Local Air Quality Forecast

Local Air Quality Forecast

Low-Energy Indoor/Outdoor "Deep" Casts of Air Quality powered by NCS

Artificial Intelligence

  • 0 Collaborators

  • 0 Followers

    Follow

Description

Context

According to the Centers for Disease Control and Prevention, asthma is a common disease on the rise with significant health care costs. Nearly 1 in 12 Americans (26 million) have asthma. In the last decade, the proportion of people with asthma grew by nearly 15% [1]. The concern is even higher in Hawaii: in 2008, an estimated 95,665 adults had asthma. Also, according to the same source, adult lifetime asthma prevalence was 16.1% and adult current asthma prevalence was 9.6% compared with U.S. rates of 13.3% and 8.5% respectively [2], but not constrained to it. From wildfires in California [4] to an alarming reduction of life expectancy by 3 years in China [5], air quality is a concern closer to us than we would expect.

According to the United States Environmental Protection Agency (US EPA) slightly above average air quality due to excessive ozone, carbon monoxide, sulfur dioxide, and particulate matter (PM 2.5, PM 10) suffice to put at risk children, older adults and people active outdoors [6].

Motivation

Despite alarming consequences and percentage of the affected population, air quality is not something often seen indoors or outdoors as being measured like a thermometer. This may partially be due to the fact that when first introduced (2003), "NowCasts" was calculated based on a 24 hour time window, and little was known about how to report: PM 2.5 hourly to the public, and the implication of short-term exposure to high concentrations of PM 2.5 on health [7]. More recently, however, new methods have been proposed to not only measure the current air quality in a given location through instant cast [8], but also forecasting it [9].

Despite the exciting progress that has been made in providing air quality measurements to account for when air quality changes rapidly, such as nowcasts still being dependent on the availability of local station data, which greatly reduces accuracy and user satisfaction (and safety!). For example, EPA's own APP has been criticized [11] in user reviews for inaccuracy due to stating "good air quality" despite a large amount of smoke being seeing outside the reviewer's house. What if the instant and forecasts not only relied on station data (sometimes miles away), but also on a low-power device and air quality sensors?

Proposal

This project proposes a low-energy air station built on top of a Rhasberry Pi for a local sensor measurement, whether indoors or outdoors, and powered by Movidius NCS to provide nowcasts sensible to the local surroundings.

Low-energy and low-cost projects using Raspberry PI to create air quality stations have been proposed and implemented all over the web [12-23], including for EPA workshops for 5th-12th graders [24] and even combining both Arduino and Raspberry Pi's [25]. This project builds on top of existing ideas of measuring air quality indoor and outdoors by using the Movidius Neural Computer Stick to extend its ability to forecast.

I also see opportunity in other related projects and studies limitations, such as those involving people who suffer from asthma. For instance, Apple's asthma study using a mobile device proved accurate when compared to existing research [26], however, a major limitation of the study was the number of participants when compared to the initial pool (from 50k downloads to 2,317 users who filled multiple surveys). I believe the main limitation of interest by participants was the amount of manual effort required and low reward by filling out the surveys. A Raspberry Pi and Arduino, however, provide low-cost and easy physical interfaces (such as a button) to indicate asthma attack events or breathing discomfort. Such additional health information could then be used by the NCS to provide better-tailored predictions to health threats for a particular household based on disease severity, as the Air Quality Index considers populations suffering from chronic diseases to be more susceptible to air quality fluctuations. Finally, station data is available for users through both EPA's [26] and Airnows APIs [27].

Please note the images used in this project are based on the chosen project for the early prototype available at:

https://www.rigacci.org/wiki/doku.php/doc/appunti/hardware/raspberrypi_air#components_and_connections

Why this project

Due to the limit of characters in DevMesh, the rest of the proposal is available here (including full list of references):

https://docs.google.com/document/d/1BeE4Fy0St91AtTNKVt-4nGWE5RI4750XetD2LPL6Us0/edit

Medium linkedin avatar

Carlos P. created project Local Air Quality Forecast

Medium c3000f54 a01f 4a3b ae25 1ba678aa9d18

Local Air Quality Forecast

Context

According to the Centers for Disease Control and Prevention, asthma is a common disease on the rise with significant health care costs. Nearly 1 in 12 Americans (26 million) have asthma. In the last decade, the proportion of people with asthma grew by nearly 15% [1]. The concern is even higher in Hawaii: in 2008, an estimated 95,665 adults had asthma. Also, according to the same source, adult lifetime asthma prevalence was 16.1% and adult current asthma prevalence was 9.6% compared with U.S. rates of 13.3% and 8.5% respectively [2], but not constrained to it. From wildfires in California [4] to an alarming reduction of life expectancy by 3 years in China [5], air quality is a concern closer to us than we would expect.

According to the United States Environmental Protection Agency (US EPA) slightly above average air quality due to excessive ozone, carbon monoxide, sulfur dioxide, and particulate matter (PM 2.5, PM 10) suffice to put at risk children, older adults and people active outdoors [6].

Motivation

Despite alarming consequences and percentage of the affected population, air quality is not something often seen indoors or outdoors as being measured like a thermometer. This may partially be due to the fact that when first introduced (2003), "NowCasts" was calculated based on a 24 hour time window, and little was known about how to report: PM 2.5 hourly to the public, and the implication of short-term exposure to high concentrations of PM 2.5 on health [7]. More recently, however, new methods have been proposed to not only measure the current air quality in a given location through instant cast [8], but also forecasting it [9].

Despite the exciting progress that has been made in providing air quality measurements to account for when air quality changes rapidly, such as nowcasts still being dependent on the availability of local station data, which greatly reduces accuracy and user satisfaction (and safety!). For example, EPA's own APP has been criticized [11] in user reviews for inaccuracy due to stating "good air quality" despite a large amount of smoke being seeing outside the reviewer's house. What if the instant and forecasts not only relied on station data (sometimes miles away), but also on a low-power device and air quality sensors?

Proposal

This project proposes a low-energy air station built on top of a Rhasberry Pi for a local sensor measurement, whether indoors or outdoors, and powered by Movidius NCS to provide nowcasts sensible to the local surroundings.

Low-energy and low-cost projects using Raspberry PI to create air quality stations have been proposed and implemented all over the web [12-23], including for EPA workshops for 5th-12th graders [24] and even combining both Arduino and Raspberry Pi's [25]. This project builds on top of existing ideas of measuring air quality indoor and outdoors by using the Movidius Neural Computer Stick to extend its ability to forecast.

I also see opportunity in other related projects and studies limitations, such as those involving people who suffer from asthma. For instance, Apple's asthma study using a mobile device proved accurate when compared to existing research [26], however, a major limitation of the study was the number of participants when compared to the initial pool (from 50k downloads to 2,317 users who filled multiple surveys). I believe the main limitation of interest by participants was the amount of manual effort required and low reward by filling out the surveys. A Raspberry Pi and Arduino, however, provide low-cost and easy physical interfaces (such as a button) to indicate asthma attack events or breathing discomfort. Such additional health information could then be used by the NCS to provide better-tailored predictions to health threats for a particular household based on disease severity, as the Air Quality Index considers populations suffering from chronic diseases to be more susceptible to air quality fluctuations. Finally, station data is available for users through both EPA's [26] and Airnows APIs [27].

Please note the images used in this project are based on the chosen project for the early prototype available at:

https://www.rigacci.org/wiki/doku.php/doc/appunti/hardware/raspberrypi_air#components_and_connections

Why this project

Due to the limit of characters in DevMesh, the rest of the proposal is available here (including full list of references):

https://docs.google.com/document/d/1BeE4Fy0St91AtTNKVt-4nGWE5RI4750XetD2LPL6Us0/edit

No users to show at the moment.

No users to show at the moment.