Photonic Data Science EN

Our research is seated at the interplay between mid-infrared spectroscopy (IR) and artificial intelligence (AI) techniques. We explore, develop, and employ machine learning (ML) and deep learning (DL) approaches to enhance and extend the versatility and capability of IR in biomedical investigations and clinical applications. Foremost, data pipelines are developed for IR spectroscopy, from measurement and pre-treatment to evaluation and analysis, to extract biomedically relevant knowledge from IR data. Furthermore, strategies for IR spectral data standardization are systematically investigated to make the technology broader widespread in real applications. Moreover, we aim at maximizing the performance of IR spectroscopy by co-modelling between different IR modalities (e.g., FTIR, O-PTIR). More importantly, experimental protocols, from sample preparations to IR measurements, are optimized with AI techniques such as reinforcement learning (RL) and large language models (LLMs). Last but not the least, we use AI for knowledge discovery by linking IR signals and other data modalities to the underlying biomedical properties of a sample and the mechanisms and dynamics behind a biomedical process.