Abstract: Radiotracers containing [18F]-labeled electron-rich aromatic rings are among the most highly sought- after PET imaging agents but have been historically challenging to synthesize. Recent efforts have sought to improve the late-stage labeling of (hetero)arenes with [18F]fluoride. In particular, transition metal-mediated reactions using high molar activity [18F]fluoride have changed the way radiochemists form C–18F bonds, and copper-mediated radiofluorination (CMRF) has proven one of the most versatile of approaches. In the previous award, we reported 10 new CMRF reactions, validated them for Good Manufacturing Practice (GMP), and used them to synthesize FDA-approved clinical doses. However, despite many successes, several key challenges remain for the widespread clinical application of CMRF: (a) many important organic scaffolds are incompatible with existing CMRF processes, (b) yields of automated CMRF methods are typically moderate and thus unsuitable for commercial distribution, and (c) disposable cassette technologies are not available for CMRF on automated radiosynthesizers, limiting radiotracer production for routine clinical use. All of these challenges will be addressed in this renewal proposal. The overall objective is to develop robust methods for clinical production of diverse PET radiotracers. Our central hypothesis is that CMRF is uniquely positioned to enable us to achieve this goal. The proposed research will identify new reactions to radiofluorinate scaffolds that are incompatible with existing CMRF (Aim 1), use cutting edge machine learning techniques to improve automated CMRF yields (Aim 2), and develop cassette technologies for reliable GMP production of clinical radiotracers using CMRF (Aim 3). The research is significant because it entails development of methods for radiolabeling bioactive molecules containing functionality that is incompatible (or low yielding) with existing CMRF (heterocycles like pyridine and morpholine, drug molecules like GW405833), as well as optimized automated methods and cassettes for radiotracers that have been challenging to access for decades (e.g. [18F]FDOPA). The viability of the proposed efforts is supported by extensive preliminary results that provide groundwork for the exciting new research directions. Our team has been collaborating for 7 years and our expertise in transition metal catalysis (Sanford), radiochemistry (Scott), and machine learning (Doyle) uniquely positions us to accomplish the proposed research. The project goals will be accomplished through a variety of innovations including: (1) developing methods for labeling challenging electron-rich (hetero)arenes from new precursors (C–H bonds, aryl halides), (2) the first application of machine learning to radiochemistry, and (3) development of automated cassettes for conducting CMRF using the newest generation of radiosynthesizers designed for plug-and-play production. Overall, this project will deliver multiple new methods ...