Humans excel at causal reasoning, however our “neurological machinery” has limitations, especially when trying to make decisions in situations of elevated risk, complexity, and uncertainty: in these circumstances we resort to biases and heuristics, for example we can see the “follow the leader” approach (aka prestige bias or social desirability bias) to decisions being made by US citizens today.
Artificial Intelligence can provide predictions or classifications but is far from being “intelligent” in the human sense. However, good news for us dumb humans, the emerging discipline of Decision Intelligence (DI) seeks to overcome both the constraints of the human brain and the cognitive limitations of AI. Its adoption is becoming more widespread since the discipline’s inclusion in Gartner’s Key Trends for the last 2 years.
The current Covid-19 pandemic is an exemplar of how Decision Intelligence can be applied in a real-world situation, where businesses need to make informed, high-stakes decisions about which mitigation strategies to invest in. But, to know they are making the right decisions around the novel coronavirus, businesses need to explore volumes of data, multiple epidemiological and infection models, and work out how those apply to their particular workspace. In most cases it’s just easier to apply the minimum regional government guidance and hope this will be sufficient.
The science is pretty conclusive that Covid-19 is transmitted through exhalation of respiratory droplets by an infected person. While large droplets fall on surfaces, small droplets or particles can be suspended in the air for hours causing a significant risk of airborne transmission and made worse by “breathing the same air” as other people in inadequately ventilated, indoor spaces.
While this risk is relatively easy to imagine in small offices or shops, you may think that large office spaces are safer given extensive air conditioning. However, the features of buildings and their influence on air flows could be causing concentrations of the Coronavirus in certain parts of these facilities. Just as facilities where many people congregate, potentially expose their occupants to fire risk, air quality risk, and water safety risk, it is now clear that any large, enclosed structure may pose a bio-safety risk to its occupants under epidemic or pandemic conditions.
Our initial project work in this area has revealed that these risks have been continuously present but overlooked until now. This has only really been exposed through the running of AI over thousands of simulations of the building in question. There are four key building blocks to our solution and we’ve already posted some articles on these to our LinkedIn page, however we’d like to share some insight on the two data-driven aspects of any potential solution: processing of CAD diagrams into simulation spaces, and the use of agent-based modeling, as I describe below.
Firstly, this is all made possible by technology advances and a unique approach to simulation. We are used to seeing immersive environments in computer games, and we have adapted this for the business world. Computer-Aided Design (CAD) has been used for many years in the construction industry, so any modern facility is likely to have multi-layer electronic CAD drawings. As well as the dimensions of the building and interior, these drawings usually contain information on important features such as air conditioning vent locations; this information is relatively easy to access via engineering drawing packages.
To build a model of an interior space so that we can do this analysis is not straightforward. CAD drawings need some special processing to be converted to an interior building space and then used for Coronavirus simulation.
This requires Building Information Modelling (BIM) – the standard for 3D modelling of new building designs. Using this standard, we will be able to more-readily automate the creation of the simulation environment, creating a very concise means to deliver our WorkSafeAI solution.
Secondly, our objective is to advise business leaders on what they can do to prevent Covid-19 transmission through their facilities, requiring a model for how infection spreads within indoor spaces. Although a large volume of research has been done on the epidemiology of the virus, no single efforts have integrated infection, human movement and facility engineering modelling which can be instantly applied to workspaces. Creating this holistic view has been a key product of our prototype development.
Key to being able to suggest effective mitigation advice is understanding the business operation and objectives. From this we can create an Agent Based Model for occupant movement within the facility. Having built a number of standard mitigations into our simulation, including their effect on the Coronavirus transmission, we can run thousands of simulations at a wide range of values of the mitigation parameters to observe the effect and model the outcomes over time.
I mentioned earlier that we are currently focusing on the design of new buildings. However, our solution is also more broadly applicable to other infections transmitted through workplaces. While Covid-19 is an immediate threat, is it just the latest in a series of epidemics across the globe: we have just seen the 11th outbreak of Ebola being fought in Africa, Europe is recovering from a 3-year outbreak of measles and a peak in West Nile fever across Turkey and eastern Europe. SARS-COV-2 has made it to “pandemic” status and has cost the world economy trillions of pounds and according to experts such as World Health Organisation (WHO) Emergencies Director Dr Mike Ryan, “…this virus may never go away”.
Even if we consider Covid-19 as a worst-case scenario for pathogens in humans, we shouldn’t overlook other disease. We already live with the seasonal flu epidemics which, again according to the WHO, costs over 650,000 lives each year for both the flu itself and complications. Although we currently accept these impacts and may have historically lacked the tools to mitigate them, we now have evidence from the southern hemisphere that it largely avoided its flu season due to implementing the same mitigations as have been proven to be effective at the Coronavirus transmission. AI now provides us with the tools to counter the impacts of transmission through the workplace, which also benefits businesses in terms of lost productivity. Society as a whole also benefits from reduced statutory sick pay and healthcare costs.
These business and personal improvement possibilities may appear ambitious. However, new technology including AI and DI raises the bar on what we can accomplish, if it is designed and deployed correctly. It is now possible to reach a new level of safety, productivity, and health.
A previous version of this article originally appeared at datainnovation.ai.