Research map / Vision-language and multimodal / Multimodal LLMs
MedSPOT: A Workflow-Aware Sequential Grounding Benchmark for Clinical GUI
Rozain Shakeel, Abdul Ali, Muneeb Ganie, Tausifa Jan Saleem, Tajamul Ashraf
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 |
|---|---|---|---|
| score with sequential early-termination (TCA) instead of independent step scoring (ablation) | Qwen3-VL-Instruct 8B · MedSPOT-Bench (sequential protocol) | accuracy: 63.0 S1A -> 35.0 TCA | Hurt |
| score with sequential early-termination (TCA) instead of independent step scoring (ablation) | GUI-Actor 7B · MedSPOT-Bench (sequential protocol) | accuracy: 65.0 S1A -> 43.5 TCA | Hurt |
| use GUI-specialized models instead of general-purpose MLLMs | Aguvis/Os-Atlas vs Qwen2-VL, Llama 3.2 · MedSPOT-Bench (sequential protocol) | TCA: 26.7 / 26.6 vs 0.0 | Improved |
| use closed-source frontier model | GPT 5 · MedSPOT-Bench (sequential protocol) | TCA: 2.8 TCA, 16.4 SHR | No clear effect |
| evaluate GUI-Actor with top-3 candidates instead of top-5 (ablation) | GUI-Actor 7B · MedSPOT-Bench (sequential protocol) | TCA: 43.5 -> 33.64 | Hurt |
| evaluate GUI-Actor with a single candidate instead of top-5 (ablation) | GUI-Actor 7B · MedSPOT-Bench (sequential protocol) | TCA: 43.5 -> 28.04 | Hurt |
| vary small/medium/large target-area thresholds (ablation) | all models · MedSPOT-Bench (sequential protocol) | step hit rate by size | No clear effect |
Closest papers by meaning
- MedSG-Bench: A Benchmark for Medical Image Sequences Grounding (NeurIPS 2025 · similarity 0.86)
- WinDeskGround: A Benchmark for Robust GUI Grounding in Complex Multi-Window Desktop Environments (ICML 2026 · similarity 0.85)
- Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis (NeurIPS 2025 · similarity 0.85)
- Towards GUI Agents: Vision-Language Diffusion Models for GUI Grounding (CVPR 2026 · similarity 0.85)
- MedReasoner: Reinforcement Learning Drives Reasoning Grounding from Clinical Thought to Pixel-Level Precision (AAAI 2026 · similarity 0.85)
- MedMO: Grounding and Understanding Multimodal Large Language Model for Medical Images (CVPR 2026 · similarity 0.85)
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.
