Data collection using unmanned aerial vehicles (UAVs) in construction and heavy civil projects is subject to compliance with strict operational rules and safety regulations. Both the US Federal Aviation Administration (FAA) and the European Union Aviation Safety Agency (EASA) require drone operators to keep the drone in sight and avoid flying near people or other objects. From the perspective of the operator, remaining in standing or sitting position while always looking up to monitor the drone movements can cause awkward body postures, stress, and fatigue. Coupled with the mental load resulting from delegated tasks, this could potentially put the drone mission, people, and property at risk. This research investigates the reliability of using the drone operator’s physiological indexes and self-assessments to predict performance, mental workload (MWL), and stress in immersive virtual reality training and outdoor deployment. A user study was carried out to collect physiological data using wearable devices and design general population and group-specific prediction models. Results show that in 83% of cases, these models can predict performance, MWL, and stress levels accurately or within one level. This paper contributes to the core body of knowledge by providing a scalable approach to objectively quantifying performance, MWL, and stress that can be used to design adaptive training systems for drone operators. Personalized models of physiological signals are presented as reliable indexes to describing the outcome of interest. Scalability is achieved through the application of generalizable machine learning models that learn the interdependencies between physiological and self-assessment inputs and their association with corresponding outcomes.