Premise
Monitoring the health of south Asian Migrant construction workers in order to detect life threatening heat strokes.
Synopsis
When it comes to adopting modern technologies, the fields of architecture and construction tend to lag behind, and the topic of safety on construction sites is no exception. While safety improved in the recent years, there is no doubt much more can be done to avoid preventable injures – and in some cases fatalities, especially among marginalized migrant workers. This project aims to highlight this issue and offer an affordable solution using IoT technology.
Preface
Later this year the 22nd version of the FIFA world cup will take place in Qatar. It is a spectacle that people and competing teams flock from all over the world to attend, and while it is much anticipated, it came at a very high cost. A Guardian article revealed that as of 2021 more than 6,500 SA (south Asian) migrant workers have died since the hosting rights were awarded to Qatar back in 2011 [1]. This highlights a much bigger problem throughout the world and in the GCC (Gulf Cooperation Council) region specifically. Similar trends can be seen throughout the region as it has been witnessing an ever growing construction boom in the past decades. While not all of these deaths are related to poor working conditions of SA construction workers, official governmental figures attribute around 70% of deaths to natural causes or cardiac arrest with little to no background information over the underlaying causes [2]. This number is staggeringly high given that these workers are expected to pass physical fitness examinations prior to traveling.
A 2019 study that investigated the mortality rates of migrant workers and climate in the GCC region found a strong correlation between CVD deaths and high WGBT, which indicates that a high percentage of these deaths are most likely due to severe heat stress. Other factors that play into this are the long work shifts that strain the body and the fact that many of these workers come from colder and drier regions such as Nepal, which is a mountainous country where the average altitude is above 3000 meters.
“Global studies show that approximately 15% of deaths in the age group 25–35 years are due to CVD causes. However, in this NMW population, the figures were 22% during the cool season and 58% during the hot season.” [3]
The human body maintains a core temperature of 36.5-37C. Body temperature rises during physical activity and warm temperature, and the body cools down through different mechanisms (e.g. sweating, radiation, etc..) known as thermoregulation. These mechanisms can be weakened though when ambient temperature reaches 33C, which can frequently occur in the gulf region during summer months. While such temperature is not common throughout the world, it has been increasingly observed around the globe in recent years due to climate change. Another factor that affects thermoregulation negatively is high humidity, which is the case in the gulf region[4]. Once the body core temperature reaches 40C, it is considered a heat stress. This is when immediate medical attention is required to cool down the body as it loses its ability to thermoregulate. Delayed medical attention leads to brain damage and multiple organ failure, and eventually death[5]. While a core body temperature of 40C doesn’t necessarily constitute a heat stress, it can be easily confirmed through measuring ambient temperature and heart rate.
Process
Since core body temperature is the gold standard to identify heat stresses, I started by looking for suitable methods to measure it. Some of the more conventional methods such as rectal and oral thermometers are deemed too invasive and highly impractical, especially if the aim here is to monitor temperature in real time. Other less invasive methods would be body surface or the tympanic membrane ( in-ear), and are suitable for real-time monitoring as measurements could be acquired using infrared technology without need for contact[6].
1st Iteration
Having decided to go with Arduino and the aforementioned methods for monitoring core temperature, the aims of this iteration were to, firstly get familiar with Arduino’s functionality, environment and components, and secondly to determine a suitable method to measure core temperature. For this iteration I used an Arduino Uno board and a Melexis MLX90614 infrared thermopile for measuring temperatures. I also brought an Idyl oral thermometer to help me determine which method seems to be more accurate, body surface or the tympanic membrane.
Firstly I tested the functionality of the components through connecting them to the laptop and uploading test codes through Arduino IDE software to make sure they are working. Afterwards I started taking readings of my core temperature using MLX module from two areas, the tympanic membrane in my ear, and the forehead and compared them to readings taken by the Idyl oral thermometer.
The Idyl oral thermometer kept giving a stable reading of 36.9C. Forehead readings using MLX thermopile kept fluctuating between 34.8C and 37.1C with some readings reaching as high as 40C. This is consistent with what some studies show that forehead readings could give a normal or lower than normal readings despite a higher core temperature[7]. Tympanic readings fell in a range of 36.6 and 37.3 with a fluctuation of +- 0.4C. This fluctuation is probably due to ambient temperature and lack of insulation of the MLX module as most readings hit the 36.9-37C mark. These readings could be optimized through insulating the module properly and using an algorithm to filter out noisy readings. I concluded through this iteration that measuring core temperature through tympanic membrane is superior and more suited for real time readings, specially that placing the sensor on an exposed area to sun such as the forehead will probably lead to more wrong readings given the environment where workers are in. I have also cultivated a basic understanding of Arduino’s functionality through this iteration.
2nd Iteration
Once I decided on a method for taking readings, the next step was to verify whether a recorded high temperature is due to heat stress or not. Elevated heart rate and reduced ability of sweating are symptoms of a heat stroke that can be measured using sensors. My first thought was to measure heart rate, but after thoroughly experimenting with a couple of pulse sensors that I purchased online, they were both faulty and gave me wrong readings. After reading about the issue, there are apparently quite a few faulty batches circulating in the market. Eventually I ruled out heart rate as a verifying factor due to time limitations.
As mentioned before, there is a strong correlation between hot weather and heat strokes, so my next step was to factor in ambient temperature. The MLX thermopile has the ability to return two temperature readings, one is for the object and the other is for the surrounding. Upon experimenting though, the ambient temperature reading was inconsistent and didn’t reflect the actual room temperature. It also seemed to be affected by the object temperature which deemed it unreliable, so I opted for using a second thermometer that can be placed away from the object to give more accurate readings. I obtained a real time clock module (DS3231 RTC module by Maxim) that comes with a thermometer and integrated it into the circuit. Another advantage of such a module is that it also keeps track of time which is helpful for data logging purposes. In a practical scenario, ambient temperature could be obtained through a WBGT device that measures the wet bulb globe temperature and send it to the prototype wirelessly. These devices are easy to implement and are already commonly used in construction sites. The following diagram shows the wiring of the components where the MLX sensor returns object temperature readings while the RTC module provides ambient temperature as well as time and date.
3rd Iteration
Now that the prototype manages to take accurate readings of both core body and ambient temperatures, I decided to interface it with an OLED display to show the readings in real time. I also connected a bone conduction transducer (speaker) which gives a warning when the core temperature exceeds a safe limit. This transducer is basically a metal rod wrapped with a voice coil that transmits sound through the skin without the need to place it inside the ear. The reason I opted for it instead of a conventional speaker is due to its non-invasive nature. Blocking the hearing canal with an in-ear speaker is potentially unsafe as workers need to hear sounds in their environment clearly without obstruction. Also since the thermopile is located very close to the ear canal, there is already too little space for an earphone.
Upon testing the prototype, it succeeded in giving real time temperature readings and send audio warnings once the temperature rose to a dangerous level. In order to prevent the system from sending false warnings in case a noisy reading has been detected, I modified the code to only send the warning when at least 10 consecutive high temperature readings have been recorded over the span of 5-10 seconds.
Conclusion
For future iterations I picture integrating a pulse sensor to improve feedback and monitor fatigue. It is also important to integrate a highly autonomous algorithm that processes the data locally due to privacy concerns.
Given the simple nature of the prototype, the relative affordability and availability of the components, and the solid scientific backing behind the aforementioned methods, it is truly heartbreaking to see how easily preventable many of these deaths are. May they rest in peace and may technology keep helping us make the world a safer and a better place.
References
[1] Revealed: 6,500 migrant workers have died in Qatar since World Cup awarded – The Guardian
[2] Qatar’s failure to ivestigate, remedy and prevent migrant workers’ deaths – Amnesty International
[3] Heat Stress Impacts on Cardiac Mortality in Nepali Migrant Workers in Qatar
[5] Heat exhaustion – Symptoms and Causes – Mayo Clinic
[6] Core Temperature Measurement—Principles of Correct Measurement, Problems, and Complications