2025-10-27 Structure-activity landscapes and synthetic accessibility in machine learning-guided organic semiconductor discovery

Norah Salem Alsaiari, Bilal Siddique, Ejaz Hussain, Muhammad Faizan Nazar, Talal M. Althagafi, Amir Badshah, M.S. Al-Buriahi,
Structure-activity landscapes and synthetic accessibility in machine learning-guided organic semiconductor discovery,
Journal of Photochemistry and Photobiology A: Chemistry,
Volume 473,
2026,
116877,
ISSN 1010-6030,
https://doi.org/10.1016/j.jphotochem.2025.116877.
(https://www.sciencedirect.com/science/article/pii/S1010603025006173)
Abstract: The search for efficient organic semiconductors within vast chemical spaces is often a tedious and resource-intensive task. In this study, we present a machine learning–based framework for property prediction and compound screening, aimed at accelerating the discovery process. Fast learning algorithms were applied to predict light absorption behavior, and multiple models were evaluated to identify the most effective approach. Using an automated design strategy, a diverse collection of novel polymers was generated, and their optoelectronic properties were forecasted with the trained models. Compounds exhibiting red-shifted absorption were prioritized for selection. To gain deeper insight into structure–property relationships, the Structure Activity Landscape Index (SALI) was employed alongside chemical similarity analyses using cluster plots and heatmaps. Finally, the synthetic feasibility of the proposed candidates was assessed through Synthetic Accessibility Scores (SAS), which indicated that the majority of selected polymers are likely to be readily synthesizable.
Keywords: Machine learning; organic photodetectors; polymers; absorption maxima