ABSTRACT
Knowing how much money is invested in funding mental health research, and in which areas, is essential to inform strategy and track trends to achieve the best allocation of limited resources. However, no comprehensive categorisation system for mental health research is available and, therefore, national and international data on mental health research funding are minimal and not comparable. In this Health Policy paper, we consider the complexities involved in generating such data and propose an approach to classify mental health research grants. We then describe a method using search terms and algorithms for automatic identification and categorisation of mental health research grants listed in a major international database (Dimensions, Digital Science). The automated approach was validated using manually categorised grants data from funders based in the UK, which showed that the accuracy of this approach is satisfactory and comparable to manual classification. Finally, we consider areas of research that are difficult to classify, and how the automated approach can be refined using machine-learning. We argue that agreed definitions and automated approaches could facilitate collaborative reporting of mental health research funders nationally and internationally and improve the strategic dialogue in this area of research.
