Coronavirus: AI Is Assisting The Global Effort

Coronavirus: AI Is Assisting The Global Effort

By: USQRD in AI

From tech giants to data science start-ups, academic researchers to part-time hobbyists, the AI community is coming together to help combat the corona crisis. We explore how Machine Learning technologies and AI systems are supporting the global effort by monitoring transmission, assisting healthcare operations and maintaining societal order.

Determining Coronavirus Prevalence

Crucially, before forecasters can begin predicting how and where a virus may spread, they must first understand the nature of its current prevalence.

Ideally, this would be done via extensive testing. However, the novelty and sudden pervasiveness of Covid-19 has impeded progress in this regard – the necessary equipment remain in short supply while logistical challenges hamper scale-up operations. As such, authorities must rely on alternative information sources to understand Corona’s prevalence and inform strategic decisions.

Accordingly, tech companies offer valuable insights by utilising natural language processing (NLP) technologies to monitor relevant online discussions. These NLP networks scrape data from Google searches, Facebook mentions and product purchases (thermometers, associated medication etc.) and feed results to machine learning algorithms. In turn, these algorithms deliver predictions regarding corona’s current incidence and help identify potential virus hotspots.

Of course, output should be taken with a hint of scepticism as these simplified estimations are vulnerable to the unpredictable behaviours of the collective crowd. However, the delivered insights can supplement the limited information gained from testing statistics to guide researchers as they make informed estimations.

Similarly, NLP technology can extract and summarize relevant online data to help authorities understand public perception. For example, systems can scrape posts from Facebook or conversations from Reddit to reveal how the public is reacting to the outbreak and inform on the responses being mobilized at the local level. This can then advise policymakers as they make key strategic decisions aimed at curtailing corona’s adverse effects.

Predicting Corona’s Transmission Trends

Coronavirus’ abrupt prevalence and rapid growth has rendered traditional AI-led methodologies for estimating virus spread insufficient.

Crucially, standard machine learning techniques for predicting infections would analyse relevant past data to estimate future trends. While beneficial for diseases such as the flu where researchers can call upon 20+ years of historical data, such an approach would be ineffective for a novel virus unheard of a mere 4 months prior to writing.

Accordingly, researchers from Carnegie Mellon University have formulated predictions by supplementing ML methodologies with the ‘Wisdom of Crowds’ theory. The theory dictates that the aggregation of non-expert estimations made by ordinary people tends to produce remarkably accurate outcomes. Volunteers are presented with graphs depicting a disease’s current trend line and are instructed to add their own line representing their predictions of future scenarios. Of course, being laypersons, individuals are unlikely to perform well. However, they will still be exercising common sense reasoning as they apply knowledge gained from the news, friends and the state of their local area. Crucially, those who are wrong in one direction (i.e. overestimate prevalence in a given region) are offset by the number of incorrect responses in the other direction. Thus, aggregating hundreds of these responses delivers solid estimations that are often far more accurate than those produced by a single expert researcher.

Professor Roni Rosenfeld from Carnegie Mellon University argues that combining this data with M-L algorithms can deliver strong predictions about future infection rates and guide governments towards optimum strategies for combating its spread.

Delivering Critical Supplies

With nation-wide lockdowns being imposed across the globe, the safe delivery of food, necessary supplies and medical equipment has become a prominent concern.

As they strain under a dramatic influx in demand, delivery systems will be looking to local voluntary operations to help ensure deliveries reach beneficiaries in a timely manner. AI-powered smart phone applications will prove a useful tool in this regard. These apps utilize artificial neural networks and self-organising maps to optimize route planning processes. They compute the most efficient order in which to deliver to houses and suggest plausible routes to complete deliveries. Their affordability makes them accessible to local operations, helping ensure their delivery systems are as effective and efficient as possible.

Furthermore, autonomous drones could prove a treasured tool in ensuring a continual service. Indeed, these devices are already transporting supplies between disease control centres and the people’s hospital in Shanghai. With this method, the risk of contamination via delivery personnel is reduced while the continued supply of goods is sustained.

Enforcing Compliance

Maintaining good public health and supporting a stretched NHS requires every citizen to follow stringent rules – stay at home and minimise excursions as much as possible. However, enforcing these measures is a considerable challenge.

Chinese authorities have responded by using machine learning systems to bolster compliance during the coronavirus crisis. One monitoring system, named ‘HealthCode’, analyses big data to assess individuals’ risk of infection based on multiple diverse factors – e.g. travel history, time spent in virus hotspots and potential exposure to infected people. The system then assigns each citizen a colour (green, red or yellow) that signifies whether they should be quarantining or not. Of course, the intrusive nature of the application renders this a highly controversial technique. However, it is likely that similar, less imposing versions may be used in western cultures to minimise the spread of the virus once lock-downs are eased.

Closing Remarks

Seemingly, AI and Machine Learning technologies will be valuable assets beyond the scope of healthcare and epidemiology. We, as a society, will be looking to AI to maintain order and bolster morale during this Covid-19 pandemic.

Indeed, as has been covered in our previous articles, AI systems will prove essential across industries and in all walks of life. They will help preserve supply chain integrity with logistics technologies and agricultural aids. They’ll empower law enforcement by ensuring compliance and uphold journalistic standards by reducing the dissemination of misinformation. They will help residents stay connected in their homes, enable citizens to maintain exercise routines and ensure families stay entertained during periods of isolation.

In short, as coronavirus continues to pose some of the greatest challenges for generations, the importance of AI will be laid to bare now more than ever. Crucially, the current crisis supports the safe integration of sophisticated intelligence systems into all industries, as they will prove essential in keeping economies moving and societies safe during future pandemics.