Prolonged exposure to toxic heavy metals leads to deleterious health outcomes including kidney injury. Metal exposure occurs through both environmental pathways including contamination of drinking water sources and from occupational hazards, including the military-unique risks from battlefield injuries resulting in retained metal fragments from bullets and blast debris. One of the key challenges to mitigate health effects in these scenarios is to detect early insult to target organs, such as the kidney, before irreversible damage occurs. High-throughput transcriptomics (HTT) has been recently demonstrated to have high sensitivity and specificity as a rapid and cost-effective assay for detecting tissue toxicity. To better understand the molecular signature of early kidney damage, we performed RNA sequencing (RNA-seq) on renal tissue using a rat model of soft tissue-embedded metal exposure. We then performed small RNA-seq analysis on serum samples from the same animals to identify potential miRNA biomarkers of kidney damage. We found that metals, especially lead and depleted uranium, induce oxidative damage that mainly cause dysregulated mitochondrial gene expression. Utilizing publicly available single-cell RNA-seq datasets, we demonstrate that deep learning-based cell type decomposition effectively identified cells within the kidney that were affected by metal exposure. By combining random forest feature selection and statistical methods, we further identify miRNA-423 as a promising early systemic marker of kidney injury. Our data suggest that combining HTT and deep learning is a promising approach for identifying cell injury in kidney tissue. We propose miRNA-423 as a potential serum biomarker for early detection of kidney injury.