 learning at the intersection of AI, physiology, EEG, our environment, and well-being. So without further ado, please take it away. Hi, good morning everyone. Thanks very much for having us here. So my name is Ahmed. I'm from a research team at the National Institute of Education. That's focused on helping learners design their own inquiry experiments into learning more about their local environment from the perspectives of maker culture and citizen science. So Min-Tuan is a student who has been since 2021 working on a very interesting project where he's actually investigating the impact of our local microclimate on human health using a variety of sensor devices which he and his partner have designed by themselves. Good morning everyone. I hope everyone's having a good day. So my name is Tuan and I'll be presenting to you guys about the Life2Well project. So this is a study about the environment, physiology, and EEG. So do anyone of you know what is microclimate? In the dark right, the microclimate is the climate of a very small or restricted area and it affects humans in their physiological well-being. For example, concentration and thermal comfort. And the diagram to the left really shows how the diagram of the gears, really shows the relationship or the correlations between the environment as well as our physical and mental health. And in the past as well as recently there has been a lot of research on the correlations between the environment and our health. And there has also been a rising popularity in this research. So you may be asking what's the value at of our research then. So our research really focus on another approach which is the citizen science approach. So we focus primarily on three areas. Physiology, environment, and actual intensive biography which is EEG. So for physiology, we use smart watches such as Fitbit and Huawei to collect biometric data such as heart rate, oxygen saturation, blood oxygen saturation, and at best body temperature. For our environmental device, we assembled it ourselves using a different sensor and it can measure noise, light, temperature, humidity, pressure, CO2, and dust concentration. And finally for EEG, this one we took it a step further and as you can see to the little image to your right, we actually assembled it with sponges and tapes. And we only buy the dry out sensors as well as the satin board to measure our EEG data. So this to the bottom left part, the top left part is an image of our own environmental sensors and that is a schema of the different sensors that we use. And the image to the top right is actually the EEG has said that we assemble ourselves. Now after we have collected data, how do we analyze the relationship then? We use machine learning models for the relationship between biometric data and the environmental data. We use a random forest regression model and for the relationship between EEG environmental and EEG environmental data, we use deep learning models. So these are the different features that we extract from the EEG which we are not going to detail because time is very limited. So you may be wondering, if we use our own EEG has said that it's assembled from sponge tapes and elastic bands, it is quite unreliable. So how do we make sure that the data covered is actually useful and meaningful so that we can further investigate the different relationships between EEG and environmental sensors, EEG and biometric. So we came up with our own analysis on validation of EEG data and the plans is as such. And we actually carried it out and these are the results of the NOVA test which is quite congruent. So the advice we use to compare it with is the ant neural EEG headset. We use three different methods to validate which is visual statistics and classifications. Here are the results of validation and NOVA. These are the topology heat maps of the ant neural EEG headset, random forest classification. So these are all the results of validation and they are very congruent. So I will now talk about the results. So this is until the beginning of June last year where we collected more than 100,000 points which included 20,000 EEG windows and they already are from these microclimate devices and smartwatch are shown at the table below. So now for the relationship between the environment and our biometric data. So for the interest of time, I'll just explain one example, one of the findings that we have found. So this is a Shapley assembly plot which we used to interpret the results of our regression model. So for carbon dioxide, for hot rate, they are affected the most by carbon dioxide concentration, ambient sound and ambient temperature and the extent of which it affects hot rate can be seen as such. So overall we found that the most significant microclimate factors are dust concentration, carbon dioxide concentration, ambient noise and temperature. These are for the hot rate and EEG. This is what EEG data suggests about hot rate. As you can see, these are different features that affect hot rate. Also for CNN models, these are the CNC maps for the prediction of CNN about our EEG hot rate. Okay, so what is so meaningful about this? Well we found that this can be used to interpret into an ecosystem of smart homes, personal health devices, as well as to improve students' lives satisfaction and productivity. And it can be used in case studies to also improve productivity and attentiveness in many cases. And we hope that this can speed up the process of democratizing environmental knowledge to the mass where we can be better aware of how the environment affects us. And in turn, it can really help the government to implement more green policies and with more obedience from the mass. So this is all our references. Thank you so much for listening. If you have any questions, feel free to ask. Thank you. Thank you very much, Muntan. Does anyone have any questions? Yep. I was just wondering what's been the most challenging part of working on the project? So I think the more challenging aspect is that this is actually a multidisciplinary study. And since we are only, so me and my partner, we, at that point in time, we are only secondary students. So we didn't have much expertise and resources in terms of how to approach this as well as how we can make we can turn our vision into a real product as well as a real study. So actually what we did is that we reached out to many professionals and we asked our professors to connect us with them so that we can learn more about them. And I think the more challenging part is actually persevere through the mistakes because these are actually very complicated schemas as well as designs that we need to constantly review and renovate after every iteration of a collection of data. So I think that's the challenging part because we didn't have enough expertise to really function this project on our own. Thank you. Next question. Yeah. So what was your main motivation in starting this project? Thank you for the question. Okay, so me and my partner, we were both students from Vietnam. And so when we came to Singapore, we realized that somehow all the allergies that we had as a result of climate has been like magically cured. So we were wondering what happened to what happened? What's the reason behind that? Because it's quite interesting as well in our opinion. And then at the same time, we feel like we can do something meaningful and contribute back to the community. Why? Because we think the citizen science approach is not really a approach that is very popular. And actually it can be very beneficial because we can actually motivate others to follow our footsteps as well. So I think that's like a double benefits that motivated us to actually carry out this project. Thank you. We have time for one last question. Yeah, so in that point about aerating and areas to go, you made flash it by made misted. These sensors mostly in HDBs, condos, the CBD, like what's the makeup of where you're collecting samples. And then on top of that, then if you're, if you did, if resources were to constrain time more to construct what other areas, which we want to expand out to collect data from. Okay, so for the collection of data, we actually this was only conducted. On students. So basically, I asked my friends and my peers to wear these devices when we went to the same school. So basically, we try to mitigate the pattern of data collection such that the area where each of our participants went through during their day or their activities are quite the same. So we cannot limit it so that we can eliminate those outliers. And in terms of expansion of data collection as well as our future planning, we are planning to integrate another type of data that is less intrusive and easy data, which is electro demo activity. So we're trying to find, we have actually assembled our sensor, but then we, again, we need to validate this because it's again made from very low cost and as well as they are in a way they are very different from research equipment. So we need further additional net to prove that it's actually working. And so then we can integrate this into our overall models and investigate more about how these factors are related to each other.