Substance misuse and drug overdose deaths have emerged as major public health threats in the United States over the past two decades. Despite a recent decline in overdose mortality, the overall burden remains alarmingly high, with over 80,000 overdose deaths in 2024 alone. While substantial efforts over a wide range of intervention strategies are underway to abate the harms of opioid and substance misuse in communities on the practical front, many existing and emerging questions require research utilizing advanced analytic modeling tools to guide evidence-based decisions. In this talk, I will present several data and decision analytic modeling studies spanning from measurement to resource allocation in the substance use epidemic. First, to inform local stakeholders with the burden of opioid misuse in communities, we develop a Bayesian hierarchical framework that integrates multiple publicly available data sources to provide nationwide county-level prevalence estimates of opioid misuse. Next, we will discuss a problem of allocating opioid settlement funds that originates from a real-world decision-making setting. We introduce two fairness metrics for the allocation and formulate the fair allocation decisions as a convex optimization problem. Finally, motivated by the challenges encountered by the addiction medicine team in the local health center, we study the prioritization decisions of peer recovery specialists in substance use treatment. We formulate a Markov decision process model to optimize which patients to release early and more efficiently utilize the limited capacity of peer recovery services.