This site uses cookies and other tracking technology to assist with navigation and your ability to provide feedback, analyse your use of our products and services, assist with our promotional efforts, and provide content from third parties. You can read more in our privacy policy.

DHI MSc Blog: Indoor CO2 analytics for healthy living

Part of the blog series from MSc students funded by the Digital Health & Care Institute

Wednesday, May 30, 2018
Chika Eucharia Ugwuanyi

Indoor CO2 analytics for healthy living

Evidence abound that people spend greater percentage of their time indoors. These indoor environments includes places of work, public transports, houses, classrooms etc.. One major poisonous gas that are often found in an enclosed space with limited air supply is carbon dioxide (CO2). According to ASHRAE, typical outdoor CO2 concentration level is between 250 - 350 parts per million(ppm) with ambient air, whereas a typical considerable indoor CO2 concentration level could range from 350ppm - 1000ppm with good air exchange (see Figure 1).

Figure 1: Air Quality by ASHRAE
Source: American Society of Heating and conditioning Engineers [2016]

From Figure 1, it is obvious that once the indoor CO2 is above 1000ppm, occupants ought be concerned about the quality of the air they breath. Sometimes these indoor environments are poorly ventilated and most of the time it’s almost impractical to actually say for sure how healthy the air, that the occupants of such places breath is until you scientifically prove that. The main reasons for this are that CO2 is a colourless and odourless gas; that can’t be seen nor perceived.

One example of such places is on major highways, it has been observed that there are high rate of concentrations of Volatile Organic Substances (VOS) such as CO2 inside long-distant buses (Figure 2). In 2004 alone, road traffic CO2 emission accounts for approximately three-quarter of the whole emission from global energy use Woodcock et al. [2009]. Interestingly, according to vehicle certification agency (VCA), these CO2 emissions are directly proportional to the quantity of fuel consumed by an engine Browne et al. [2014]. This could imply that, the more vehicles we have on the road, the more fuel they consume and consequently, the more risky the inside of our transport systems are likely to become as a result of increased CO2 concentrations, when there are limited fresh air supply. This could adversely affect peoples’ health without them knowing about it.

Figure 2: There are high CO2 concentration in public transports

Another known indoor place where increase in CO2 concentrations could be a cause for concern is in an office or a room with inadequate ventilation (Figure 3). Sadly, most employees or occupants are being exposed to these high level of this poisonous gas daily unknowingly. One adverse effect of high indoor CO2 in a poorly ventilated office is its negative impact on human cognition (ability for human to perform their duties sucessfully) Gall et al. [2016]. Employees in such conditional state of environment are less productive which in turn could negatively affect the revenue or success of the organizations they are working for.

Additionally, another place that might have potential for increased CO2 concentrations is a class room and/or lecture hall (see Figure 4). Consequently, these poorly ventilated classrooms with high number of students could produce stale air thereby causing students to contract airbone diseases such as sneezing, runny nose, sore throat and headaches etc.. Rudnick and Milton [2003].

Figure 3: Poorly ventilated offices


Figure 4: CO2 increases in a populated lecture halls with limited air supply

In order to curb these indoor diseases’ symptoms that often emanate when the indoor CO2 concentration levels are high, various environmental sensors have been developed and made commercially available to help occupants to become more aware of their indoor environmental conditions with respect to CO2 concentrations. These sensors are able to track, monitor and record the level of indoor CO2 concentrations and present the results to occupants. However, meaningfully utilizing these recorded sensed data for occupants and/or stakeholders’ utmost health and well-being is a known academic challenge.

To bridge this gap in a pragmatic way, we propose a situation whereby, these sensed data could be periodically tracked, recorded and examined methodically for more insight. This process is often refer to as data analysis. Applying ensemble regression methods as one of the machine learning methods on the sensed dataset (CO2, temperature, humidity, pressure, noise and dew-point), we were able to predict indoor CO2.

The historical observation (data) used was recorded with a sensor for a period of three months and transferred to a database hosted in the cloud for easy retrieval and analysis. The ensemble models used in this study achieved upto 97% in predicting the indoor CO2 better than the traditional linear regression method. The evidence of the results’ outcome is shown in Figure 5

Figure 5: Graphical representation of the performance

The graphical representation shown in Figure 5 is an empirical evidence of relationship between indoor CO2 (dependent variable) and other indoor environmental (independent) variables such as temperature, humidity, pressure, noise, dew-point, and that indoor CO2 concentration levels can be predicted in advance once other independent variables are known. The actual indoor CO2 values are shown in blue and the predicted values of the CO2 are shown in red.

Conclusively, as our society struggle to reduce outdoor carbon (greenhouse) emissions in our environments more efforts and awareness should be created towards making sure that the CO2 in our indoor environment is always at a level that is recommended by ASHRAE, and future prediction approach is one of such methods of dealing with the situation pragmatically.

Chika Eucharia Ugwuanyi
PhD Research Student, University of Strathclyde

Refrigerating American Society of Heating and Air conditioning Engineers. Standard 62.1-2016 – ventilation for acceptable indoor air quality (ansi approved). 2016.

Michael Browne, Christophe Rizet, and Julian Allen. A comparative assessment of the light goods vehicle fleet and the scope to reduce its co2 emissions in the uk and france.
Procedia-Social and Behavioral Sciences, 125:334–344, 2014.

Elliott T. Gall, Toby Cheung, Irvan Luhung, Stefano Schiavon, and William W. Nazaroff. Real-time monitoring of personal exposures to carbon dioxide. Building and Environment, pages 59 – 67, 2016.
URL //

SN Rudnick and DK Milton. Risk of indoor airborne infection transmission estimated from
carbon dioxide concentration. Indoor air, 13(3):237–245, 2003.

James Woodcock, Phil Edwards, Cathryn Tonne, Ben G Armstrong, Olu Ashiru, David Banister, Sean Beevers, Zaid Chalabi, Zohir Chowdhury, Aaron Cohen, et al. Public health benefits of strategies to reduce greenhouse-gas emissions: urban land transport.
The Lancet, pages 1930–1943, 2009.