"Intelligent Scaffolding in Mobile Game-Based Learning Environments;Exploring a frugal implementation model for Medical Training
Lead Research Organisation:
University of Oxford
Department Name: Education
Abstract
This research will examine the role of Artificial Intelligence in Education (AIEd) (Luckin, 2016) for gamebased medical training on mobile platforms. The proliferation of smart phones in all resource settings and the growth of game-based learning provides a wealth of data for social researchers to explore this. This is particularly true for instructional scaffolding, which can benefit from the availability of big data from digital platforms recording the learning progression of students. One such mobile game-based learning platform is LIFE (Life-saving Instruction for Emergencies) project(Oxford University, 2015) which aims to use smartphones and low-cost Virtual Reality headsets to deliver simulation training on the management of medical emergencies in low-resource settings beginning in Kenya. LIFE is a scenario-based mobile gaming platform that aims to train healthcare workers to identify and manage medical emergencies, using game-like training techniques to reinforce the key steps that need to be performed by a healthcare worker to save the life of a newborn baby in distress(Oxford University, 2015). The LIFE project is a collaboration between the University of
Oxford's Department of Education, the Nuffield Department of Medicine and KEMRI-Wellcome Trust Research Programme. Early versions of the LIFE application are being tested with potential users, enabling data to be gathered on how learners progress through simulated scenarios, how they use feedback and how frequently they use the application (Oxford University, 2015). However, little is known regarding context-specific ways to implement scaffolding due to scarcity of studies looking at effectiveness of game-based learning. This is partly attributed to the lack of consensus on approaches, methodologies and descriptions of gaming for educational purposes (Vandercruysse, 2012). This lack of formal specification of necessary inputs and the availability of datasets from the LIFE project makes it a promising case study of how game-based learning implementation can be enhanced to deliver optimal training to medical students. In summary, the research questions of this project are:
1. What feedback theories and methodologies benefit from statistical learning techniques and datadriven scaffolding?
2. How does game-based learning proficiency correspond with learners' knowledge acquisition and beliefs about self-efficacy?
3. In what ways does theory-informed context-aware adaptations from these analytics affect the users' learning experience?
The research will adopt a mixed methods sequential transformative methodological design. The study design will be a prospective cohort with a nested intervention design and a qualitative component to explore how and why the intervention works. The first phase of this research will involve determining how game-based learning has been previously implemented and how theory informed the implementation process, if at all it did. A synthesis of systematic reviews of game-based learning in health sciences will be used to identify adaptive feedback theories that have implemented AIEd. This will help answer the first research question, and ideally, lead to the generation of an overarching implementable conceptual model for mobile based training that will be tested in subsequent phases.
The second phase will explore how identified theoretical concepts can be evaluated in an experimental design. This phase will begin with using Self-Administered Questionnaires (SAQs) administered to study participants before, during and after course duration to gather identify key behavioural traits relevant in the game-based learning process that reflect theoretical concepts identified in phase 1.
Oxford's Department of Education, the Nuffield Department of Medicine and KEMRI-Wellcome Trust Research Programme. Early versions of the LIFE application are being tested with potential users, enabling data to be gathered on how learners progress through simulated scenarios, how they use feedback and how frequently they use the application (Oxford University, 2015). However, little is known regarding context-specific ways to implement scaffolding due to scarcity of studies looking at effectiveness of game-based learning. This is partly attributed to the lack of consensus on approaches, methodologies and descriptions of gaming for educational purposes (Vandercruysse, 2012). This lack of formal specification of necessary inputs and the availability of datasets from the LIFE project makes it a promising case study of how game-based learning implementation can be enhanced to deliver optimal training to medical students. In summary, the research questions of this project are:
1. What feedback theories and methodologies benefit from statistical learning techniques and datadriven scaffolding?
2. How does game-based learning proficiency correspond with learners' knowledge acquisition and beliefs about self-efficacy?
3. In what ways does theory-informed context-aware adaptations from these analytics affect the users' learning experience?
The research will adopt a mixed methods sequential transformative methodological design. The study design will be a prospective cohort with a nested intervention design and a qualitative component to explore how and why the intervention works. The first phase of this research will involve determining how game-based learning has been previously implemented and how theory informed the implementation process, if at all it did. A synthesis of systematic reviews of game-based learning in health sciences will be used to identify adaptive feedback theories that have implemented AIEd. This will help answer the first research question, and ideally, lead to the generation of an overarching implementable conceptual model for mobile based training that will be tested in subsequent phases.
The second phase will explore how identified theoretical concepts can be evaluated in an experimental design. This phase will begin with using Self-Administered Questionnaires (SAQs) administered to study participants before, during and after course duration to gather identify key behavioural traits relevant in the game-based learning process that reflect theoretical concepts identified in phase 1.
Organisations
People |
ORCID iD |
| Timothy Ng'Ang'a (Student) |
Publications
Tuti T
(2020)
Evaluation of Adaptive Feedback in a Smartphone-Based Game on Health Care Providers' Learning Gain: Randomized Controlled Trial.
in Journal of medical Internet research
Tuti T
(2021)
The counterintuitive self-regulated learning behaviours of healthcare providers from low-income settings
in Computers & Education
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| ES/R501037/1 | 30/09/2017 | 29/09/2021 | |||
| 1926841 | Studentship | ES/R501037/1 | 30/09/2017 | 29/06/2020 | Timothy Ng'Ang'a |
| Description | This research set out to evaluate the effect of adaptive feedback within a smartphone-based serious game on the learning gains of healthcare providers from low-income countries. The effect of adaptive feedback on learning gain was found to be 0.644 (95% CI: 0.347 to 0.941, p-value: <.001) when clinician-learning interaction and differential time effect was controlled for. This is moderately higher effect size compared to using standardised feedback. After controlling for the moderating effect of the healthcare provider's self-regulated learning on both time on task, and performance on previous learning opportunities, the odds of getting the next learning attempt correct on the first try was four times significantly higher when the learner received reflective feedback through an adaptive mechanism compared to when they received detailed feedback (or no feedback) through a standard mechanism. This research also planned to generate insights about Self-Regulated Learning (SRL) by healthcare providers in low-income countries on digital learning environments, whether they be computer- or smartphone-based. In summary, using Latent Profile Analysis (LPA) we demonstrated that there are four SRL profiles in this cohort of learners: high, above-average with low help-seeking, average, and low SRL profiles. Consultants had three times the odds of being in the low SRL group, and online computer-based learners had 13 times the odds of being the average SRL profile. The above-average with low help-seeking SRL learners were 89% less likely to have used computer-based learning platform and at least 65% less likely to be clinical officers or consultants. Only 10.1% of healthcare providers who took part in the research would be regarded as having low self-regulated learning. When knowledge gain was considered for learners where that information was available (i.e. LIFE users), compared to healthcare providers in the low SRL profile (who were mostly consultants), the other three SRL profiles had lower learning gains which were never maximised even with more repeated learning cycles. Even within self-directed learning, it might require more instructional support to ensure maximum learning gains are realised. A key concern in the models used is that with increasing model performance, there is also an increase in complexity in demonstrating how the model works or how to transfer the modelling approaches to other learning contexts. However, if the latent factor associated with "knowledge" is agreed to be complex and multi-dimensional, simpler knowledge tracing models are more likely to harbour reductionism, which hinders the extent they can be useful for formative diagnosing of learning needs as illustrated by their poor calibration and discrimination scores |
| Exploitation Route | Linking the level of adaptive feedback provided to healthcare providers to how they space their learning and their clinical level might yield a larger intervention effect both at the group and individual learner level. For the feedback content itself, as an alternative to using reflective hints on what the right answers might be, elaborating why the healthcare providers' responses were wrong might enhance understanding of the learning content. This can be explored in future research. Learning also takes place in social practice, and therefore tutors and colleagues are vital for any healthcare provider's self-regulated learning in their context of practice. However, in this study, the interest was in how individual healthcare providers perceived themselves to have overt control over their self-regulated learning on digital platforms as opposed to how others in the clinical context of practice influence them. Future research in this context can look into if and how the nature of learning in social and collaborative contexts might influence SRL behaviours (e.g. Help-seeking) of healthcare providers' in LICs. This might explain external learning regulation from using clinical peer influence for learning support after catering for individual learning motivation on their preferred digital learning platform. Additionally, as a knowledge tracing intervention that is linked to skill performance mastery, while deep knowledge tracing approach provides a necessary starting point for bridging knowledge gaps in healthcare providers, adding a layer of multi-modal learning using in situ high fidelity simulation training provided by platforms such as Virtual Reality (VR), or Mixed Reality (MR) might arguably enhance hands-on experiential learning. This would transform the learning experience into a more meaningful one aimed at building performance skills, confidence and self-efficacy of the healthcare provider in providing necessary life-saving care |
| Sectors | Digital/Communication/Information Technologies (including Software) Education Healthcare |