אפקה - פרויקטי גמר 2020

Allergy Prediction App

Motivation- The Problem

Allergic rhinitis (AR) is an inflammatory disease of the nasal mucosa, induced by

an immunoglobulin E (IgE)-mediated reaction in allergen-sensitized subjects. AR

is characterized by sneezing, rhinorrhea, nasal congestion and nasal pruritus.

Worldwide, allergic rhinitis affects between 10% to 30 % of the population,

reduces quality of life and has a significant impact on society and economics.

The Solution

Meteorological factors have been shown to affect the physiology, distribution and

amounts of inhaled allergens and therefore affect the incidence of allergic rhinitis

in patients.

Developing an app that will predict the probability of the symptoms appearance

will help AR patients manage their disease. These features will allow the user to

become more familiar with his disease and prepare better for an upcoming

symptoms.

The Method

Developing an algorithm that combines prediction models built by researchers

based on weather parameters and individual user feedback that allows us to

improve the prediction accuracy and to account for the individual nature of this

disease

Shay Sharvit , Shani Marco

Advisor: Dr. Nir Andelman

Medical Engineering

An application that allows the user to view the

allergy symptoms appearance prediction for the next

few days. The app combines prediction models built

by researchers based on weather parameters and

individual user feedback that improve the prediction

accuracy

Data Collection

We gets the weather data from an external API. For this project we decided to

use data from OpenWeatherMap.com.

The Algorithm

Step 1 - CMP Calculation:

based on weather data such as

vapor pressure, average minimum temperature and average

sea-level pressure.

Step 2 - Algorithm No.1 Calculation:

base on Ouyang research in Beijing

where y1, y2 and y3 represent the predicted

values for AR incidence in pollen, dust mite

and mold respectively and ‘X’ represents CMP

value.

Step 3 - Algorithm No.2 Calculation:

calculating a weight function based on the

number of feedbacks previously given, then calculating the final probability taking

into account the average feedbacks, their weights and algorithm no.1 results

Results

Using responsive app developing concepts we were able to develop a web-

based app that functions and looks great both on mobile phone and desktops.