Research map / Vision-language and multimodal / Cross-modal learning
DeepGB-TB: A Risk-Balanced Cross-Attention Gradient-Boosted Convolutional Network for Rapid, Interpretable Tuberculosis Screening
Zhixiang Lu, Yulong Li, Feilong Tang, Zhengyong Jiang, Chong Li, Mian Zhou, Tenglong Li, Jionglong Su
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Add CVPEM cross-validated probability embedding (ablation) | CNN backbone · TB cough dataset | AUROC: 0.809 -> 0.822 | Improved |
| Add IMDM multimodal fusion module (ablation) | CNN backbone + CVPEM · TB cough dataset | AUROC: 0.822 -> 0.889 | Improved |
| Add CM-BCA bidirectional cross-attention (ablation) | CNN + CVPEM + IMDM · TB cough dataset | AUROC: 0.889 -> 0.901 | Improved |
| Add TRBL false-negative-weighted loss (ablation) | DeepGB-TB without TRBL · TB cough dataset | AUROC: 0.901 -> 0.903 | No clear effect |
| Add TRBL false-negative-weighted loss (ablation) | DeepGB-TB without TRBL · TB cough dataset | F1-score: 0.845 -> 0.851 | Improved |
| Set TRBL lambda to 3 instead of 1 (ablation) | DeepGB-TB · TB cough dataset | AUROC: 0.901 -> 0.903 | No clear effect |
| Add audio to tabular-only model (ablation) | DeepGB-TB · TB cough dataset | AUROC: 0.840 -> 0.903 | Improved |
| Add tabular data to audio-only model (ablation) | DeepGB-TB · TB cough dataset | AUROC: 0.825 -> 0.903 | Improved |
| Use DeepGB-TB instead of late-fusion CNN-LightGBM | CNN-LightGBM Ensemble · TB cough dataset | AUROC: 0.792 -> 0.903 | Improved |
| Use DeepGB-TB instead of Qwen-Omni 3B | Qwen-Omni 3B · TB cough dataset | AUROC: 0.900 -> 0.903 | No clear effect |
| Exclude Mel-Spectrogram feature (ablation) | DeepGB-TB leave-one-out · TB cough dataset | AUROC: 0.887 -> 0.902 (p<0.01) | Mixed |
| Exclude Heart Rate feature (ablation) | DeepGB-TB leave-one-out · TB cough dataset | AUROC: 0.887 -> 0.882 (p=0.01) | Hurt |
Closest papers by meaning
- Revisiting Computer-Aided Tuberculosis Diagnosis (TPAMI 2023 · similarity 0.85)
- Learning Low-Rank Feature for Thorax Disease Classification (NeurIPS 2024 · similarity 0.82)
- Deep Reinforcement Learning for Early Diagnosis of Lung Cancer (AAAI 2024 · similarity 0.81)
- NaFV-Net: An Adversarial Four-view Network for Mammogram Classification (AAAI 2025 · similarity 0.81)
- Adversarial Learning Under Hybrid Perturbations for Robust Acute Lymphoblastic Leukemia Classification (AAAI 2025 · similarity 0.80)
- Cross-Modal Interactive Perception Network with Mamba for Lung Tumor Segmentation in PET-CT Images (CVPR 2025 · similarity 0.80)
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.
