ORBIT: A Prognostic World Model for Ocular Reasoning Based on Imagined Trajectories

Jiangtao Yan, Yanlin Qu, Yansheng Qiu, Shujian Gao, Wei Yu, Zheng Wang, Xiaodong Sun, Huixun Jia, Diping Song
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:143921-143949, 2026.

Abstract

The longitudinal management of blinding fundus diseases constitutes a Partially Observable Markov Decision Process (POMDP) necessitating a critical precision-risk trade-off between intervention and over-treatment, as true pathology is often obscured in static observations. However, existing paradigms fail to address this complexity. Traditional vision models remain uninterpretable and memoryless, and while Vision-Language Models (VLMs) excel in semantic understanding, they rely on unsafe open-loop text reasoning lacking the anatomical grounding essential for clinical safety. Furthermore, robust learning is hindered by the scarcity of process supervision in sparse clinical records. To bridge this gap, we introduce the Logic-Constrained Abductive Data Engine. Operating on a “Propose-and-Verify” paradigm, it validates MLLM-proposed biomarkers against clinical and temporal logic to reconstruct dense pathological states from sparse outcomes. Building on this foundation, we propose ORBIT, the first ophthalmic Prognostic World Model. Uniquely, ORBIT employs counterfactual visual foresight to imagine anatomical futures under different treatments, anchoring decisions in Closed-Loop Anatomical Verification rather than linguistic probabilities. Experiments demonstrate that ORBIT effectively captures disease evolution and establishes a new paradigm for human-in-the-loop longitudinal decision support and anatomically grounded treatment planning.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-yan26p, title = {{ORBIT}: A Prognostic World Model for Ocular Reasoning Based on Imagined Trajectories}, author = {Yan, Jiangtao and Qu, Yanlin and Qiu, Yansheng and Gao, Shujian and Yu, Wei and Wang, Zheng and Sun, Xiaodong and Jia, Huixun and Song, Diping}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {143921--143949}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/yan26p/yan26p.pdf}, url = {https://proceedings.mlr.press/v306/yan26p.html}, abstract = {The longitudinal management of blinding fundus diseases constitutes a Partially Observable Markov Decision Process (POMDP) necessitating a critical precision-risk trade-off between intervention and over-treatment, as true pathology is often obscured in static observations. However, existing paradigms fail to address this complexity. Traditional vision models remain uninterpretable and memoryless, and while Vision-Language Models (VLMs) excel in semantic understanding, they rely on unsafe open-loop text reasoning lacking the anatomical grounding essential for clinical safety. Furthermore, robust learning is hindered by the scarcity of process supervision in sparse clinical records. To bridge this gap, we introduce the Logic-Constrained Abductive Data Engine. Operating on a “Propose-and-Verify” paradigm, it validates MLLM-proposed biomarkers against clinical and temporal logic to reconstruct dense pathological states from sparse outcomes. Building on this foundation, we propose ORBIT, the first ophthalmic Prognostic World Model. Uniquely, ORBIT employs counterfactual visual foresight to imagine anatomical futures under different treatments, anchoring decisions in Closed-Loop Anatomical Verification rather than linguistic probabilities. Experiments demonstrate that ORBIT effectively captures disease evolution and establishes a new paradigm for human-in-the-loop longitudinal decision support and anatomically grounded treatment planning.} }
Endnote
%0 Conference Paper %T ORBIT: A Prognostic World Model for Ocular Reasoning Based on Imagined Trajectories %A Jiangtao Yan %A Yanlin Qu %A Yansheng Qiu %A Shujian Gao %A Wei Yu %A Zheng Wang %A Xiaodong Sun %A Huixun Jia %A Diping Song %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-yan26p %I PMLR %P 143921--143949 %U https://proceedings.mlr.press/v306/yan26p.html %V 306 %X The longitudinal management of blinding fundus diseases constitutes a Partially Observable Markov Decision Process (POMDP) necessitating a critical precision-risk trade-off between intervention and over-treatment, as true pathology is often obscured in static observations. However, existing paradigms fail to address this complexity. Traditional vision models remain uninterpretable and memoryless, and while Vision-Language Models (VLMs) excel in semantic understanding, they rely on unsafe open-loop text reasoning lacking the anatomical grounding essential for clinical safety. Furthermore, robust learning is hindered by the scarcity of process supervision in sparse clinical records. To bridge this gap, we introduce the Logic-Constrained Abductive Data Engine. Operating on a “Propose-and-Verify” paradigm, it validates MLLM-proposed biomarkers against clinical and temporal logic to reconstruct dense pathological states from sparse outcomes. Building on this foundation, we propose ORBIT, the first ophthalmic Prognostic World Model. Uniquely, ORBIT employs counterfactual visual foresight to imagine anatomical futures under different treatments, anchoring decisions in Closed-Loop Anatomical Verification rather than linguistic probabilities. Experiments demonstrate that ORBIT effectively captures disease evolution and establishes a new paradigm for human-in-the-loop longitudinal decision support and anatomically grounded treatment planning.
APA
Yan, J., Qu, Y., Qiu, Y., Gao, S., Yu, W., Wang, Z., Sun, X., Jia, H. & Song, D.. (2026). ORBIT: A Prognostic World Model for Ocular Reasoning Based on Imagined Trajectories. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:143921-143949 Available from https://proceedings.mlr.press/v306/yan26p.html.

Related Material