Share

The COVID-19 Detect & Protect Challenge

Designing a low-cost, open-source system for detecting community spread of COVID-19

28th Apr 2020

With extra time on my hands, and a box full of open source COTS sensors and displays, I figured it might be fun to play around with a submission to the UNDP "COVID-19 Detect & Protect Challenge" I guess this counts as Article 02 in a series of posts on household inventions!

This is very early days, and this post will continue to evolve as I work on this project. Right now, there are quite a few challenges ahead, not least of which is getting the missing components that I need for my idea. Unfortunately Arduino's aren't considered essential deliveries on Amazon, so there is a long lead time to build with parts I don't already have on hand.

The Problem

From the official website: "The UNDP is calling on Hackster's global community to support developing countries through the sharing and transfer of open source technology. This challenge has three priority actions:

  • Design replicable, low-cost tools and resources to aid viral detection
  • Flatten the curve in communities with preventive solutions
  • Reduce the disease's impact on the economies of these vulnerable areas

The Solution

I'm a scientist; because I invent, transform, create, and destroy for a living, and when I don't like something about the world, I change it.

Rick Sanchez

Based on what I've learned from working with the UNDP for a few years now on various projects, the key to my submission is likely going to be around building a low-cost IoT network for developing areas of the world that have low or no traditionally connectivity. My prototype will be two parts:

  1. A low-cost, non-contact, temperature sensor built with off-the-shelf parts and a 3D printed shell. Basically, a home-built version of those little gun temperature sensors that airport security uses to check if you have a fever when coming off an international flight. Mine probably won't be wildly accurate, but the goal is to get within +/- .5°C. Enough to check if you have a fever. Importantly, these non-contact temperature guns are sold out EVERYWHERE and are also too expensive to be used in many developing areas of the world. This temperature gun would also have a long-range LoRA radio chip that sends the anonymized data to the community hub.
  2. A LoRA based community IoT network that syncs the data from these temperature guns up in a single location that does have an internet connection (wired ethernet or WiFi). This location will be the community 'hub' for recording both individual and group average body temperature readings, looking for spikes in fevers that might indicate an increase in an extremely contagious fever-based virus such as COVID-19.

While it's difficult to claim that any commercial off-the-shelf sensors will be sensitive enough to be medically accurate. However, I imagine that the real benefit of this platform is by leveraging large data sets over an Edge IoT network, looking specifically for group fever spikes. With a virus like COVID-19, it seems that the 'community' average temperature is actually more important than any single temperature reading.

I'm putting a great emphasis on designing something that I believe would be implementable in any developing community, using relatively simple to find off-the-shelf parts and 3D printed shells, and also not rely on 3G/4G/WiFi to push the data to the cloud. By using LoRA as the communication framework, which has a 2-3km range in urban environments and 5-6km range in rural settings, each community can collect and store group data for a geographic zone. Additionally, these developing communities are often those who do not have ready access to a single hospital or clinic where this type of community spread data would be collected.

The Build

Update - May 18

Right now the project is coming along, built around the Adafruit Feather 32u4 RFM96 LoRa Radio - 433MHz module powered by a 4400 mAH LiPo battery. Still using the AMG8833 8x8 IR Thermal Camera as a sort of 'display', I've managed to enable a 64x64 interpolation working, giving me significantly higher resolution. However, because the sensor is only accurate to +/- 2.5° C, I'm going to rely on the GY-906 MLX90614 Non-contact Infrared IR Temperature Sensor as a primary temperature variable tool as it offers a +/- .5° C accuracy - good enough!

The display is a Adafruit 1.44" Color TFT LCD Display with MicroSD Card breakout (ST7735R), and there is also a laser diode for knowing where you're pointing the thing! The big change since last post was that I've decided to put the project into a rather, ahem, interesting shell. I think it would be pretty fun to have a world-changing, life-saving, IoT enabled LoRA smart contactless thermometer inside of a Rick & Morty Portal Gun. Seriously - I think this would be both helarious and awesome.