A scoping review of explainable artificial intelligence for medical multimodal data

Multimodal AI models that integrate diverse data such as imaging and clinical records are advancing rapidly in healthcare, yet a significant disconnection persists between these complex predictive architectures and the explainable AI (XAI) techniques used to interpret them. We conducted a scoping review over four bibliographic databases to investigate the use of explainability methods in cross-modal medical AI studies.

Across 82 included studies, the landscape remains dominated by independent feature attribution that scores each modality in isolation, with most studies relying on post-hoc methods that treat the model as a black box. Emerging trends such as visual grounding and model reasoning show promise, but a critical gap remains in explaining the underlying reasoning process. Standardised evaluation is missing in the majority of studies, which rely solely on qualitative measures, and only a minority achieve good reproducibility with a public codebase. We suggest the field transition from individual and post-hoc XAI toward intrinsically explainable designs where reasoning logic is built into the model architecture.

Read the paper in npj Digital Medicine

Recommended citation: Kaiyuan Hu, Xingyue Fu, Yupeng Zhang, Adam G. Dunn, Jinman Kim. (2026). "A scoping review of explainable artificial intelligence for medical multimodal data." npj Digital Medicine. doi:10.1038/s41746-026-02953-3 https://www.nature.com/articles/s41746-026-02953-3