A Data- and Knowledge-Driven Approach for Mental Health Policy-Making at World Health Organization

Understanding the current situation of the mental health system in a certain country or region is critical to design the appropriate policy and associated intervention plan to be deployed. This talk shows the application of a new data science process to the identification and characterization of the behaviour of mental health systems in low and middle income countries, as a support for the design of the intervention plans conducted by the World Health Organization to promote the development and improvement of mental healthcare services in those countries. Rather than developing predictive models, this methodology covers a previous goal of understanding a complex system through descriptive data science tools, thanks to a mixed data- and knowledge-driven approach. The proposed methodology covers all steps from the very early stage of data cleaning to the very final step of knowledge production and decision support, bridging the gap between raw data mining results and effective decision-making, and resulting in an understandable typology of mental health systems and an actionable domain ontology for mental health systems in low and middle income countries. Including domain experts in the process is part of the methodology itself and is critical to guarantee both the validity and the understandability of results. Although the process is presented in the particular case of the World Health Organization's mental health data, the methodology can be also applied to other domains that can contribute to improving human life, from policies to improve air quality, to designing healthy lifestyle recommendations, among others.