Between 2000 and 2019, the United Nations Office for Disaster Risk Reduction (UNDRR) estimated a record number of global catastrophic disasters, with the majority of the resulting deaths occurring in developing countries. Sudden-onset disasters lead to the displacement of 25 million people globally each year and cause billions of dollars in damages that disproportionately impact low- and middle-income countries. The United Nations estimates that economic losses from global disasters increased sevenfold between the 1970s and 2010s. Climate change has also increased the frequency and severity of global disasters. This has had a variety of consequences, including increasing the vulnerability of low-lying megacities and deltas in countries such as Bangladesh, the Maldives and the Philippines.
Research indicates that the last-mile portion of medical delivery systems in low- and middle-income countries is often costly and inefficient, resulting in delayed access to blood and emergency medical supplies, negative health outcomes and many preventable deaths for women, children and residents of rural communities. To address this issue, AI-powered UAVs, commonly known as drones, are now being used to deliver critical healthcare supplies to hospitals and medical centres. Drones can overcome challenging terrain and limited infrastructure to rapidly deliver lifesaving medications and supplies, including blood samples, gloves, oxytocin, vaccines and other types of medication. For instance, Zipline, a company based in the United States, has been providing a 24-hour medical commodity service in hard-to-reach parts of Africa since 2016. Zipline’s drones can make up to 600 on-demand delivery flights daily and serve about 22 million people across over 2,000 healthcare facilities in sub-Saharan Africa.
AI-powered UAVs can provide last-mile medical deliveries in low- and middle-income countries by providing extended emergency resource delivery and dispatch, regardless of topographic and infrastructural obstacles. UAVs utilize AI to analyse weather conditions, terrain and air traffic restrictions to determine optimal flight paths for deliveries, monitor opportunities to improve flight systems and make real-time decisions to reduce the risk of accidents. They also have widespread potential to lead global immunization campaigns and increase vaccination rates in last-mile health facilities in developing countries due to their high-speed logistical efficiency and reduced lead time.
AI for Good in focus: Integrating AI into Geographic Information Systems workflows
“AI really helps us see things that we can’t see [in an image or in text] to help us learn new workflows in each model. […] It helps accelerate our reading capabilities, analysis and creation of different layers.” - Rami Alouta, ESRI.
AI is primarily used in geospatial analysis in two key ways: (1) enhancing GIS workflows by rapidly automating data extraction at scale, and (2) powering AI assistants and agents that understand user intent, generate insights, perform GIS tasks and create geospatial content. The AI for Good webinar, “Unlocking the Power of Geospatial Artificial Intelligence for humanitarian use cases,” explores how GeoAI solutions can accelerate geospatial analysis across a wide range of applications. In this session, Rami Alouta, Geospatial Enterprise Systems Expert at ESRI, explained how AI integrates complex ML and deep learning workflows to enable cohesive analysis of large-scale geospatial data.
Over 150 pre-trained models and foundational models to create, manage and analyse spatial and geographic data are currently available on ArcGIS, a growing open-source deep learning library developed by ESRI. These models provide powerful, ready-to-use tools, such as flood detection, imagery and remote sensing, and time-series forecasting, that allow users to leverage AI to prevent and mitigate humanitarian harms. Object detection and text extraction features accelerate the analysis of multidimensional models to uncover hidden insights that enhance decision-making and support timely intervention.
