Four-module AI teaching platform for medical imaging¶
Design · All designs
Evidence · no claims cited
Description¶
A closed-loop AI teaching system comprising four modules: (1) cognitive diagnosis using BKT (pyBKT) and DKT (custom LSTM in TensorFlow) for real-time mastery updates and misjudgment modeling; (2) case generation & selection from public image datasets with synthetic perturbations creating difficulty gradients and confusion pairs; (3) explainable guidance using Grad-CAM saliency maps and LLM-generated three-level cues; (4) adaptive scheduling via custom-coded PPO in PyTorch with a 15-feature state space and custom reward function.
Design Implications¶
Context¶
Requirements¶
- Public image datasets with expert annotations for case generation
- Pilot data for calibrating BKT parameters (forgetting rate and guessing probability)
- Historical interaction sequences for DKT training (20 pilot learners)
Constraints¶
- DKT trained on historical interaction sequences from only 20 pilot learners
Target Learners¶
- Postgraduate medical imaging students
Learning Goals¶
- Fine-grained knowledge point mastery
- Diagnostic accuracy
- Reduction of systematic misjudgment errors
Claims¶
Related Designs¶
Examples¶
-
Key Sources¶
- An Jie, Leng Xi, Liu Yujie, Zhou Jie, Liang Yi, Chen Yuna, Qiu Shijun (2026). Rebuilding ability acquisition in medical imaging teaching: a multi-level AI-driven teaching paradigm based on cognitive-information theory. BMC Medical Education. https://doi.org/10.1186/s12909-026-09002-3
Page metadata
| Field | Value |
|---|---|
type |
design |
id |
four-module-ai-imaging-teaching-platform |
title |
Four-module AI teaching platform for medical imaging |
description |
A closed-loop AI teaching system comprising four modules: (1) cognitive diagnosis using BKT (pyBKT) and DKT (custom LSTM in TensorFlow) for real-time mastery updates and misjudgment modeling; (2) case generation & sel... |
status |
review |
generated |
by process:settle-candidates on 2026-11-05 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — an-jie-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.