Liberia is amid a water crisis. One third of its citizens – some 1.3 million people -- lack access to even basic water services. Eight in ten, nearly 3.7 million people, don’t have access to decent toilet facilities, escalating the spread of diseases, such as cholera. A 14-year long civil conflict, ravaged much of the country’s infrastructure and severely damaged the water distribution system, leaving most of the country, even in the capital of Monrovia, without access to safe running water. Five years later UNICEF and the Liberian government conducted a national survey and found that only 8% of the population had access to water pipes—none of which were connected to a national water plant—showing that much of the country’s water infrastructure is highly decentralized. Starting in 2014, the Ebola outbreak in West Africa hit Liberia hard, resulting in almost 5,000 deaths across the country. The Ebola outbreak shed more light on the inadequacies in sanitation and water systems, which in turn further elevated improving these systems as a national priority.
The U.S. Water Partnership sent two experts to introduce the power of artificial intelligence to help make Liberia’s future more water secure. Chandler McCann, a data scientist with DataRobot, and Geert Soet, a data project manager with Akvo, introduced a program in which water sector professionals use automated machine learning to develop critical insights and analyses of water point data across the Liberia. Working with government and NGO representatives, their multi-day training program focused on the importance of data collection, sharing, and analysis in evidence-based decision-making regarding water security. They led a workshop with 20+ users from the national government, municipal governments, and relevant NGOs on why data sharing is important, establishing which data points have been collected, and training people on the ground how to use the tools available. Automated machine learning uses artificial intelligence to create models that make predictions about the future of water points using patterns within existing data. For example, these models can be used to forecast which water points are at risk and likely to fail. This allows the government and NGOs to be data-driven and, most importantly, proactive in their management of water points. Those on the ground are not responsible for building these predictive models, which continue to be managed by DataRobot. The government and non-governmental representatives who partook in the workshops, however, are responsible for continuing to upload data to the Water Point Data Exchange (WPDx) and sharing their knowledge of these new tools with other relevant parties.
To get a better understanding of the breadth and depth of Liberia’s water issues the Worlds’ Bank Water and Sanitation in tandem with the Liberian government carried out one of the first national water point inventories on the African continent in 2011. Between the 2011 and 2017 national surveys 40,000 data points from rural parts of the country. Collecting these data points is only the first step. It is the analysis of these data that paves the way for coordinated action, more informed policies, and smarter investment decisions. The analysis of the data points—prior to the Water Experts Program exchange —focused primarily on summarizing trends, rather than transforming the data in ways that directly improve practical decisions being made on a regular basis. However, because of the training that participants received through the exchange, they will be able to directly apply the data available to making practical decisions about the country’s water infrastructure.
Several accomplishments came out of this project, one of the most important being the collective agreement on data sharing and national water standard for Liberia. Standardizing the process of what data need to be collected and how to distribute them to the relevant stakeholders has enormous implications. The National Water, Sanitation, and Hygiene Promotion Committee (NWSHPC) of Liberia has been tasked to coordinate and execute policies based on the analysis of the data collected throughout the country. This mandate provides the needed centralization of these processes to maximize the benefits of applying these tools. This program also brought several systemic challenges to the surface which are not confined to Liberia. Data sharing and frequent collection remains an obstacle in many areas, which impedes the efficiency of large-scale initiatives. This is particularly true in countries, like Liberia, with distributed infrastructure where systems are managed independently of one another. Communicating the importance of data is no small feat, but it is possible using the program in Liberia as a test case. While challenges remain, this program is replicable if scaled for other contexts. In fact, other models have already been built using data from WPDx. By developing relationships with additional governments and relevant stakeholders automated machine learning can support more countries in making evidence-based decisions on water.
The Water Experts Program, a cooperative program between U.S. Water Partnership and the U.S. Department of State, has identified and deployed American hydrologists and experts to high-priority countries on short-term to help improve water security. Experts provide advice to key host-country official through meetings, workshops, briefings, and assessments. These experts specialize in various water-related fields such as, water and climate security, transboundary water cooperation, sanitation, water quality and efficient, integrated management of surface and ground water, grey and green infrastructure, and the water-energy-food nexus.