Online learning
Can AI prediction dashboards help online groups?
Open access · cc by · source: Europe PMC
Combining AI performance prediction with learning analytics reshaped analysis of online graduate groups' collaboration in an engineering course.
Study at a glance
- Design
- Other — AI performance prediction and learning analytics applied to online graduate group work
- N
- N=62 · 62 graduate students in 15 groups over an 8-week course
- Population
- Online graduate engineering students in collaborative writing groups
- Outcome
- Collaboration patterns and reflections under AI-supported performance prediction
Structured fields used in claim comparison tables when every cited study has a complete layer.
Key findings
Most social-network metrics differed significantly across analytic comparisons, including density. The integrated approach is framed as a way to identify at-risk students and support instructional decision-making during online collaborative work.
Methodology
In an eight-week online graduate course, sixty-two students worked in fifteen groups on weekly collaborative writing. Researchers gathered discussion logs, applied AI-supported performance prediction and analytics, and analysed social-network measures plus students' end-of-course reflections.
Limitations
Findings come from one graduate engineering course rather than a multi-institution randomised trial. Predictive analytics are decision aids, not guarantees of later achievement for every learner.
How this study connects
Role on claims
Each row is a claim on a concept or method page where this paper supports, challenges, or qualifies the statement. Roles are hand-checked — not a model guess.
In an eight-week online graduate course, 62 students (43 master’s, 19 doctoral) in 15 groups produced 105 discussion files and 62 reflections while AI-supported prediction and social-network analysis ran on the logs. Most SNA metrics differed across analytic comparisons, including density (F = 4.90). Reflections and network numbers are decision aids in one engineering course, not guarantees of later achievement.
Evidence for the claim as stated.
Not every paper under this label is doing the same analytic job. Ableism and EMBARC analyses interpret identity and culture in talk. EQUIP focus groups and interviews judge feasibility of competency rating (ENACT concerns easing with practice). The AI-analytics paper’s headline numbers are SNA metrics on 62 students, with reflections alongside. Shared 'theme' vocabulary is not a shared finding.
Evidence for the claim as stated.
Sixty-two graduate students in 15 groups (four to five each) produced discussion logs over eight weeks; AI-supported prediction and SNA (density F = 4.90) plus 62 reflections were used to flag at-risk collaborators. Analytic comparisons of network metrics are not a multi-institution randomised trial of dashboards, and prediction is not a guarantee of later achievement.
Evidence for the claim as stated.
Outcomes disagree in kind. ITS moves a fraction post-test. DES and MIM target engagement/interaction. Faculty training moves self-reported attitudes. AI-analytics moves SNA metrics and offers a decision aid. Importing 'the quasi-experiment worked' across those endpoints overclaims.
Evidence for the claim as stated.
Combining AI performance prediction with learning analytics can reframe analysis of online collaboration and support identification of at-risk students.
Evidence for the claim as stated.
Open questions
Tensions this paper is part of
From concept pages' “where studies disagree.” Disagreement means the same question; scope means different assays, populations, or outcomes.
Not every paper under this label is doing the same analytic job. Ableism and EMBARC analyses interpret identity and culture in talk. EQUIP focus groups and interviews judge feasibility of competency rating (ENACT concerns easing with practice). The AI-analytics paper’s headline numbers are SNA metrics on 62 students, with reflections alongside. Shared 'theme' vocabulary is not a shared finding.
Outcomes disagree in kind. ITS moves a fraction post-test. DES and MIM target engagement/interaction. Faculty training moves self-reported attitudes. AI-analytics moves SNA metrics and offers a decision aid. Importing 'the quasi-experiment worked' across those endpoints overclaims.
Related papers in this topic
Same topic cluster — not a recommendation engine.