Applications of QSPR and Machine Learning in Molecular Photonics

Abstract In the past decade, quantitative structure–property relationship (QSPR) techniques combined with machine learning (ML) have gained significant traction in photochemistry and photophysics, offering new strategies for the rational design of photoactive molecules. This progress has been driven by advances in artificial intelligence, automation of research workflows, and the availability of large, curated datasets containing experimental and computational data on molecular structures, electronic transitions, spectra, and related properties. These developments enable pre‐synthetic screening, reducing time, cost, and material consumption in experimental studies. QSPR and ML have emerged as powerful tools for accelerating discovery in optical materials and functional dyes design, with applications ranging from UV–vis absorption and fluorescence to intersystem crossing (ISC), singlet oxygen generation, photosensitization efficiency, and phototoxicity prediction. Such models help prioritize candidates and eliminate low‐potential compounds early in development. This review summarizes recent advances in the application of QSPR and ML to molecular photonics, highlighting the most effective statistical methods and ML architectures. Best practices in model development and validation are outlined, and the most informative molecular descriptors and fingerprints for different photochemical processes are identified, providing practical guidance for integrating ML into photonic materials research.