant et al.’s study was followed for transmission radius and level settings . Therefore, a new transmission theory is shown in Fig. 5.
Fig. 5. A simplified transmission pattern.
Fig. 5 shows that the SARS-CoV-2’s transmission range is a circle around the center of the superspreaders. The width and height of the corridor and the radius of contagion will form the shape of a piece of cheese. Coronaviruses are distributed in this space; however, the proportion of alive coronaviruses becomes lower as the radius grows, as calculated by the following equation:
In Equation (2), P is the proportion of alive coronaviruses and l is a transmission level parameter. OR3 is the maximum transmission range for the SARS-CoV-2, OR2 is 2/3 of OR3, and OR1 is 1/3 of OR3. When the n is social distance, in the R3R2, R2R1, and R1O segments, n = 30, n = 20, and n = 10, respectively. The value of m is typically 1. Where C is combination formula and m is the number of combinations.
As an example, a Level 4 SARS-CoV-2 variant with a maximum transmission distance of 5 m would have an alive coronavirus percentage of 16.7% at 3.5 m.
After all the model’s modifications, it was again validated using the informatics class schedule for the fall semester of 2021. In the improved model, each agent’s motion trajectory was significantly improved, the unreasonable zigzagging, wandering, or other actions that appeared in the old model was fixed, and the model was considered to be suitable for simulation.
IV. SIMULATION RESULTS AND OPTIMIZATION PROCESS
A. Simulation Results for Unoptimized Course Schedule
Before the simulation process, many parameters must be set, with transmissibility a priority. In this model, all transmissibility relates only to the physical dimension (as illustrated in Fig. 5) and does not discuss the vaccine protective dimension. The shortlist of transmissibility parameters is set by all the SARS-CoV-2 variants’ transmissibility from the beginning of December 2019 to the end of February 2022, refer to in “Investigation of SARS-CoV-2 variants: technical briefings” . Currently (February 26, 2022), since only the SARS-CoV-2 Omicron and SARS-CoV-2 Delta variants are mainly spread in Japan, these are on the final list.
In addition, subvariants of the Omicron variant are being observed and maybe more transmissible than the original Omicron variant . All subvarieties are excluded from the list due to the current lack of credible conclusions. Therefore, this study set the transmission levels for the original Omicron and original Delta variants to Levels 7 and 4, respectively, as the default values. As with transmissibility, the transmission radii for the Delta and Omicron variants were set at 4 meters and 7 meters, respectively, as the default values. Another parameter is students’ moving speed: the value is equal to 1 for running and 0 for standing still; the default value was set to 0.5 in this study, i.e., normal walking speed.
Finally, only two parameters are yet to be set in this model: the number of superspreaders and the class schedule. According to the course plan of the Information and Computer Education Center of Mejiro University, in the spring semester of 2022, 60 classes per week will be taught in five informatics classrooms with a maximum capacity of 40 users. The unoptimized schedule is shown in Fig. 6.
Fig. 6. Spring semester informatics education course plan.
As shown in Fig. 6, there are 9, 15, 10, 11, and 15 classes on Monday, Tuesday, Wednesday, Thursday, and Friday, respectively, for a total of 60 classes per week. After the above schedule was entered into the simulator, results were obtained for the number of high-risk students. Fig. 7 shows a histogram of the number of superspreaders with the Omicron variant and the number of high-risk students after the simulation process had been repeated 25 times.
On the left side of Fig. 7, each bar represents the number of high-risk students that are transfected with the Omicron variant by the superspreaders within a school day. On the right side of Fig. 7, the total number of high-risk students in an academic week (fractions are rounded up) is shown. Two facts can be easily observed in Fig. 7: the number of high-risk students increases with the number of superspreaders, and the number of high-risk students is higher on Fridays, even if the same 15 classes are scheduled for Tuesday. The latter fact illustrates that the unoptimized schedule causes unnecessary close contact among students, and that there is room for further optimization of the course schedule.
Fig. 7. Spring semester informatics education course plan.
B. Optimization Process
Course schedule optimization is a complex problem that lies at the intersection of the natural and social sciences and is subject to a variety of conditions. It is essentially a combinatorial optimization problem, and using search algorithms to solve the problem is a popular approach , . Considering that this kind of algorithm is more stressful for computation, this study adopts nonlinear optimization using the augmented Lagrange method to deal with the course schedule optimization problem .
To smoothly perform the optimization, a simple organization of the available data is required. First, the number of informatics classrooms used in a term and that of high-risk students in the simulation results must be approximated as a function. In this study, the interpolation method was used . The approximation functions constructed are shown in Fig. 8, where f(x) is an approximation function for Terms 1-4 and g(x) is an approximation function for Term 5.
Fig. 8. Approximation function construction.
The x indicates the number of classrooms used at this time point. f(x) and g(x) are simulation results for the number of infections under the classroom arrangement. The minimum value of the sum of all f(x) and g(x) is then the objective function for this optimization process. The constraint functions as shown in Equations 3–4.
In Fig. 8, the number of high-risk students is zero for g(x), because students are no longer allowed to remain in the corridor after the last term of a school day. Meanwhile, scheduling all classes in the last term of a school day is not allowed, which is an additional constraint.
Once the objective function and the upper and lower bounds have been determined, the non-linear optimization process can be carried out. As there are not many bounds for the objective function, in this case, the open-source R package Nloptr developed by Rahul Bhadani is used for optimization .
C. Simulation Results for the Optimized Course Schedule
The optimization process outputs a new course schedule, which is shown in Table 1.
THE NEW COURSE SCHEDULE
The number in each cell of Table 1 indicates the number of classrooms in teaching during every term of each school day. However, the optimized course schedule could not output the recommended classroom number. The above table was entered into the simulation, and the new results are shown in Fig. 9.
Fig. 9. Comparison of results optimized and unoptimized.
In Fig. 9, the study compares the results from 25 simulations under each of the optimized and unoptimized course schedules. It is evident from the graph that the number of high-risk students in the optimized schedule is lower than in the unoptimized schedule. All the results for these 5 groups were tested using the t-test by JASP (0.16.2.0 version), and p < 0.01, which was statistically significant.
Although there are many simulation models related to COVID-19, there are fewer transmission models for offline education. However, as the post-COVID-19 era begins, offline education is gradually being resurrected. While educators are looking forward to the return of offline courses, they are concerned about the risk of transmission; thus, a model that can evaluate the risk of offline courses is urgently required, hence this study. The present study observed that students were generally vaccinated, and that there were reliable methods to stop splash and aerosol transmission in the classroom, which are the main challenges in returning to offline teaching, given the highly transmissible nature of the Omicron variant and the social distance required after class. Due to the high transmissibility of the Omicron variant, this return cannot be achieved in a short time; in this study, a multi-agent-based model is developed to address the problem of controlling the social distance between students by organizing offline informatics education, thus reducing the spread rate.
In the proposed model, superspreaders infected with the SARS-CoV-2 Omicron variant move through classrooms and corridors alongside healthy students. Healthy students can become high-risk students as the social distance between the students and superspreaders changes. This is demonstrated by the simulation results, as can be observed in Fig. 7, where different distribution methods lead to different results even when the same number of courses are presented. This is consistent with the initial assumption in this study: If the density of the courses cannot be rationalized, the high transmissibility of the Omicron variant will hinder the plan to return informatics education to offline teaching. In response to the above simulation results, this study proposed a nonlinear optimization method to optimize the course schedule. After the simulation was repeated, it was found that the optimized schedule showed a significant decrease in the number of high-risk students in the same model, as shown in Fig. 9. This result indicates that, as expected, the simulation model, coupled with the course schedule optimization method proposed in this study, can help undergraduate informatics education combat the COVID-19 pandemic. The model will also soon be applied in the scheduling of the informatics education courses at the university in the autumn.
Despite the good results, several limitations of this study remain to be addressed. First, the validation of this model was determined only by observing the students’ trajectories in the corridor, without any clear indicators for reference. This study initially intended to track the students’ trajectories using face or object recognition, from validating and improving the model. However, two challenges caused this experiment to be postponed to a future study. One was that the recognition accuracy was drastically reduced due to the fact that everyone was wearing a mask; thus, the trajectory was interrupted. The second was that, because the informatics course is a mandatory course with a teaching load of over five hundred students in the spring semester, it was impossible to obtain the required informed consent from all the students in a short time. Therefore, this validation will be added to future studies in conducting the simulation model study for the no-mask offline education .
Another limitation appears in the layout of the classrooms and the organization of the courses. Because this study constructs the model based on a layout from only one university, there is room for improvement regarding the generality of this model. The next focus of this study is to parameterize the layout of the classroom to adapt it to those of different universities. Additionally, the course arrangement cannot be absolutely idealized, considering both students’ and teachers’ physical strengths and learning efficiencies.
In addition, as the Centers for Disease Control and Prevention (CDC) in the U.S. announced on April 18, 2022, that it would stop the mask order, an increasing number of countries have started a discussion on stopping wearing masks . However, there is a rapid increase in the number of positive patients during the initial period without masks; thus, it has become increasingly important to control the population density, especially in public areas such as schools and universities. The greatest advantage of the model proposed in this study is that the parameters can be added at any time to adapt to different simulation conditions. Therefore, the future direction of this study lies in the simulation and optimization of offline informatics education courses when no masks are worn and when the variety of the subvariants of Omicron is increasing. This study believes that the model will play an important role in the recovery of offline teaching in the post-COVID-19 era.
To return informatics education to offline courses while controlling the risk of SARS-CoV-2 transmission, this study proposes a NetLogo-based simulation model to predict the risk under the existing course schedule. The model predicts the number of high-risk students caused by superspreaders carrying the Omicron variant by inputting the course schedule. Non-linear optimization was used to optimize the course schedule and successfully reduced the number of high-risk students. The proposed model means that the countdown to the return of offline education has begun. The use of the model will specify a more rational and higher quality arrangement of courses for informatics education in the post-COVID-19 era.
 M. E. Caspersen, J. Gal-Ezer, A. McGettrick, and E. Nardelli, “Informatics as a fundamental discipline for the 21st century,” Commun ACM, vol. 62, no. 4, pp. 58–63, April. 2019. DOI: 10.1145/3310330.
 S. J. Daniel, “Education and the COVID-19 pandemic,” PROSPECTS, vol. 49, no. 1–2, pp. 91–96, Oct. 2020. DOI: 10.1007/s11125-020-09464-3.
 P. A. Christensen et al., “Delta Variants of SARS-CoV-2 Cause Significantly Increased Vaccine Breakthrough COVID-19 Cases in Houston, Texas,” The American Journal of Pathology, vol. 192, no. 2, pp. 320–331, Feb. 2022. DOI: 10.1016/j.ajpath.2021.10.019.
 K. Komagata, “Explanation of the ‘Information’ entrance examination for the common test starting in 2025 (in Japanese),” ICT Online. Avaialable:
https://project.nikkeibp.co.jp/pc/atcl/19/06/21/00003/100400271/ (accessed March. 10, 2022).
 Y. Mitsumori, “Learning Innovation through Online Lectures: A Study on Two-way Real-Time Lectures via Zoom,” in 2021 IEEE Region 10 Symposium (TENSYMP), Aug. 2021, pp. 1–6. DOI: 10.1109/TENSYMP52854.2021.9550869.
 P. Epple, M. Steppert, M. Florschütz, and P. Dahlem, “Partition walls as effective protection from bio-aerosols in classrooms – an experimental investigation,” GMS Hyg Infect Control, vol. 16, pp. 1–10, 2021. DOI: 10.3205/dgkh000380.
 M. Mirzaie, E. Lakzian, A. Khan, M. E. Warkiani, O. Mahian, and G. Ahmadi, “COVID-19 spread in a classroom equipped with partition – A CFD approach,” Journal of Hazardous Materials, vol. 420, pp. 1–18, Oct. 2021. DOI: 10.1016/j.jhazmat.2021.126587.
 B. Chen, P. Jia, and J. Han, “Role of indoor aerosols for COVID-19 viral transmission: a review,” Environmental Chemistry Letters, vol. 19, no. 3, pp. 1953–1970, Jun. 2021. DOI: 10.1007/s10311-020-01174-8.
 S. Kwon et al., “Association of social distancing and face mask use with risk of COVID-19,” Nature Communications, vol. 12, no. 1, pp. 1–10, Dec. 2021. DOI: 10.1038/s41467-021-24115-7.
 D. Lewis, “COVID-19 rarely spreads through surfaces. So why are we still deep cleaning?,” Nature, vol. 590, no. 7844, pp. 26–28, Feb. 2021. DOI: 10.1038/d41586-021-00251-4.
 U. Wilensky, “NetLogo,” Center for Connected Learning and Computer-Based Modeling, Northwestern University, 1999. Available: http://ccl.northwestern.edu/netlogo/ (accessed Mar. 10, 2022).
 M. Jaxa-Rozen and J. H. Kwakkel, “PyNetLogo: Linking NetLogo with Python,” Journal of Artificial Societies and Social Simulation, vol. 21, no. 2, 2018. DOI: 10.18564/jasss.3668.
 D. Sornette, A. B. Davis, K. Ide, K. R. Vixie, V. Pisarenko, and J. R. Kamm, “Algorithm for model validation: Theory and applications,” Proceedings of the National Academy of Sciences, vol. 104, no. 16, pp. 6562–6567, Apr. 2007. DOI: 10.1073/pnas.0611677104.
 K. Kahn, “An Introduction to Agent-Based Modeling: Modeling Natural, Social, and Engineered Complex Systems with NetLogo,” Physics Today. vol. 68. 2015. DOI: 10.1063/PT.3.2884.
 R. Tellier, “COVID-19: the case for aerosol transmission,” Interface Focus, vol. 12, no. 2, pp. 1–10 Apr. 2022. DOI: 10.1098/rsfs.2021.0072.
 C. R. Alimboyong, “Modeling virus spread on a network using NetLogo for optimum network management,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 23, no. 1, p. 370–377, July. 2021. DOI: 10.11591/ijeecs.v23.i1.pp370-377.
 D. Chumachenko, V. Dobriak, M. Mazorchuk, I. Meniailov, and K. Bazilevych, “On agent-based approach to influenza and acute respiratory virus infection simulation,” in 2018 14th International Conference on Advanced Trends in Radioelecrtronics, Telecommunications and Computer Engineering (TCSET), Feb. 2018, pp. 192–195. DOI: 10.1109/TCSET.2018.8336184.
 M. Z. Bazant and J. W. M. Bush, “A guideline to limit indoor airborne transmission of COVID-19,” Proceedings of the National Academy of Sciences, vol. 118, no. 17, pp. 1–12, April. 2021. DOI: 10.1073/pnas.2018995118.
 UK Health Security Agency, “Investigation of SARS-CoV-2 variants: technical briefings,” GOV.UK. Available: https://www.gov.uk/government/publications/investigation-of-sars-cov-2-variants-technical-briefings (accessed: April. 20, 2022).
 E. Takashita et al., “Efficacy of Antiviral Agents against the SARS-CoV-2 Omicron Subvariant BA.2,” New England Journal of Medicine, vol. 386, no. 15, pp. 1475–1477, April. 2022. DOI: 10.1056/NEJMc2201933.
 G. B. Dantzig, M. N. Thapa, Linear Programming 1: Introduction. New York: Springer New York, 1997. DOI: 10.1007/b97672.
 H. Soma, M. Hori, and T. Sogou, “Schedule optimization using fuzzy inference,” in Proceedings of 1995 IEEE International Conference on Fuzzy Systems. March, 1995, pp. 1171–1176. DOI: 10.1109/FUZZY.1995.409831.
 Y. Ye, “Interior Algorithms for Linear, Quadratic, and Linearly Constrained Non-Linear Programming,” PhD dissertation, Stanford University, Stanford, 1987.
 D. W. Scott, “The New S Language,” Technometrics, vol. 32, no. 1, pp. 103–104, Feb. 1990. DOI: 10.1080/00401706.1990.10484610.
 R. Bhadani, “Nonlinear Optimization in R using nlopt,” Jan. 2021, arXiv: 2101.02912v2. DOI: 10.48550/ARXIV.2101.02912.
 S. G. Coelho et al., “Return to school and mask-wearing in class during the COVID-19 pandemic: Student perspectives from a school simulation study,” Paediatrics & Child Health, vol. 27, no. Supplement_1, pp. S15–S21, May 2022. DOI: 10.1093/pch/pxab102.
 Centers for Disease Control and Prevention, “Order: Wearing of face masks while on conveyances and at transportation hubs,”. Available: https://www.cdc.gov/quarantine/masks/mask-travel-guidance.html (accessed: April. 29, 2022).
is currently a lecturer in Organization of Liberal Arts Education at Mejiro University for Informatics and computer courses. He graduated from Tokyo Metropolitan University with a Ph. D degree in engineering. And the M.Sc. degree is received in innovation for design and engineering from Advanced Institute of Industrial Technology, Japan, in 2019. His research interests include signal processing and simulation model building.
received his M. Sc. and Ph.D. from Keio University in 1991 and in 1997, respectively. He worked as a researcher in the National Institute of Advanced Industrial Science and Technology from 1997 to 2003, and also worked as an associate professor in Nagoya University from 2003 to 2009. In 2009, he joined Tokyo Metropolitan University, where he is currently a professor in the Faculty of Systems Design. His research interests include rehabilitation engineering, biomechanics and simulation model building.
is currently an assistant in Informatics and Computer Education Center, Mejiro University. She received her bachelor’s degree from Mejiro University. She is primarily responsible for operating and managing the informatics classroom, and her research interests include information science education.
is currently an assistant in Informatics and Computer Education Center, Mejiro University. He received his bachelor’s degree from Mejiro University. He is primarily responsible for running the informatics classroom, and his research interests include information science education.
is current an associate professor in Faculty of Media Studies, Mejiro University and he is the director in Informatics and Computer Education Center, Mejiro University. He received his M.A from Naruto University of Education. His research interests include information science education and educational technology.