Open-ended Medical Visual Question Answering on MIMIC
58.1F1 ScoreLobA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LobAModel Category=Adversarial Techniques2025.04 | 58.1 | 59.3 | 57.2 | |
| VCD + LLaVA-MedModel Category=Adversarial Techniques2025.04 | 56.6 | 58.3 | 53.8 | |
| PH + LLaVA-MedModel Category=Adversarial Techniques2025.04 | 56.3 | 58.9 | 54 | |
| LLaVA-MedModel Category=Open-source Models2025.04 | 55.1 | 54.1 | 56.5 | |
| CRG + LLaVA-MedModel Category=Adversarial Techniques2025.04 | 54.4 | 56.3 | 55.4 | |
| LobA w/o self-promptModel Category=Adversarial Techniques, Self-prompt=false2025.04 | 54.4 | 55.7 | 53.5 | |
| LISAModel Category=Open-source Models2025.04 | 52.4 | 55.6 | 51.1 | |
| CheXagentModel Category=Open-source Models2025.04 | 42.5 | 45 | 41.6 | |
| BioMedGPTModel Category=Open-source Models2025.04 | 39.7 | 38.7 | 41.2 | |
| MedFlamingoModel Category=Open-source Models2025.04 | 36.5 | 39 | 31 | |
| GPT-4-OModel Category=Proprietary Models2025.04 | 10.9 | 14.3 | 8.9 | |
| Gemini-1.5-Flash-8BModel Category=Proprietary Models2025.04 | 9.1 | 12.7 | 2.4 | |
| GPT-4-VisionModel Category=Proprietary Models2025.04 | 8.7 | 9.3 | 6.7 |