motivations

Facilitating development and use of predictive modeling tools for climate-sensitive disease forecasting, early warning and response

Early warning and response systems have been proposed as a cost-effective and reliable solution to improve climate resilience of national health systems. Several spatio-temporal models have been developed for forecasting incidence of vector- and water-borne diseases like malaria, dengue and cholera. While these approaches often employ sophisticated statistical modeling and show good predictive accuracy, they require substantial manual effort to develop for each country and disease setting, and are implemented through bespoke software tools that are not integrated with the routine Health Management Information Systems (often based on DHIS2) that are used in LMICs for planning, monitoring, and evaluating health program. This makes it challenging to operationalize these tools over longer periods and to scale beyond individual country and disease contexts. While there are potentially useful open-source models available, there is a gap in translating this research into automated, packaged tools. In addition, the majority of LMICs face gaps in the availability of the granular, digitized climate data these models require. The result is that early warning and response systems for climate-sensitive diseases have not reached widespread use in LMICs.

To address this gap, the HISP Centre has formed a team of machine learning, software engineering, and information systems experts, led by Geir Kjetil Sandve, to develop CHAP (Climate Health Analytics Platform), a collaborative and open platform for harmonizing climate and health data, as well as importing, training, tuning, assessing, and sharing predictive models for climate-informed disease forecasting. CHAP is developed by HISP to integrate seamlessly with DHIS2, combining health data in national DHIS2 systems with climate, weather, and environmental data from a variety of sources to run and compare predictive models and power early warning systems that are both locally customizable and scalable across disease programs and countries. Rather than limit this platform to DHIS2 systems, CHAP is intended to be available as generic software that climate and health stakeholders and researchers can use to run and assess their models using global and local climate and health data.

Building an interdisciplinary network for global and local collaboration

To provide the interdisciplinary expertise required for this project, the HISP Centre is partnering with subject matter experts from internationally recognized research groups and global health organizations working at the intersection of climate and health. We have also convened an interdisciplinary working group at the University of Oslo including experts on digitalization, machine learning, data/statistics, health and climate, and are forging partnerships with stakeholders and research partners in LMICs.