Smaller ai model improves skin lesion diagnosis with missing patient data
Masked Privileged-Information Distillation for Multimodal Skin Lesion Classification Under Missing Clinical Metadata
Computer Vision and Pattern RecognitionMachine Learning
Summary
Skin lesion diagnosis can be improved by combining photos with patient information, but sometimes this information is missing. The authors created a smaller AI model that learns from a bigger model using complete patient data, but trains with some information randomly hidden. This helps the smaller model still do a good job even when some patient details are missing at use time. Their approach works well on a real skin lesion dataset and keeps its focus on the actual lesion when patient data is less available.
What this means in practice
- •For medical device developers: Build compact AI models for skin lesion diagnosis that remain accurate when clinical patient data is partially missing at clinics or in the field.$Commercial implications: Enables sellable AI-powered diagnostic tools that work efficiently with incomplete patient metadata in resource-limited healthcare settings.
- •For mobile health app developers: Develop smartphone apps for skin lesion analysis that perform well even when users do not provide full clinical information.
Tested on one dataset.
Authors
Anirban Barua, Md Mahir Abrar Khan, Ayman Iktidar, Md. Sajjatul Islam
Abstract
Multimodal skin lesion classification combines clinical images with patient metadata to improve diagnostic accuracy. However, complete metadata available during training may be only partially accessible at deployment, and resource-constrained settings additionally require computational efficiency. We address these challenges with a privileged-information distillation framework in which a multimodal teacher trained on complete metadata supervises a 9.2x smaller student trained with randomly masked clinical fields. Clinical fields are masked as whole groups at a per-sample rate drawn from U(0,1), so one training run covers the full metadata availability range. Fusion is residual, with metadata added as a gated correction to an unconditional image base. On the PAD-UFES-20 dataset, distillation under masked training improves balanced accuracy over cross-entropy training at every availability level, by an average of +4.7 points versus +1.8 points without masking. The masked student loses only 7.8 balanced-accuracy points as metadata decreases from complete to absent, compared with 36.6 points for the same student trained on complete metadata, highlighting the role of masked training in graceful degradation beyond distillation alone. Grad-CAM visualizations further show that the masked student's attention generally remains lesion-centered as metadata is withdrawn. The resulting compact model targets point-of-care settings, where clinical metadata is often incomplete.