COVID TRACKER

About our project

COVID Tracker is a mobile application which predicts the risk of COVID19 infection. This app was mainly developed to collect data from Peruvian regions so it has the most recent information of different variables provided by specialized databases. The novel model for calculating the percentage is based on a multi-layered approach with demographic, environmental and other variables (such as vaccination status and reinfections) supported by recent literature. After obtaining the percentage by GPS, messages are shown for prevention in relation of a certain range. This solution has the potential to be used in people who would like to know the risk percentage and what measures can be applied to them; it can also be used in other parts of the world as long as they count with enough data to be updated on a daily basis. The purpose of this site is to show our study and how our application works.

Download our prototype (in APK) from here:

https://drive.google.com/file/d/1VHPxLjN15btq6yDqWv7zH7Z1qHl9zVR0/view?usp=sharing

Procedure

The steps that a user would follow after downloading our app are the following:

Activate your GPS feature.
Specify your name, age, gender, whether you have got COVID before and if you have got a vaccine.
After a while, if the user is not in Peru, a message will appear saying if you want to choose a place manually; otherwise, the risk appears in the screen from 0 to 100% and with one color out of a palette (blue when the risk is low and red when it is high).
Personalized advice can be seen from an option of the app depending of your risk level, while general advice appears for every user.
Change your configuration setting whenever you want in order to alter your risk factor. You can also change the interface language: English, Spanish, Quechua and Aimara.

The interaction between our users and the objects of the system is described in the pictures.

Fig. 1. Sequence Diagram

Fig. 2. Use Case Diagram

App mock-up

COVID TRACKER uses a palette composed of green - turquoise colors. A logo was designed and a series of mock-ups were created for log in/signing up, calculating the risk, changing settings, displaying general and personalized recommendations and displaying graphs from general COVID variables around the world by linking it to RACE-ESA database. The Figma mock-up can be seen in this link.

Our model

The app's main algorithm was coded in Python and it has a total of three groups of utils identified as layers. The first layer contains epidemiological factors obtained by OpenCovid Peru database and the second layer consists of environmental variables provided by Waqi database; both data are updated daily. Both of them are used in a function called getIndex(city) which allows us to obtain a raw infection rate (also called Rate 1), then we transform this rate to a final output thanks to the third layer, which considers vaccination and reinfection factors. This last layer contains constant data provided by reliable sources.

It is important to mention the parameters that we are considering for this model.

Layer 1: Epidemiological

All the threshold values of these variables were provided by HHS Protect Public Data Hub.

Cases per 100K: Confirmed cases of COVID19 per 100 thousand people. It is defined as the number of individuals who had a viral test via a saliva test, throat or nose swab, and the test confirms the presence of SARS-CoV-2.
Confirmed Deaths per 100K: Confirmed deaths by COVID19 per 100 thousand people. It is one of the most important parameters for any model of COVID19 rate calculator along with the previous variable.
Positivity Rate: Proportion of positive tests within a number of COVID19 tests. It is a useful measure of the virus spreading in a determined community.
Tests per 100K: Number of COVID19 tests conducted per 100 thousand people.
Beds Occupied: Percentage of beds occupied by region until the last Sunday.
Beds UCI Occupied: Percentage of ICU (Intense Care Unit) beds occupied by region until the last Sunday.

Layer 2: Environmental

%PM10: PM(10) refers to dust particles and inorganic compounds dispersed in the atmosphere. A 10 ug/m3 increasing in this variable is associated with a 1.76% increasing in confirmed COVID-19 cases. [1]
%PM2.5: PM(2.5) refers to respirable airborne particles weighing less than 2.5 microns. A 10 ug/m3 increasing in this variable is associated with a 2.24% increasing in confirmed COVID-19 cases. [1]

Layer 3: Vaccine and Reinfection

Vaccine brand and dose: We consider the percentage of effectiveness of the brands Pfizer, AstraZeneca, and Sinopharm to calculate the risk of infection. We also take into account the number of doses that the user applied.
Reinfection rate: This parameter considers whether the user had previously been infected with COVID or not. According to a study of 25 661 health workers, the risk of getting infected the second time is 84% lower than the risk of the first infection; hence, we consider our final rate as the product of 0.16 and the original rate. [2]

Prototype

An APK file was created as a Minimum Viable Product. Its main purpose is to collect your personal data and to predict a COVID19 risk factor. Other features as recommendations and setting were not implemented yet as well as other personal data. You can download it from the following link.

https://drive.google.com/file/d/1VHPxLjN15btq6yDqWv7zH7Z1qHl9zVR0/view?usp=sharing

Main Insights

The correlation between the variables are significant because they have an R number greater than 0.6.
The app obtains data in real time to create the dynamics epidemiological model of risk.
The app calculates the risk for the user at the moment with updated daily data
The app follows the rules of UX and UI design.
The app shows recommendations to prevent the Covid 19 depending on the user's risk scale.
The app pretends to be translated to quechua, aymara (the languages with the most speakers in the country) and English (a global language for tourists).
The use of React Native allows us to be compatible in Android and iOS.
In the future, the app can be replicated in other countries.
Our model will be improved with mobility data from ESA (European Space Agency).

Page updated

Report abuse