Validating therapeutic targets for complex diseases requires investigating gene functions within native tissue architectures rather than reductionist in vitro models. Here we present a modality-agnostic AAV-based in vivo high-throughput screening platform capable of delivering knockouts, gain-of-function, and synthetic miRNA knockdowns directly to cells within the diseased environment. This system scales to hundreds of perturbations and is adaptable to diverse species and organ systems. To translate high-dimensional screen data into therapeutic assessment, we established a curated analysis framework that scores single-cell transcriptomes against human disease-specific molecular signatures. This method enables quantitative ranking of targets across distinct biological domains ranging from structural fibrosis to inflammatory signaling, to narrow in on the therapeutic potential of each intervention. We applied this strategy to screen loss- and gain-of-function libraries in a murine pulmonary fibrosis model and within the spontaneously osteoarthritic joints of aged horses, identifying metabolic, anti-fibrotic, and immunomodulatory targets. Importantly, our analysis framework successfully predicted functional outcomes in orthogonal human ex vivo tissue models, including soluble collagen reduction in lung slices and glycosaminoglycan restoration in cartilage, thus establishing a powerful paradigm for prioritizing therapeutic targets by uniting human disease signatures with highly multiplexed in vivo functional genomics.