Research map / Vision-language and multimodal / Human and embodied modeling
Condensed Memory Networks for Clinical Diagnostic Inferencing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use condensed hop representations instead of standard multi-hop memory | Memory networks · MIMIC-III, 50 labels, 5 hops | macro AUC: 0.804 -> 0.833 | Helped |
| Use condensed hop representations instead of standard multi-hop memory | Memory networks · MIMIC-III, 100 labels, 5 hops | macro AUC: 0.720 -> 0.767 | Helped |
| Use condensed hop representations instead of standard multi-hop memory | Memory networks · MIMIC-III, 50 labels, 5 hops | Average Precision@5: 0.40 -> 0.42 | Helped |
| Use condensed hop representations instead of standard multi-hop memory | Memory networks · MIMIC-III, 100 labels, 5 hops | Hamming Loss: 0.11 -> 0.05 | Helped |
| Use weighted-average retention of earlier hop states instead of standard KV memory (ablation) | KV MemNN · MIMIC-III, 50 labels, 5 hops | macro AUC: 0.775 -> 0.804 | Helped |
| Use weighted-average retention of earlier hop states instead of standard KV memory (ablation) | KV MemNN · MIMIC-III, 50 labels, 3 hops | macro AUC: 0.761 -> 0.762 | No clear effect |
| Increase memory hops from 3 to 5 (ablation) | C-MemNN · MIMIC-III, 50 labels | macro AUC: 0.785 -> 0.833 | Helped |
| Increase memory hops from 3 to 5 (ablation) | C-MemNN · MIMIC-III, 100 labels | macro AUC: 0.697 -> 0.767 | Helped |
| Increase memory hops from 3 to 5 (ablation) | A-MemNN · MIMIC-III, 50 labels | macro AUC: 0.762 -> 0.804 | Helped |
| Increase memory hops from 3 to 5 (ablation) | KV MemNN · MIMIC-III, 100 labels | macro AUC: 0.679 -> 0.697 | Helped |
| Replace softmax memory addressing with sigmoid addressing (ablation) | KV MemNN · Multi-label diagnosis classification | memory-state suitability | No clear effect |
| Use gated addressing over memory slots (ablation) | C-MemNN · MIMIC-III clinical diagnosis | classification performance | Helped |
Closest papers by meaning
- Memorize and Rank: Elevating Large Language Models for Clinical Diagnosis Prediction (AAAI 2025 · similarity 0.85)
- Visual and Domain Knowledge for Professional-level Graph-of-Thought Medical Reasoning (ICML 2025 · similarity 0.85)
- MEPNet: Medical Entity-Balanced Prompting Network for Brain CT Report Generation (AAAI 2025 · similarity 0.83)
- DiReCT: Diagnostic Reasoning for Clinical Notes via Large Language Models (NeurIPS 2024 · similarity 0.83)
- Beyond Classification Accuracy: Neural-MedBench and the Need for Deeper Reasoning Benchmarks (ICLR 2026 · similarity 0.83)
- Reasoning-Enhanced Healthcare Predictions with Knowledge Graph Community Retrieval (ICLR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.