Insights

Solving Social Distancing Through Spatial Intelligence

March 18, 2020


George Shaw
Founder / CEO at Pathr™.

We are looking to bring the community together to discuss and try to solve this complex problem. Pathr does not imply in any way that we can solve this problem by ourselves.

In order to understand the spread of disease, it’s common to use statistics and simulation. Epidemiologists have many tools at their disposal to understand the spatial patterns of disease outbreak and spread. These tools are blunt force instruments, however, due to the coarse granularity of the underlying data. We know 5 people in our county that caught the disease; however, what we aren’t focused on are some retrospective questions: was there a way for them to avoid catching it? Is there a way for others to lower their risk given asymptomatic individuals? How do we make our public spaces safer, implement these new social distancing practices, and still try to enjoy our day to day life? The answer lies in Micro-location and spatial intelligence. The power in these simulations below showcase to operators of public spaces important insights and tools to make their spaces safer in our current pandemic.

Most simulations are like these (https://www.washingtonpost.com/graphics/2020/world/corona-simulator/), that are powerful and take into account important macro properties to showcase the grand scale a virus like this can infect. However, we want to bring this closer to home and show simulations of real public spaces that you might visit every day, and how we can #flattenthecurve with simple optimizations like decreasing occupancy or placing hand sanitizer and face mask stations so that you can still visit your favorite physical spaces and feel safe, even during an outbreak like COVID-19.

By accurately simulating physical space and real-world disease transmission in our micro-location framework, we can study the movements of people and how to optimize a space for safety in a pandemic.

We start with the floor plan of a typical physical space like a mall, and we build a digital version of the space. We then spawn our AI agents into this environment at a rate that is typical, based on real world observations. In the case of a mall, our agents represent customers and staff. We assume that some percentage of these agents is infected, either symptomatic or asymptomatic, upon entering the mall — this percentage corresponds to the current rate of infection in the general population, as estimated by the CDC.

Disease is transmitted from agent to agent using a programmable transmission function that would typically take into account the proximity of the agents, as well as the a priori probability of a disease transmission.

Our first simulation depicts the outbreak that could unfold over a day of shopping with just 2.5% of shoppers bringing a disease to the mall with them. By the end, nearly everybody (87.5%) is carrying the virus!

Next, we modeled an increase in social distance. We allowed fewer agents into the synthetic mall, and those agents that were there had an increased tendency to move away from each other, seeking out increased personal space. This simulation saw a lower rate of infection (33%), but still not stellar containment of the virus.

Finally, we used our spatial intelligence to place hand sanitizer and mask-distribution kiosks at optimal locations throughout the mall, and to model the results. It turns out that, at least in our synthetic environment, mask distribution at strategic locations is an effective means of controlling the spread of disease.

Moving forward, we can supplement the simulated environment to assess behaviors as they happen in a space in real-time using existing sensor networks such as security cameras. With existing camera feeds, we can assess the optimal number of people in a space, determine compliance with social distancing, allow for A-B testing effectiveness of policies and procedures, create a risk assessment in real time based on age, and dispatch employees to disinfect zones with high probability of infection based on traffic. Adding in thermal sensors, such as thermometers, would allow us to anonymously track people with an elevated body temperature, where they go, how many interactions they have, and where those occur.

In addition to these real time public health analytics, we can also create AI that assists healthcare. The potential application includes identifying healthcare and critical infrastructure workers versus standard population, routing triage patients based upon initial fever screening (thermal sensor), identifying and dispatching staff to sanitize a room, bed, or bathroom, and determining the optimal location of healthcare professionals and staff based upon the exact needs at any given moment.

We fully understand that we are not experts in epidemiology and do not have the expertise to stop the spread of infectious diseases. However, we believe that showcasing how spatial intelligence and simulation, combined with public health and healthcare expertise can #flattenthecurve in physical spaces. If you are interested in us recommending how to re-work your space to increase social distance, install preventative solutions at the proper locations, please reach out at info@pathr.ai

We are here to help!

Stay Safe,

The Pathr team

Read on MediumInsights