Abstract

The escalating healthcare impact of chronic wounds has prioritized the development of smart dressings for precise diagnosis and treatment. While wearable colorimetric sensors offer convenient diagnostics, their clinical accuracy is severely bottlenecked by ambient light interference. In this work, we report a bioinspired Janus smart bandage with light-adaptive deep learning for precise wound monitoring and accelerated healing. The smart bandage utilizes its Janus substrate to drive unidirectional exudate drainage, maintaining a dry wound bed while continuously delivering fresh analytes to multiplexed colorimetric sensors. We also integrate an LSTM-based deep learning model that autonomously calibrates colorimetric signals under varying color temperatures, eliminating illumination-induced inaccuracies and enabling high-fidelity quantification of wound pH, temperature, and glucose concentration with >96% accuracy. In vivo results demonstrate that this smart bandage not only ensures precise wound monitoring but also significantly accelerates wound closure. The integration of a bioinspired Janus structure with the adaptive deep learning model offers a compelling strategy for smart dressings, potentially advancing high-precision therapeutics in digital healthcare.