FACT: Fuzzy Alignment with Comorbidity Topology for Reliable Multi-Label Medical Image Diagnosis

Yingyu Chen, Yongqiang Huang, Yang Qin, Ziyuan Yang, Lang Yuan, Maosong Ran, Yi Zhang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:14625-14650, 2026.

Abstract

In clinical practice, patients often present with multiple co-occurring diseases, yet most existing Multi-Label-Diagnosis (MLD) methods treat diagnosis as a rigid discriminative partitioning task, implicitly assuming that overlapping pathologies are separable. This assumption is problematic in medical images, where identical or highly similar visual observations may simultaneously support multiple disease labels, and disease concepts are inherently correlated rather than independent. Enforcing hard decision boundaries under such overlap suppresses shared evidence, biases feature representations, and ultimately undermines model reliability. To address this limitation, we propose Fuzzy Alignment with Comorbidity Topology FACT, a novel paradigm that reformulates MLD as a fuzzy alignment problem between atomic visual evidence and disease semantic anchors. FACT is characterized by three key features: (1) modeling visual polysemy through shared and reusable atomic visual evidence; (2) encoding disease correlation via semantic anchors structured by comorbidity topology; and (3) employing a metric-based fuzzy membership function for non-discriminative visual-semantic alignment. Extensive experiments on three public clinical benchmarks demonstrate that FACT consistently improves diagnostic performance while delivering clinically plausible predictions. The code is available at https://github.com/yyuChen9/FACT.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26at, title = {{FACT}: Fuzzy Alignment with Comorbidity Topology for Reliable Multi-Label Medical Image Diagnosis}, author = {Chen, Yingyu and Huang, Yongqiang and Qin, Yang and Yang, Ziyuan and Yuan, Lang and Ran, Maosong and Zhang, Yi}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {14625--14650}, 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/chen26at/chen26at.pdf}, url = {https://proceedings.mlr.press/v306/chen26at.html}, abstract = {In clinical practice, patients often present with multiple co-occurring diseases, yet most existing Multi-Label-Diagnosis (MLD) methods treat diagnosis as a rigid discriminative partitioning task, implicitly assuming that overlapping pathologies are separable. This assumption is problematic in medical images, where identical or highly similar visual observations may simultaneously support multiple disease labels, and disease concepts are inherently correlated rather than independent. Enforcing hard decision boundaries under such overlap suppresses shared evidence, biases feature representations, and ultimately undermines model reliability. To address this limitation, we propose Fuzzy Alignment with Comorbidity Topology FACT, a novel paradigm that reformulates MLD as a fuzzy alignment problem between atomic visual evidence and disease semantic anchors. FACT is characterized by three key features: (1) modeling visual polysemy through shared and reusable atomic visual evidence; (2) encoding disease correlation via semantic anchors structured by comorbidity topology; and (3) employing a metric-based fuzzy membership function for non-discriminative visual-semantic alignment. Extensive experiments on three public clinical benchmarks demonstrate that FACT consistently improves diagnostic performance while delivering clinically plausible predictions. The code is available at https://github.com/yyuChen9/FACT.} }
Endnote
%0 Conference Paper %T FACT: Fuzzy Alignment with Comorbidity Topology for Reliable Multi-Label Medical Image Diagnosis %A Yingyu Chen %A Yongqiang Huang %A Yang Qin %A Ziyuan Yang %A Lang Yuan %A Maosong Ran %A Yi Zhang %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-chen26at %I PMLR %P 14625--14650 %U https://proceedings.mlr.press/v306/chen26at.html %V 306 %X In clinical practice, patients often present with multiple co-occurring diseases, yet most existing Multi-Label-Diagnosis (MLD) methods treat diagnosis as a rigid discriminative partitioning task, implicitly assuming that overlapping pathologies are separable. This assumption is problematic in medical images, where identical or highly similar visual observations may simultaneously support multiple disease labels, and disease concepts are inherently correlated rather than independent. Enforcing hard decision boundaries under such overlap suppresses shared evidence, biases feature representations, and ultimately undermines model reliability. To address this limitation, we propose Fuzzy Alignment with Comorbidity Topology FACT, a novel paradigm that reformulates MLD as a fuzzy alignment problem between atomic visual evidence and disease semantic anchors. FACT is characterized by three key features: (1) modeling visual polysemy through shared and reusable atomic visual evidence; (2) encoding disease correlation via semantic anchors structured by comorbidity topology; and (3) employing a metric-based fuzzy membership function for non-discriminative visual-semantic alignment. Extensive experiments on three public clinical benchmarks demonstrate that FACT consistently improves diagnostic performance while delivering clinically plausible predictions. The code is available at https://github.com/yyuChen9/FACT.
APA
Chen, Y., Huang, Y., Qin, Y., Yang, Z., Yuan, L., Ran, M. & Zhang, Y.. (2026). FACT: Fuzzy Alignment with Comorbidity Topology for Reliable Multi-Label Medical Image Diagnosis. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:14625-14650 Available from https://proceedings.mlr.press/v306/chen26at.html.

Related Material