# Training LLMs with Reinforcement Learning over Digital Twin Representations for Reasoning-Intensive Surgical VideoQA

arXiv: https://arxiv.org/abs/2606.17279 · HTML: https://mlsanity.com/paper/2606.17279/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Clinical and biomedical applications, one of 17 directions in LLM reasoning.
- Builds on 23 of its 38 references that are in the map.
- First posted on arXiv 15 Jun 2026.

## Placed in

- LLM reasoning / [Clinical and biomedical applications](https://mlsanity.com/lr/clinical-and-biomedical-applications/)

## Builds on (most cited first, 23 of 23)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Depth Anything V2](https://mlsanity.com/paper/2406.09414/) (2024)
- [VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding](https://mlsanity.com/paper/2501.13106/) (2025)
- [MedGemma Technical Report](https://mlsanity.com/paper/2507.05201/) (2025)
- [Grounded-VideoLLM: Sharpening Fine-grained Temporal Grounding in Video Large Language Models](https://mlsanity.com/paper/2410.03290/) (2024)
- [Surgical SAM 2: Real-time Segment Anything in Surgical Video by Efficient Frame Pruning](https://mlsanity.com/paper/2408.07931/) (2024)
- [Surgical-VQLA: Transformer with Gated Vision-Language Embedding for Visual Question Localized-Answering in Robotic Surgery](https://mlsanity.com/paper/2305.11692/) (2023)
- [Surgical-LVLM: Learning to Adapt Large Vision-Language Model for Grounded Visual Question Answering in Robotic Surgery](https://mlsanity.com/paper/2405.10948/) (2024)
- [Online Reasoning Video Segmentation with Just-in-Time Digital Twins](https://mlsanity.com/paper/2503.21056/) (2025)
- [Position: Foundation Models Need Digital Twin Representations](https://mlsanity.com/paper/2505.03798/) (2025)
- [SurgViVQA: Temporally-Grounded Video Question Answering for Surgical Scene Understanding](https://mlsanity.com/paper/2511.03325/) (2025)
- [Neural-Symbolic VideoQA: Learning Compositional Spatio-Temporal Reasoning for Real-world Video Question Answering](https://mlsanity.com/paper/2404.04007/) (2024)
- [When to Trust the Answer: Question-Aligned Semantic Nearest Neighbor Entropy for Safer Surgical VQA](https://mlsanity.com/paper/2511.01458/) (2025)
- [Reasoning Text-to-Video Retrieval via Digital Twin Video Representations and Large Language Models](https://mlsanity.com/paper/2511.12371/) (2025)
- [Text-Driven Reasoning Video Editing via Reinforcement Learning on Digital Twin Representations](https://mlsanity.com/paper/2511.14100/) (2025)
- [Recognize Any Surgical Object: Unleashing the Power of Weakly-Supervised Data](https://mlsanity.com/paper/2501.15326/) (2025)
- [Constructing and Interpreting Digital Twin Representations for Visual Reasoning via Reinforcement Learning](https://mlsanity.com/paper/2511.12365/) (2025)
- [SurgSLOT: Segment Anything in Surgical Videos via Semantic Long-term Tracking](https://mlsanity.com/paper/2511.16618/) (2025)
- [SMILE: A Composite Lexical-Semantic Metric for Question-Answering Evaluation](https://mlsanity.com/paper/2511.17432/) (2025)

## Abstract (arXiv)

Surgical video question answering requires multi-step reasoning across semantic, spatial, and temporal dimensions. Existing methods architecturally compress videos into discrete token representations and couple visual perception with reasoning. This approach fragments continuous spatial-temporal relationships and has been shown to restrict multi-step reasoning capabilities. We introduce a reinforcement learning (RL) framework that trains large language models (LLMs) to decouple perception from reasoning by operating over digital twin representations constructed from surgical foundation models. Additionally, we introduce hierarchical representations across frame, temporal window, and procedure levels with probabilistic uncertainty estimates. Finally, we propose a novel reward that combines format validation with accuracy assessment through clinical plausibility evaluation and uncertainty-aware calibration for training. To demonstrate the capabilities of this approach, we introduce REAL-Colon-Reason, a colonoscopic benchmark with 2000 question-answer pairs across three complexity levels. We achieve state-of-the-art performance on REAL-Colon-Reason and two existing surgical VideoQA benchmarks REAL-Colon-VQA and EndoVis18-VQA.