Abstract The goal of this research is to develop a new spectroscopic technique – Time-dependent Infrared Reflection Spectroscopy Assay (TIRSA) – that will enable us to measure the real-time spectral response of a cell to various pharmacological compounds. The uniqueness of TIRSA is that it exploits infrared-reflective plasmonic (gold) metasurfaces to detect, as a function of time, biochemical changes in close proximity of cellular membranes, as well as cytoskeleton reorganization occurring. The optical spectra obtained over the course of several hours after administering the drug will be used to identify its effects on signaling pathways without any labels because vibrational fingerprints of biomolecules are the natural labels. Unlike time-consuming cytotoxic assays, the TIRSA can be potentially as short as several hours, thereby addressing one of the most serious deficiencies of phenotypic assays: their low throughput. Powerful techniques of machine learning (ML) will be used to process the enormous amount of biochemical information obtained by TIRSA, and to conduct supervised and unsupervised data analysis that will be based on large libraries of spectral cellular responses to single-target chemical compounds such as kinase inhibitors and others. New laser-based hardware for TIRSA will enable high sensitivity, time and space resolution.