Hybrid Flexible (HyFlex) learning space design and implementation at graduate level: An iterative process

Abstract This paper investigates the process of designing HyFlex classrooms at Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), a graduate-level research university located in Abu Dhabi, United Arab Emirates, underpinned by the application of the EDUCAUSE Learning Space Rating System (LSRS). This investigation takes the form of a case-study and specifically focuses on the rationale, planning, design, and technology behind the implementation of the flexible HyFlex spaces as deployed in several classroom environments at MBZUAI. Iterations’ performance was assessed with the LSRS—V3. The findings should make an important contribution to the field of HyFlex learning spaces and technology-enhanced classroom design at graduate level.


Introduction
Technology in the past 20 years has enhanced learning spaces to the point where students can work, learn, and interact in any environment, whether it is in a classroom or virtually.The choices given to students allow them to customize their learning experience to suit their individual needs.Existing research recognises the critical role played by thoughtful learning space design to provide flexibility (Christou et al., 2023) and maximise student learning and academic teaching experience.A decade ago, Keppell et al. (2011) highlighted for instance the need to include flexibility and adaptability to any learning space.In addition, Duvivier (2019) argued that the flexibility technology provided allowed "students to take a significant proportion of their learning off-campus" (p.1).Finkelstein et al. (2016, p.28) proposed five principles for student engagement, namely 1) academic challenge, spaces that support various teaching modalities and foster interaction with content 2) learning with peers, spaces that support peer interaction, individual and collaborative work, 3) experiences with faculty, spaces that support student-instructor interactions, 4) campus environment, future-proofed spaces, aligned with university strategy and goals, and 5) high-impact practices, spaces that would support a variety of active learning and teaching approaches.Although some research has been carried out on the design of learning spaces (Angelone et al., 2020;Beatty, 2007;Ellis & Goodyear, 2016;Eyal & Gil, 2020;Leijon et al., 2022) in higher education, a search of the literature revealed few studies, apart perhaps studies from Zeivots and Schuck (2018), Detyna et al. (2023), andSanchez-Pizani et al. (2022) which focus on HyFlex classroom design at graduate level.The research to date

Theoretical framework
The theoretical framework used to understand the synchronous learning and teaching experiences of distance and residential learners and academics in our context, blended graduate courses, is the Community of Inquiry (CoI) framework.Col was developed by a research group at the University of Alberta from 1996-2001.Garrison et al. (2010) extended the framework to include blended learning, emphasizing three critical elements in creating effective and meaningful online learning experiences.The CoI framework combines teaching, social, and cognitive presences to explore students' blended learning experiences (Cuadra & Bernal, 2023).The teaching presence refers to the design, facilitation, and direction of the online learning experiences by the instructor.It involves the selection of the learning materials, the facilitation of the activities, and the feedback guidance to the learners.Cognitive presence refers to the extent to which the learners can construct and ascertain meaning through reflection, critical thinking, problem-solving, and knowledge creation in the class.Social Presence refers to the degree to which the learners feel a sense of belonging and connectedness within the online learning environment.It also includes using communication tools and strategies to develop social interaction, collaboration, and support among the learners (Cuadra & Bernal, 2023).The (Col) framework provides a comprehensive understanding of the interrelationship among these presences and their impact on students' learning processes (Lakhal et al., 2021).
The CoI framework also provides a research-based framework for designing, facilitating, and directing collaborative, constructive learning environments in online and blended learning settings (Wang et al., 2023).It has been used to interpret educational experiences and address challenges such as isolation, low morale, and the absence of community-building in online learning environments (Wilson & Berge, 2023).
Additionally, the framework has been used to assess the effectiveness of blended synchronous learning (BSL) in higher education institutions, including during the COVID-19 pandemic (Zhang, 2020), hence its relevance to our study.It helps to explore students' perspectives on collaboration, critical thinking, and knowledge construction in a purposely designed blended learning environment (McGinn, 2019).Using the CoI framework, researchers can gain insights into BSL's pedagogy, technology, and organization/logistics aspects and identify challenges and opportunities for implementing synchronous online learning effectively (Boateng, 2020).

COVID-19 in the UAE and the switch to online learning
The Coronavirus pandemic , emerged at the end of 2019, began in Wuhan, China (Temsah et al., 2020) and spread to more than 200 countries (Shen et al., 2020).On 11 March 2020 the World Health Organization characterized COVID-19 as a pandemic, the first caused by a coronavirus (WHO, 2020).Lockdowns and curfews were implemented and millions of people across various sectors were directed by their governments to stay safe by quarantining partially (Cooper et al., 2020).United Nations reported that nearly 1.6 billion students worldwide were affected by the closure of educational institutions and 80-85% of the classes were moved to online learning, whereas only 50% made the shift in low-income countries (United Nations, 2020).
In the UAE, on 3 March 2020, the Ministry of Education closed schools and universities for one month commencing 8 March, with an early two-week Spring break on 8 March instead of 29 March.On 30 March all Higher Education Institutions (HEIs) in the UAE were advised to continue with online learning until the end of the 2019-2020 academic year (UAE Government Portal, 2020).All universities were directed to switch to a distance learning mode, they were given rules, regulations, and guidelines on online learning and assessment methods.
COVID-19 had a severe impact on the normal education processes at universities and sped up the reform of online education through the development of innovative content, and quick adaption to state-of-the-art technology.All teaching had to be done remotely and students and faculty were required to adjust (Shisley, 2020;Sun et al., 2020;Vlachopoulos, 2011).Calonge et al. (2022) argued that for "many teaching staff the pandemic presented their first experience with delivering teaching remotely and online (p.183).Ashour (2021) declared the UAE HEIs all had the tools required for online lectures, teleconferencing, and online exams but were reluctant to change.The pandemic hastened the shift, although faculty required training to be more creative in the approach of using the tools (Adedoyin & Soykan, 2020).To provide academics and personnel with teaching responsibilities with the skills to facilitate online classrooms, the Ministry of Education coordinated with a local university to launch the "Be an online tutor in 24 hours" online course [https://cloudcampus.hbmsu.ac.ae/enrol/ index.php?id=4].Whilst the course's main focus was to quickly upskill teaching personnel on the use of Learning Management Systems (LMS) and content authoring, it was not fit for purpose because (1) it only lasted 4 hours with no interactions (asynchronous), (2) it had no content on how to design pedagogically sound and engaging teaching and learning activities and assessment for the fully online format, and (3) it was system focused.

HyFlex
There is a growing body of literature that recognises the importance of the Hybrid Flexible (HyFlex) modality, particularly since the start of the pandemic (O'Ceallaigh et al., 2023).HyFlex is an innovative educational framework that has gained prominence in recent years, particularly in the context of higher education.It offers a versatile approach that combines face-to-face and online modalities, allowing students to choose their preferred mode of participation in a course.At its core, the HyFlex model aims to provide learners with maximum flexibility and agency in their educational journey while maintaining the quality of instruction.It recognizes the importance of accommodating diverse student needs, preferences, and circumstances, ultimately fostering a more inclusive learning environment.HyFlex's relationship with hybrid synchronous teaching and learning is integral to its success.In a hybrid synchronous setting, instructors simultaneously engage both in-person and remote students, leveraging technology to bridge the physical and virtual classrooms.This approach ensures that students, regardless of their location or circumstances, have access to real-time interactions, discussions, and collaborative activities, fostering a sense of belonging and participation in the learning community.HyFlex leverages the principles of hybrid synchronous teaching to offer students the choice to attend classes synchronously or asynchronously, tailoring their learning experiences to their unique situations and contexts.
In summary, the goal of HyFlex is to provide a) flexibility, b) an alternative (but equal) learning experience, and c) a choice to either attend face-to-face (synchronous) class sessions or complete course teaching and learning activities online (synchronously or asynchronously), without the need to physically attend classes.In short, HyFlex offers choice, an equivalent and flexible experience.HyFlex allows for "greater control over their learning and course engagement modes" (Kohnke & Moorhouse, 2021, p.232) and supports students "who choose not to, or are unable to, participate in traditional classroom instruction" (Beatty, 2020, p.1). Penrod (2022) argued that the implementation of a HyFlex strategy at university, in an increasingly competitive international higher education context, would in fact elevate "the student experience and help higher education institutions remain relevant" (p.1).Goodyear et al. (2021) described learning spaces as an intrinsic, interconnected and interdependent component of a larger ecosystem composed of various key elements: students, faculty (how they interact, synchronously and asynchronously, individually or in groups), pedagogical strategies (e.g., case-based learning, lecturing, teaming, the intended learning outcomes, etc.), tech tools (Learning management system, cameras, live polling, digital whiteboards, apps, etc.) and furniture (chairs, lecterns, etc.).They argued that the central focus needed to be on the intersection between the tasks (and teaching and learning activities) students were asked to do and the design elements in the physical and social domains.Deliberate and strategic design-thinking of the spaces (Bülow, 2022;Gobin, 2023;Nørgård, 2021;Penrod, 2023;Raes, 2022), choice of technology and pedagogy are therefore crucial to ensure a) engagement, persistence, satisfaction, learning, b) an equivalent learning experience, and c) inclusion of all students.Raes et al. (2020) indicated however that whilst the hybrid-virtual modality offered flexible alternatives to students, it was most demanding and brought a series of complex challenges, with "relatedness to peers and the intrinsic motivation" assessed as the lowest, among all modalities (p.1).Calonge et al. (2019) argued for instance that "inadequately designed peer-to-peer interaction and instructor-to-participant interaction often result in attrition" (p.100) and disengagement in courses facilitated online.This was echoed by Nõuakas et al. (2023), and specifically by Paulsen and McCormick (2020) in a study of three groups of students taking courses fully online, face-to-face (F2F) and both online and F2F from 541 U.S. institutions: They observed that peer collaboration in the online mode was inferior to the F2F mode, and that in-class learning provided "opportunities to engage with faculty that the online sphere" did not offer (p.26).Previous research by Rovai et al. (2005), Ramsey et al. (2016), Huang et al. (2017), Wang and Huang (2018), Conrad et al. (2022), andHirschmann andRiley (2023), has also established that students online often felt distanced, forgotten, disengaged and socially isolated.Additionally, Conrad et al. (2022) indicated that even though "many instructors offered synchronous lectures or sessions . . . it never provides the feeling of being socially present in the same space as other students and the instructor" (p.550).In fact, Penrod (2022) argued that "most colleges and universities that adopted HyFlex teaching during the pandemic did not have the required infrastructure to enable the equitable learning experiences and outcomes that characterize the HyFlex teaching model" (para.20).
The implementation of HyFlex also presents a set of challenges (Gatlin et al., 2021) and opportunities in terms of technical infrastructure, classroom and learning space design.On the one hand, the flexibility of HyFlex requires universities and tertiary institutions to invest in often costly technology infrastructure, such as high-quality video conferencing equipment and Learning Management Systems (LMS), to support seamless communication and content delivery.Additionally, classroom design must consider critical factors like acoustics (Detyna et al., 2023), lighting, and camera placement to ensure equitable engagement for both in-person and remote participants.On the other hand, HyFlex offers opportunities to rethink traditional learning spaces (and teaching/facilitation in those spaces).Institutions can adopt versatile, adaptable, and inclusive classroom layouts that accommodate both physical and virtual students.Moreover, HyFlex encourages instructors to attend continuing faculty development courses/workshops and/or explore innovative pedagogical approaches, such as active learning strategies, teamwork face-to face and online, and flipped classrooms, to maximize engagement and interaction in all modalities.
A search of the literature revealed few studies which (a) seek to answer the research question and (b) use the methodology set for this study 1. in the context of the UAE, and 2. at graduate level.The present study therefore seeks to address this gap, as it contributes to literature by providing a detailed implementation process of HyFlex technology-enhanced classes during and after a pandemic.

Case study
The methodology used in this article is an exploratory qualitative case study approach, as described by Yin (2003).Yin (2018) identifies three conditions that must be considered when selecting from amongst a range of different qualitative research methods: "(a) the form of research question posed, (b) the control the researcher has over actual behavioural events, and (c) the degree of focus on contemporary as opposed to entirely historical events" (Yin, 2018, p. 9).A case study is indicated where the form of the research question is of the "how" type and the focus is on a contemporary event.
The research question (RQ) being examined in this study is of the "how?" type: How do [regulators', environmental, pedagogical] constraints imposed by COVID-19, stakeholder [student, academics, professional staff] requirements, and physical and technological limitations inform an iterative approach to classroom design for HyFlex courses?
The use of an embedded single-case design, incorporating sub-units of analysis makes it possible to provide a rich description of the relationship between the imposed constraints and the classroom design process.This design allows for the exploration of several units of analysis within the case.Specifically, there are four embedded units of analysis, represented diagrammatically in Figure 1.

Embedded unit of analysis 1: classroom design at MBZUAI-a chronicle of constraints
MBZUAI secured accreditation for six graduate programs from the UAE Ministry of Education (MoE) in February 2020 and was subsequently licensed as a university by the MoE in April 2020.Programs were scheduled to be launched in August of 2020 (Fall semester) with both local (UAE) and international students.However, the COVID-19 global pandemic resulted in a decision to postpone the launch until January 2021 (Spring semester).By that time, the UAE Ministry of Health and Prevention regulations were in place (https://mohap.gov.ae/en/)requiring all teaching and learning to be conducted online, until some resolution of the ongoing pandemic crisis could be reached.Hence, the first learning experience of students was of a virtual learning space mediated via the Learning Management System (LMS), Moodle.

Constraint #1: Regulators' Requirements
The UAE has clear guidelines from the Ministry of Education on the delivery of e-learning or courses taught through e-Learning.These regulations govern the methodologies and are incorporated into the Commission for Academic Accreditation (CAA) Standards 2019, Annex 15: Courses Taught through e-Learning (CAA Standards, 2019) details how the institutions must plan the e-learning strategy, what learning resources are required, the learning environment, guidelines on budgeting for e-learning.Additionally, the guidelines provide specifics on program/course delivery, faculty qualifications, experience, training, and workload.
Constraint #2: A Safe Learning Experience -Providing an equivalent learning experience and students having to embrace a form of digital literacy that encompasses "a blend of technical savvy, creativity and complex thinking" in an online environment (Schaffhauser, 2017, para.1).
Face-to-face classes resumed in Fall of 2021, but COVID-19 restrictions required minimum separation of students of 1.5 m.Furthermore, some students were forced to isolate during this period or were prevented from travelling to the UAE so had to continue their studies online.Isolation in the distance learning environment was identified by Lister et al. (2021) as a clear barrier for students' mental wellbeing, and this would be a determining factor in the design of the HyFlex classrooms.A study by Kapasia et al. (2020) on 232 Indian undergraduate and postgraduate students' experience with lockdowns indicated for instance that 42% suffered from stress, depression, and anxiety.Poor internet connectivity (32.4%) and the absence of a suitable home study environment (12.6%) were also identified as barriers to learning (pp.3-4).
Online student-student/faculty-student interactions, and group work via WebEx was initially complicated, as (1) few academics and students had had previous extensive experience with learning and teaching online with the platform, and (2) there were numerous issues with sound and image, which hindered the learning experience.
Change needs to be technology-enabled, not technology driven.Putting pedagogy before technology was therefore a critical element of space design brainstorming and ideation.A study by Hao et al. (2021) indicated that active learning environments "were found to have little influence" (p.1) on student achievement, compared to active learning and teaching.Prior to redesigning classroom space (iteration 1), a task force initially composed of members from Facilities, IT, Finance, Graduate Student Council, academic departments, and Academic Development was formed.Focus groups with students and faculty were also scheduled to assess/gauge initial needs of the in-room and virtual experience.Comments from faculty and students are shown in Table 1.

Constraint #4: Teaching Experience -a) faculty expectations; b) small faculty size
To avoid the need for faculty to teach a class multiple times to groups of no more than 20, a decision was taken to link classrooms so that a class could be delivered in one room and broadcast to other rooms.This came with some requirements and challenges: • Two-way audio and video.
• Video tracking of instructor at the front of the room.
• Video and audio tracking of students asking questions.
Furthermore, at the inception of the university, there were just 11 faculty members.Few had experience with the UAE, teaching in a multi-modal learning environment, or had attended workshops on how to effectively facilitate learning online.
Synchronous attendance was an additional challenge during the first weeks of the online mode, possibly and partly because a) students were scattered around the world and b) WebEx and Moodle were new to students.

Constraint #5: Classrooms -small classroom size
The MBZUAI classrooms were designed for 20 or less students.There were no classroom spaces that could accommodate large class groups.There are only a total of 6 classrooms, one of which could only accommodate 15 students.

Constraint #6: Technology
During the first few weeks of the Spring '21 semester, the IT support Team received occasional challenging requests from academics and teaching assistants wanting to push the boundaries of the software.For several reasons, teaching staff sometimes needed WebEx sites combining to increase flexibility of provision, allow increased access or to simply save time and improve efficiency.Most issues encountered by academics and students therefore referred to technical glitches (image quality, sound, lag in screen share, logging-in, internet connectivity, lecturecapture), as shown in Table 2.
Figure 2 displays the interaction of these 6 different constraints.
While COVID-19 regulations and the small size of the classrooms imposed significant limitations on the physical learning environment, the HyFlex pedagogy provided technology-mediated opportunities for safe teaching and learning practices.The tension between constraints (faculty preparedness, technology infrastructure/support, and students' preferences) and opportunities (improving the teaching and learning experience for both online and F2F students) created an ongoing balancing act that necessitated flexibility and iterative accommodations by both faculty and students.

Participants
Non-random purposive sampling was used.Participants for this study consisted of academics, professional staff, and students at MBZUAI.They were purposefully selected due to different factors: they had experience teaching or learning in the original and first classroom iteration or had been involved in the technical setup of those classrooms.

Data collection
(A) To gather comprehensive data on the use of the learning space, we employed a multifaceted approach, combining interviews, focus groups, and email feedback at Version 0, Iteration 1 and Iteration 2 (see figure 14).Participants were contacted via email (English) or face-toface.Following an initial positive response to participate, an informed consent form was sent.
(1).Interviews: We conducted one-on-one interviews with participants to gain in-depth insights into their experiences and perspectives regarding the learning space (V0 and Iterations 1&2).These interviews were digitally recorded using professional-grade recording equipment.This allowed us to capture the nuances of their feedback, ensuring accuracy and completeness in our data collection.(2).Focus Groups: In addition to individual interviews, we organized focus group discussions to encourage open dialogue and feedback among participants.A total of 4 focus groups were conducted with students and faculty separately, for iterations 1 and 2. Colleagues from academic development, IT, and student affairs were invited to the faculty focus groups.These sessions were also recorded using the same The overall experience I had so far with Zoom was better than WebEx.The following are my experiences: The additional two 55" monitors networked together, strategically places opposite the teacher, made it easy to tell what the remote students see.
When I draw on the interactive monitors, my updates (for the remote students) are in real-time.
Choosing what to show in which interactive monitor rather straightforward.You can easily pick from presentation or smartboard, at a push of a button.The voice-controlled cameras in the room help identify the students sitting physically in the room (gallery-view /individual) and the video tile of whichever student is speaking is automatically enlarged, allowing other students, especially those is the remote mode, to view their in-person peers.
Saving and sharing the drawings and annotations as PDF file was easy.Some students had no issues joining class from their mobile phones, Windows, or Mac devices.(3).Email Constructive Feedback: To accommodate participants' (faculty and students) preferences and schedules, we provided the option to submit constructive and directive feedback (faculty and students input/feedback, figure 14) via email, anytime.This allowed participants to share their thoughts and experiences in a text format.We carefully catalogued and stored these emails for analysis.
(B) Data Processing after Interviews: After conducting interviews, we followed a systematic approach to process and analyze the gathered data.The steps included: (1).Transcription: All interview and focus group recordings were transcribed verbatim.(2).Member Checks: To enhance the reliability of our data, we provided participants with copies of their respective interview transcripts.This member-checking step allowed participants to review their responses for accuracy and make any necessary clarifications or additions.(3).Data Familiarization: Each member of the research team thoroughly read and re-read the transcripts to become familiar with the content.This immersion in the data helped us identify recurring themes and initial patterns.

Data analysis
Data analysis was qualitative and descriptive (Table 3, Table 4 & Table 5).The authors of this study utilized thematic analysis to investigate how participants engaged with WebEx and ZOOM.This regarding what I want to emphasize in the slide.Can we arrange an additional monitor for this in the classroom which will allow presenter view?"F. "With the new flow where all the classroom video systems simply dial in to the same WebEx meeting, everything seems to be working fine for both F2F and remote students.The only hitch now is the need to reshare the presentation after annotating a slide.This causes a few seconds delay".
F. "While basic presentation mode appears to be working OK, there are still unresolved problems with annotation.Even for annotation within PowerPoint (not using Cisco whiteboard), we were never able to get the annotations to be displayed both on the other classrooms screens and getting recorded through WebEx.Sometimes the attached error also crops up and prevents annotations to be visible altogether on the bigger screens".

PS (Professional Staff
). "I experienced some difficulties where the whiteboard would not share . . .The camera in Classroom 4 was not tracking the instructor correctly this morning".
(2) Audio S (Student).Audio clarity."The first batch of students did suffer from low quality audio in lectures" F. "Again the same voice problem during the lecture.It happened multiple times.Video remains stable, but the voice disconnects for the remote students.This is really frustrating for both the students and the instructor".
F. "there seems to be some serious issue with the voice connectivity between the class and remote students".

PS. "
The microphones above the students are still failing to capture their questions . . . .Although the positioning of microphones has been adjusted, additional microphones have not been installed".
S. "The professor is not audible, there is a lot of noise, video keeps lagging, we can barely understand 1/3 of what the professor is saying.Some students attend in person, but it's not possible for unvaccinated students.1st year students told me that their lectures were delivered from the professors' laptop, not from classrooms, and the sound quality was perfect, thus this is a new problem".
F. " . . .currently we cannot use whiteboard in Lecture room 1 due to lack of camera and microphone".
(3) Internet connection PS: "There seems to be packet loss on the call due to which the remote users would have faced audio/video quality issues.I will work with the network team to check the utilization as we need to ensure the internet stability for these calls especially during the peak hours at the university when there is almost 99% utilization of the internet link.At the same time, some of the remote students were also having connectivity issues from their end due to which they were having specific audio/video issues".
analysis followed the systematic six-step approach developed by Braun and Clarke (2006).( 1).
Coding: Following the familiarization stage, our research team independently coded the transcripts.Coding involved categorizing and labelling segments of text with a code, individually, based on emerging themes and patterns.(2).Krippendorff's alpha coefficient (Krippendorff, 2011) was used to determine the degree of inter-rater reliability among coders (.87).Code Review and Consensus: The coded data were then shared, discussed, and clarified among the research team.Through collaborative meetings and open discussions, we achieved consensus on the themes and subthemes that emerged from the data.(3).Theme Identification: Finally, we identified and defined the overarching themes and subthemes that encapsulated participants' feedback and experiences in the learning space (s).These themes formed the basis for our subsequent analysis and findings.This rigorous data collection and processing approach ensured the comprehensiveness and accuracy of our study, allowing us to provide valuable insights into the use of the learning space from multiple sources and perspectives.
6. Hybrid flexible (HyFlex) learning space design: classroom setup Hilli et al. (2019) argued that "the design of learning spaces impacts and shapes the learning activities taking place" (p.66).As such, a model that assumes the use of fixed tables both pre-determines the (1) Presentation technology • Wireless interactive projector and smartboard.
• Interactive projector to network with projectors in other classrooms and allows students to share content from their devices, OR • A mirroring management system (2) Presentation Surfaces • Presentation technology to be augmented with whiteboards that support use by both faculty and students.
(3) Recording technology • Camera(s) and microphone(s) to record faculty together with synchronized media presentation.
(4) Furniture • Mobile chair that can accommodate a laptop and incorporates some storage for personal items.
• Height-adjustable podium that can accommodate a laptop + chair for faculty that complements student chairs.classroom layout and limits pedagogical approaches Student chairs at fixed tables arranged in rows are not conducive to a flexible student experience (Fernandes et al., 2011;Hao et al., 2021).A fixed desk and chair at the front of the room for the instructor is also a limiting pedagogical factor.
The classrooms consisted of a fixed instructor's table and chair, student chairs on castors and a wireless projector presenting content from the instructor's laptop onto a conventional whiteboard (Figures 3 and 4).Hence, this original floor plan which employed student chairs fitted with castors and personal work surfaces instead of chairs at fixed tables was a modest initial step towards a more interactive/collaborative learning environment.

Version 0 physical and technology setup
The original classrooms 1, 2 and 3 (V0) shown in Figures 3 and 4 were equipped with only basic presentation hardware and were intended to accommodate class groups < 20.The instructor logged into a WebEx session.A projector was installed to project content (PowerPoint presentation) on  a whiteboard.Each instructor had to connect his/her laptop using the cables shown in Figure 4 above.A tablet was provided to allow instructors to annotate (algorithms, etc.).The only camera in the room was the instructor's laptop built-in webcam.For a microphone, either the laptop's internal microphone could be used, or a Bluetooth headset worn by the instructor.Students in the room were able to hear the   instructor and see the presentation.Remote students used their individual laptops to hear and see the instructor along with the presentation.However, remote students were unable to hear students in the room as they were too far away from the laptop or headset microphone.Three of the six classrooms retained this layout during Fall '21 and Spring '22.The other three classrooms were identified for a redesign (see Technology Setup below).To satisfy the constraints outlined above, there were both physical and technological aspects to the classroom design.

Iteration 1 -Cisco/WebEx Classroom-Physical Setup
As indicated above, a model that assumes the use of fixed tables and chairs both pre-determines the classroom layout and limits pedagogical approaches and is not conducive to active learning (Finkelstein & Winer, 2020;Finkelstein et al., 2016).Hence, the initial redesign layout proposal employed chairs fitted with castors and personal work surfaces instead of chairs at fixed tables (Figure 5).The fixed desk at the front of the room for the instructor was also replaced by a heightadjustable podium fitted with castors that could accommodate a laptop, allowing instructors to stand in a position such that they could see and engage with all students in the room as well as see and engage with remote students via mirror screens (Figures 6 and 7).

Iteration 1 -Cisco/WebEx Classroom-Technology Setup
For the technology-enhanced classrooms, a Cisco/WebEx installation was chosen because, in theory, it addressed the constraints above and integrated with Moodle (see Figure 8).Table 5 shows all the hardware.
The system installed allowed an instructor to deliver a class in one classroom but broadcast it to other, similarly equipped, classrooms or to the web so remote students could attend the class.It also allowed external (guest) speakers and external (guest) participants to join the class.However, only three classrooms (4, 5 and 6) were equipped with this setup.These classrooms were intended to accommodate class groups of 20 given COVID-19 restrictions.At the conclusion of a class, the WebEx recording was automatically processed in the cloud and then uploaded to Moodle for review purposes or for students who may not have attended the class synchronously, as shown in Figure 8.When classes first started at MBZUAI in Spring 2021, it was in a COVID-imposed online environment.No classroom linking was needed as faculty delivered classes from their offices.Although online classes were not an ideal learning environment, the technology worked well in this scenario.However, this meant that the linked classroom setup was not tested under load during the Spring semester.

Rationale 1: back to a socially distanced campus
When COVID-19 restrictions were relaxed in the Emirate of Abu Dhabi, students and faculty were gradually allowed back on campus.Classroom capacity for the F2F mode was kept small, as per the social distancing guidelines issued by the MoE.Despite a low number of infections among our student and faculty population, a few of our students were quarantined and had to attend classes remotely, synchronously.As strict travel restrictions were still in place in various countries, several students were not allowed to either leave their home country or enter the UAE and had to join the mandatory graduate core and elective classes synchronously or asynchronously.

Rationale 2: a more sustainable technology platform
A strategic decision was made by the University to move away from the Cisco/WebEx platform to a Zoom platform.This was partly due to technical difficulties already experienced in Iteration 1, as described previously.The WebEx platform also relied heavily on integration with Cisco hardware.Zoom, on the other hand, provided a more flexible and sustainable platform that was not hardware dependent.Ideas and suggestions were sought from faculty and colleagues at various local institutions.Analysis of HyFlex settings at institutions in the UAE, the UK, the U.S and Australia, as well as a systematic search of the literature, were performed.

Iteration 2 -Zoom Room-Physical Setup
The physical appearance of the redesigned Zoom Rooms is not significantly different to the Iteration 1 Cisco/WebEx classrooms.At the front of the room are two 86" SMART Board interactive display with iQ and SMART Learning Suite that can be used for presentations or as whiteboards.Two Smart 55" screens on the side of the room allow the instructor to see both remote students and the presentation that is broadcast to them.Importantly, instructors were now able to annotate directly onto presentations which was difficult in the Cisco/WebEx setup as presentations had to be restarted after annotation.Six Poly Studio E70 4K Ultra HD Smart Video Conferencing cameras were installed (including a primary Poly camera located below the 55'' TV to capture the instructor, a secondary Poly camera installed on top of the two front screens to capture onsite students) and a lectern allowed faculty to connect a tablet, a phone, or a laptop to the AV systems via a dock.
Several mock teaching sessions were organized with faculty, students, and professional staff (IT, learning and teaching) to test the technology-enhanced classrooms in F2F and remote modes, and on various devices (smartphones, laptops, desktops, tablets), indoor and outdoor (sound/video quality), with Mac and Windows.Initial feedback identified the following issues: • Quick set-up of the AV system before a class was not as rapid, automated, or seamless as expected.
• Screens' height, that may block students from seeing the presentations/annotations, despite the chairs with castors.
• Sound quality: remote students still had difficulties hearing in-class students.
• Presentation sequence: faculty and students preferred the annotations to be displayed on the righthand side screen.
• Cabling too prominent (walls, floor, TVs), for safety (and cosmetic) issues • Faculty had to turn their back to the audience to annotate the screen.
• Conventional lectern too narrow for multiple devices • Absence of regular whiteboards for teamwork • Remote students video feed was only displayed on one screen (and the instructor's laptop) Four Shure Ceiling Array Microphones and six Bose Pendant Ceiling Speakers were installed to create a more immersive experience for both onsite and remote students, particularly when students move in the classroom to work in teams.Two whiteboards were added to the sides of the classroom as shown in Figure 10.A customized height adjustable ILS Synergy Smart Lectern was also installed in the classroom, with MBZUAI's specific specs: a) Additional HDMI, USB-C, VGA and Audio on both sides, for left/right-handed faculty, b) an additional 24" touchscreen monitor to annotate (and display on screen), c) a Poly TC8 touch control for use with Poly G7500 (Figure 11), d) an electrical tilt mechanism for the 24" monitor, e) a custom side table, and f) two T-bone GM5212 Microphones.Additionally, most cabling for the lectern was hidden (floor, for a more aesthetic look) and multiple junction boxes were installed on the ceiling to distribute the cables for left speakers, right speakers, middle microphones, and the 55" TVs.
To reduce social isolation, increase social presence, sense of belonging (Rusticus et al., 2022), participation, connectedness, and instructor-students/peer-to-peer interactivity and teamwork in synchronous (in person), and synchronous (online) modes, an additional 55" screen was added to the back of the room, as shown in Figure 12.Whilst a "sense of community" (p.758) and social interaction were highlighted by Kinsley et al. (2015) as crucial for graduate students' satisfaction, Zeivots and Schuck (2018) indicated that the "intricate and fluid working practices" of research students had a significant influence or their needs.Two remote feeds (side and back screens, + students' own laptop connected to the Zoom room/breakout rooms) allow a) natural eye contact, and b) in-person and remote students to work together.

Iteration 2 technology setup
The technology setup for the second iteration (Figure 13) is almost identical to the one described earlier in iteration 1 (Figure 8).The only major difference is the design of the main classroom (Figures 9,10 ,11 and 12) and the use of Zoom to stitch together and feed content to multiple rooms, as shown below.
Table 6 shows changes across iterations related to mode, capacity of the classrooms (taking social distancing measures into consideration), the video conferencing and learning management system used, the classroom audio-visual gear.

Figure 9. New HyFlex technology-enhanced classrooms at MBZUAI (iteration 2, design concept, front).
Tens of data points were considered (e.g.ease-of-use, procurement, constraints and limitations (Figure 2), multiple modalities [fully online, Hybrid, etc.], hardware integration, multiples devices, delivery times, availability in the UAE, spacing, furniture, health and safety, luminosity, sound, monitor's orientation and placement, etc. to inform the design at every stage.Detyna et al. (2023) argued that COVID-19 had "produced a push for developing and evaluating new approaches to the university classroom" (p.13).Different methods have been proposed to design and assess learning spaces.Three of the most used in the literature are the principles for designing teaching and learning spaces (Finkelstein et al., 2016, p.29), the activity-centred analysis and design (ACAD) framework (Goodyear et al., 2021) and the Pedagogy-Space-Technology (PST) Design and Evaluation Framework (Radcliffe, 2009, p.13).The authors of this article opted for the rigorous EDUCAUSE Learning Space Rating System (LSRS) V3 (Brandt et al., 2020), as it provides a) "a set of measurable criteria to guide the planning, design, and support of learning spaces", and b) provides a "framework to measure the potential performance of a learning space, that is, to assess what the space enables learners and instructors to do in it.It serves to measure progress toward designing learning spaces that support active learning and engagement" (Felix & Brown, 2011, p.3).Several researchers have used the validated data-driven LSRS (several versions) as a valid and relevant tool to assess learning spaces in a variety of contexts (Flores-Sánchez et al., 2022;Mui et al., 2019;Støckert et al., 2016).

Embedded unit of analysis 4: the EDUCAUSE Learning Space Rating System (LSRS)
Most importantly, and as we are still in the final processes of testing, fine tuning and rolling out our designs (Figure 14), a major advantage of the LSRS is that it measures the potential  The LSRS was used independently by two of the authors to qualitatively assess the three classroom variants (V0, Iteration 1 and 2) against the instrument categories.Each category was scored as 0 points (did not meet the category criterion) through 1, 2 or 3 points depending on the category and the extent to which the assessing author deemed the criterion to be met.The scores were then compared, and any discrepancies discussed so that a consensus was reached.

Findings
Existing research (Detyna et al., 2023;Raes et al., 2020;Weitze, 2016) recognises the critical role played by video, sound quality and internet bandwidth (Cullinan et al., 2021).When classes moved into the classroom at the start of the Fall semester 2021, real F2F and remote classroom and lab testing was possible with faculty, students and professional staff and almost immediately technical issues and negative qualitative feedback related to a) video and visuals, b) audio, and c) internet connection, arose, as indicated in Table 7 (See Results).

Additional Student Feedback indicated the following:
The inconsistent classroom internet connectivity was a major obstacle to my learning.There were times when I couldn't hear or see what was happening in the virtual class, making it impossible to follow along.
Poor sound quality made it incredibly frustrating to engage in discussions.It was often challenging to hear what the professor or other students were saying, and this hindered my ability to participate effectively.
The image quality issues were a significant hindrance to my learning experience.I often missed out on important visual cues, diagrams, or shared content, which left me feeling disconnected from the class.To mitigate poor video quality, all content was routed to the cloud before redistribution to local and remote users.At the time, MBZUAI was relying on a 100Mb internet line.This was upgraded first to 400Mb then to 1Gb.This largely addressed the issues.To overcome the poor pick-up of the instructor's voice, lapel microphones were added.
The laboratories were also equipped with Cisco/WebEx technology (WebEx Codec Pro with 85" display).There were 4 labs with a COVID-imposed capacity of 15 students.However, as labs were not necessarily broadcasted or recorded, the issues inherent in the classrooms were not as critical there.
Feedback was collected from faculty, students, and professional staff.Recommendations to improve design were made, as indicated in Table 7.
Additional qualitative data from this iteration exemplifies the challenging experience faculty had with iteration 1's settings (Table 8).
Additional feedback indicated the following: • "The unreliable classroom internet connectivity was a constant source of annoyance and stress.It disrupted the flow of my lectures and made it difficult to maintain engagement with both in-person and remote students." • "Poor sound quality created communication barriers.It was disheartening to see students struggling to understand what was being discussed, and it hindered our ability to have meaningful class discussions." • "Image quality problems negatively impacted the effectiveness of my teaching.It was challenging to convey complex ideas or share visual materials when students couldn't see them clearly, leading to a less engaging learning environment." These qualitative feedback comments from both students and faculty members highlight the detrimental impact of poor classroom internet connectivity, low sound quality, and subpar image quality on the blended learning experience.Students expressed frustration at not being able to follow virtual classes due to connectivity issues, and this hindered their engagement and participation.Additionally, poor sound quality made it challenging for students to actively contribute to discussions and understand what was being taught.The image quality problems were also problematic, as they prevented students from accessing visual materials and hindered their ability to grasp complex concepts.

Table 7. Faculty feedback
When I taught using WebEx classrooms, I experienced 1 room and 2 rooms setup for two different cohorts.
During the 1-room teaching, all students were physically available in the room while in 2-rooms teaching, students were split between the two rooms.In both cases, students had the option to joining remotely, and some did.The experience I had was good, but I had a few challenges such as • The system occasionally disconnected, and remote students could not hear me.
• When I wrote on the interactive smartboard, remote students could not access content, either partially or in some cases completely.
• When two rooms were connected, I had to mute the additional room as any small noise from that room was heard loudly in the main teaching room.This was easy to do; however un-muting the second room was not easy: a person from the second room had to do it manually on the WebEx system.
• When using WebEx, my laptop was heating, and the laptop fan had to work extra hard to reduce the heat which made a lot of noise.Other colleagues had similar experience.Note: a; our laptops have high specs.
Faculty members shared the sentiment, emphasizing the stress and disruption caused by unreliable classroom internet connectivity.Poor sound quality led to communication barriers and hindered meaningful interactions in the classroom, and poor image quality impacted their ability to effectively convey information and engage students.
In sum, these feedback comments underscore the critical importance of addressing issues related to internet connectivity, sound quality, and image quality in blended learning environments, to ensure a seamless and effective educational experience for both students and faculty.

Assessment using the learning space Rating System (LSRS)-V3
In total, three classrooms (V0 and two iterations) were assessed.Overall, with scores out of 72 expressed as percentages, V0 scored 39%, iteration 1 scored 63% and iteration 2 scored 85%, as shown in Table 8 and Figure 15.
The iterative design process resulted in an almost linear improvement in the assessment scores as measured using the LSRS, as shown in Figure 15.
Student Feedback on iteration 2 -Design Concept and Implementation indicated the following: • "The HyFlex classroom environment at MBZUAI was a total game-changer for me.I loved having the flexibility to attend classes in-person or online.It made juggling my part-time job, family commitments, and coursework much more manageable." • "Collaborating with my peers in hybrid teams was surprisingly smooth and engaging.The screens allowed us to see and interact with remote team members as if they were right there with us, making group projects much more effective." • "The screens and cameras in the HyFlex classroom helped me feel like I truly belonged to the class, even when I couldn't make it in person.I never felt like I was missing out on important discussions or interactions." Faculty Feedback on iteration 2 -Design Concept and Implementation indicated the following: • "Teaching in a HyFlex classroom was a transformative experience.It allowed me to engage with students in multiple ways, tailoring my instruction to their preferences.It made teaching a more dynamic, rewarding, and inclusive process." • "The ability to form hybrid research teams of students who were both in-person and remote was fantastic.It encouraged diverse perspectives and enriched the learning experience for everyone involved." • "One of the most significant benefits of the HyFlex setup was that the screens helped bridge the physical and virtual divide.It was heartening to see students connecting and participating actively (and with their cameras on!), no matter where they were located." These qualitative feedback comments from both students and faculty members underscore the positive aspects of MBZUAI's HyFlex learning space design in terms of flexibility, collaboration, and a sense of belonging.Students appreciated the flexibility to choose their mode of participation, which enabled them to balance various commitments effectively.
Collaborating in hybrid teams was reported to be seamless and productive, thanks to the technology that facilitated interactions between in-person and remote participants.Most notably, the screens and cameras played a pivotal role in fostering a sense of belonging, ensuring that all students felt equally engaged and connected to the class, regardless of their physical presence.Faculty members also acknowledged the transformative nature of the HyFlex approach, which allowed for more dynamic, engaging, and inclusive teaching methods.Overall, these comments highlight the benefits of HyFlex classrooms in enhancing the overall learning and teaching experience at graduate level.

Discussion
The aim of the present research was to a) examine the development, the design, implementation, and evaluation of the new HyFlex technology-enhanced classrooms at MBZUAI, and b) attempt to answer the research question: How do constraints imposed by COVID-19, stakeholder requirements, and physical and technological limitations inform an iterative approach to classroom design for HyFlex courses?Redesigning pedagogically informed learning spaces during a major pandemic, taking into consideration a long list of parameters and internal/external constraints, and most importantly students and faculty needs, preferences, and digital competence, at an institution in the midst of its development was rather challenging.Key to faculty was that the setup should foster social presence, engagement, collaboration, flexibility, and adaptability, with a focus on robustness and simplicity of use.
Our exploratory case study results indicate that the final evidence-based design that was implemented at MBZUAI (item 2.2, Table 8), which included all users' perspectives, continuous feedback and tweaks at each stage of the design (and prototypes), helped offset the issues highlighted by Nõuakas et al. (2023), namely decrease of learning motivation and insufficient digital competencies, by offering engaging and flexible interactive opportunities (Bockorny et al., 2023;Eduljee et al., 2023), relatedness to peers and instructors (Athens, 2023;Raes et al., 2020), autonomy (Mentzer et al., 2023), choice/fluidity (synchronous/asynchronous/F2F/online), equivalent social presence, and stable user-friendly technology, which allowed students and faculty to learn/teach remotely, and access courses online on any device, while being physically in the classroom.In short, it catered to the preferences and needs of (a) our research-intensive faculty and teaching assistants (as shown in item 3.4 Table 8), and (b) the two types of students identified by Crossley (2022): "those who can (and want to) come to campus, and students who can't (or won't)" (para.9).
While the scoring of the last iteration was a satisfying outcome for the many stakeholders involved, opportunities for improvement were clearly identified, such as for instance furniture flexibility, improving accessibility (on-campus/virtual) for students with different physical abilities and the need for a more culturally inclusive campus.
Notwithstanding the constraints and limitations encountered, the LSRS indicates that our purposeful cognitively-inclusive design (item 7.2, Table 8) offered (1) educational opportunities for faculty and teaching assistants to learn about new techniques and technologies, take best advantage of affordances to support in-class/online strategies, and discuss opportunities and challenges with peers and support staff, and (2) opportunities to observe, discuss, evaluate, and coach faculty and teaching assistants in reviewing their teaching practices in the teaching space itself.Gregory et al. (2020) argued that online teaching competencies should be an integral part of teacher training, as "teachers of the future" ought to be "adequately prepared to teach in on-line and blended contexts" (para.11).We therefore advocate the design of continuing professional development workshops on (1) how to facilitate teamwork and group learning and teaching activities and interactions in the HyFlex modality, and (2) how to "translate/transform" face-to-face learning materials into digital formats, as these, in our context, have been identified as most challenging.

Conclusions
In conclusion, HyFlex represents a dynamic and student-centric approach to learning design, bridging the gap between in-person and online education.It promotes inclusivity, engagement, adaptability, and the seamless integration of relevant technology, while also challenging educators and institutions to strategically rethink the design of learning spaces and teaching methods.While it presents challenges in terms of initial setup and resource allocation, the potential benefits for students and the advancement of pedagogy make HyFlex a compelling model for contemporary education.
Overall, this study also strengthens the idea that a) HyFlex classroom design is a complex, lengthy, costly, cross-departmental/collaborative (multi-stakeholder), assumptions-challenging and iterative process, as described in Figure 14, and b) strategically incorporating learning and teaching strategies and tested pedagogical practices into the design of on-campus/virtual learning spaces is critical to ensure stakeholder engagement.
To date, only a limited number of articles describing the rationale, planning, design, and specific technology behind the implementation of flexible HyFlex spaces at graduate level have been identified.This research helps fill this gap.
The generalisability of our results is subject to certain limitations.First, our case study is exploratory in nature.It investigated the potential performance of our technology-enhanced classrooms.As MBZUAI is a graduate-level institution, with few compulsory classes for students to attend, the scope of this study was limited in terms of extensive testing and feedback of iteration 2. Second, although the current study is based on the MBZUAI context and the of our graduate students and faculty, these results add to the expanding field of HyFlex learning space design.More research using controlled trials is needed in other contexts (educational level, university, country).Further research will explore the Zoom room iteration, extensively tested under normal class conditions.This will take place during the Fall 2023 semester when it is expected that it will be possible to connect additional classrooms.Feedback will be collected from instructors and students and evaluated for further input into the design process.A natural progression of this work is also to analyse the impact the classrooms, as active learning environments, have on student learning, emotional, cognitive, and behavioural engagement, and satisfaction.

Figure 2 .
Figure 2. Constraints and limitations of HyFlex classroom design at MBZUAI.

Figure
Figure 3. Original classroom floorplan (V0), used during the first semester of F2F classes, in parallel with iteration 1.

Figure
Figure 5. Original classroom physical setup design of Cisco/ WebEx classroom at MBZUAI (iteration 1, design concept).

Figure
Figure 8. Technology setup design of Cisco/WebEx classroom at MBZUAI (iteration 1, design concept).

Figure
Figure 13.Technology setup design of Zoom classroom at MBZUAI (iteration 2, design concept).

Figure 15 .
Figure 15.LSRS scoring of the three classrooms.

Table 3 . Users' feedback
(1) Video and visuals F (Faculty)." . . .for some parts of my lecture, I would need to have a look at the quick pointers (notes)