A head-confined multimodal configuration (EEG, eye tracking, forehead EDA) predicts held-out learners' five-level engagement ratings at 75.0% balanced 1-off accuracy, 12.0 points above the mode baseline¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-19 (automated, 1 of 1 entry) · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? AdaBoost on the fused 166-dimensional head-confined representation achieved 75.0%±7.0 balanced 1-off accuracy and 1.043±0.158 macro-MAE on 15 held-out forehead participants, exceeding the sensor-free mode baseline (63.0%) by 12.0 percentage points; comparisons are descriptive. → Sidharth Anupkrishnan 2026

Evidence¶

Sidharth Anupkrishnan 2026¶

Sidharth Anupkrishnan, Itir Sayar, Jeongah Lee, Sri Harsha Musunuri, Guan-Ming Su, Madeline Endres, Ravi Karkar, Phuc Nguyen. (2026). E3Sense: Head-Confined Multimodal Sensing of Learner Engagement. arXiv. https://arxiv.org/abs/2609.26569 doi:10.48550/arxiv.2609.26569

q2 · i? · design · r2

Participant-independent evaluation of the E3Sense sensing study: five participant-grouped outer folds trained on both EDA sites and scored 225 segments from 15 held-out forehead participants. The study reports "75.0%± 7.0balanced 1-off accuracy" for AdaBoost against a 63.0% mode baseline; no significance test is reported.

"AdaBoost has the strongest ordinal point estimates in Table 3, reaching 75.0%± 7.0balanced 1-off accuracy and1 .043± 0.158macro-MAE. Relative to the mode baseline, these equal-fold means differ by+12.0percentage points in balanced 1-off accuracy and −0.277in macro-MAE. These comparisons are descriptive."

Discussion¶

Related Claims¶

• Chest-worn ECG provides the strongest cardiac-sensing performance for attention-difficulty prediction, and fusing all cardiac streams does not improve upon chest ECG alone — related
• Predictive performance of sensor configurations does not change monotonically with the number of sensing streams; preferred configuration depends on target metric and sensing cost — related
• Certain multimodal data combinations improve predictive model performance, with audio plus eye-tracking data most effective in one K–8 study — a broader claim this one bears on
• A modality-aware reference model outperforms sensor-free, statistical, deep temporal, foundation-model, and LLM-based baselines for estimating minute-level attention difficulty from wearable sensing — related
• Fold-local top-20 feature selection had mixed effects across model families, supporting the complete 166-dimensional representation as the primary analysis — related
• Model-family rankings depend on the prediction target: AdaBoost leads the five-level ordinal task while LightGBM holds the highest binary macro-F1 — related
• Supplying learners' own engagement-definition codes as context raised exploratory prediction from 64.6% to 71.5% balanced 1-off accuracy in an eighteen-participant cohort