The coronavirus pandemic has revealed a lack of disaster preparedness on a global scale. This is despite the fact that COVID-19 is not the first pandemic we’ve faced in recent years — between 2011 and 2018, the World Health Organization tracked 1,483 epidemics and pandemics.
Nor will the current pandemic be the last. According to a 2019 report from the Global Preparedness Monitoring Board, pandemic events are on the rise. Yet overall the global response is dangerously reactive, poorly coordinated and undermined by a “lack of continued political will.”
These same issues are evident in our response to other major disasters — whether at the local, national or global scale. Examples include large-scale weather or seismic events or geopolitical tensions.
Fortunately, we have tools at our disposal that we can leverage to our advantage. Among them is AI, which you can implement to support and extend the capabilities of existing systems and personnel at every stage in the disaster life cycle.
AI can monitor disasters in the making.
A global disaster often begins at a hyperlocal hot spot. AI models that look for a spike of mentions or events across a set of identified domains and then cross-reference these against related data points can notify us about potential disasters before they strike.
For example, in our pandemic scenario, AI can mine social media data, news reports, public health data or search engine inputs to track reports of illnesses or increased queries relating to symptoms. These, together with information such as hospital admissions data, may be used to monitor the spread of a disease, raising the alarm when a certain density is reached or when geolocation data shows that individuals from an affected region are traveling into a new region. In fact, an AI system was among the first to recognize the novel coronavirus outbreak — it just needed human eyes to understand the significance.
AI also has value in highlighting potential environmental disasters such as earthquakes. A recent neural net project detected 17 times more earthquakes than traditional methods, helping scientists build more accurate models around seismic activity. AI models can also identify cyberattacks, outages and supply issues, and more, allowing both local and international entities to prepare in case these early warning signs develop into something more.
AI can help sound the alarm.
When disaster does strike, a timely, coordinated alarm is vital. AI can tie into existing technological systems, expanding their reach and offering faster responses and reduced costs.
For example, partnerships between telecommunications operators and AI platforms are being used by countries in disaster-prone areas as an affordable, accessible way to alert citizens about earthquakes and tsunamis. In fact, Chinese company Xiaomi recently integrated an earthquake warning function into its MIUI operating system that will alert users about impending earthquakes before their impact is even felt.
Similarly, AI can help plug gaps in systems such as the U.S. Emergency Alert System, which relies on a cell phone or radio broadcast and is often unable to reach people inside buildings. One recently proposed system involves using AI to analyze CCTV footage in real time to identify emergencies such as natural disasters and sound the alarm within buildings.
In pandemic situations, AI could sound alerts when infected individuals enter an area or when cases reach a density that requires new safety measures. The former was used recently in South Korea to track down and contain “patient zeros” — although not without some intrusion upon personal liberties.
AI can support disaster relief.
Once disaster strikes, AI can be used to support relief efforts on the front lines and beyond. It can map, analyze and model disaster zones to provide updated travel advisories, help mobilize and find citizens, and ensure that disaster response teams know what resources need to be deployed.
For example, following the 2015 Nepal earthquake, Artificial Intelligence for Disaster Response used volunteer-tagged tweets and images to identify urgent needs and infrastructure damage, mobilizing resources as needed.
AI can also be used to extend the capability of emergency workers. For example, IBM Watson’s AI-supported voice-to-text recognition is already in widespread use to help overwhelmed emergency line operators handle large call volumes. The technology converts the call to text and feeds it into a program, which then guides operators on how to respond. In situations such as COVID-19, this type of technology could help support medical offices in gathering patient information to make appropriate triage decisions.
Many current relief efforts are underpinned by AI. A South Korean biotech company used AI to fast track its COVID-19 testing kits, reducing development time from three months to three weeks; similar approaches are likely to be used for treatments. Meanwhile, the U.S. has launched a machine-readable COVID-19 dataset of more than 29,000 articles to help researchers answer vital questions about the virus.
AI isn’t a panacea, but it offers vital support.
AI’s ability to quickly and intelligently analyze large datasets makes it an invaluable resource during times of disaster. By applying it to important questions and time-critical situations, human operators can move fast and make decisions based on all available data. While AI doesn’t do away with the need for coordinated, well-funded disaster preparedness, it can help fill gaps and improve outcomes at every point during the disaster life cycle.
This post originally appeared in Forbes Technology Council.