B8vs8yfw 1487506686

Algorithms could save your life when the next pandemic arrives

Algorithms could save your life when the next pandemic arrives

Vintage Tone/Shutterstock.com

The world has experienced its fair share of deadly infectious disease outbreaks in the 21st century, but there will be more. And when they come, algorithms could help to save our lives.

Infectious disease outbreaks – like HIV/AIDS, SARS, bird flu, swine flu, MERS, Ebola and now Zika – have the potential to cause widespread loss of life, inflict economic damage, and wreak societal havoc. And thanks to airplanes, the pathogens that cause them can now spread to other countries in a matter of hours. So time is of the essence when it comes to managing new outbreaks. The sooner an outbreak is detected, the greater the chance that it might be contained at source, the sooner scientists can begin to develop new medicines, and the sooner that governments can ready their healthcare system.

However, we cannot be sure that all countries will quickly report new outbreaks. A country may not possess a sufficiently robust surveillance system to quickly detect a new outbreak. Even if it does, the process of clinically confirming the presence of a new pathogen can be slow – as pathogens first have to be isolated, analysed, verified, and then passed upwards through multiple layers of state bureaucracy. It is a lengthy process that can also be prone to errors. Worse still, some governments may even try to hide new outbreaks fearful of their negative economic ramifications. There are, in short, no guarantees that a country will quickly report new infectious disease outbreaks to the international community, and that precious time will not be wasted.

Disease vector: travel in the 21st-century. Dushlik/Shutterstock.com

Using the internet

But there is a way to get around this problem. In our information-rich world, there are a growing number of unofficial sources of openly available information that can be scanned for the earliest signs of a new outbreak. A recent study carried out in the Centre for Global Health Policy at the University of Sussex examined three such internet-based syndromic surveillance systems. Their story starts in 1994, when the Program for Monitoring Emerging Diseases (ProMed-mail) first emerged as a new platform to exchange such unofficial information. Using email and the emerging world wide web, ProMed-mail enabled experts around the world to informally exchange information about new infectious disease outbreaks on a daily basis. The system has continued to grow ever since and today has a membership extending to more than 150 countries.

Another influential outbreak detection system is The Global Public Health Information Network (GPHIN). It mostly focuses on news stories in the belief that the media nowadays may report suspicious events and outbreaks quicker than government channels. By monitoring media outlets around the world, it might be possible to pick up early warning signals even before a government makes an official notification.

In late 2002, GPHIN was able to detect signals of the spread of SARS before the Chinese government reported it to the World Health Organisation. Doing so, however, requires GPHIN to continuously monitor, process and filter vast amounts of data acquired from a wide array of media sources (and in multiple languages). That can only be done through the help of algorithms that initially assess reports of disease outbreaks and then flag up the relevant ones to experts for further analysis. Whereas ProMed-mail still largely relied on human moderation, the introduction GPHIN signalled a critical shift to algorithmic disease surveillance.

This trend towards greater reliance on algorithms in outbreak detection continued with the introduction of yet a third system called HealthMap. Unlike GPHIN, HealthMap does not just focus on media reports; instead, it uses algorithms to collate an even wider array of infectious outbreak information taken from both official and unofficial sources. It also goes one step further by using algorithms to break the data down by precise geographic location, which it then displays using a Google Maps plugin. During the recent Ebola outbreak in West Africa, HealthMap was able detect a signal of the Ebola outbreak one week before the Guinean Ministry of Health officially confirmed the outbreak.

The evolution of these new outbreak detection systems shows that our ability to rapidly detect deadly and infectious disease outbreaks is becoming increasingly reliant on algorithms. The new systems are certainly not foolproof. Indeed, they have to make a critical trade-off between accuracy and speed by relying on unofficial and more indirect sources of information. But what they lose in accuracy, they can gain in speed. During the next outbreak, algorithms may well end up helping to save our lives.