Mad at work? Supporting mental wellbeing at work with new technology

so mad at work mental health and productivity boosting in the workplace and the agenda for today is as follows first I will go through briefly uh the method work project what it is and then we will have the keynote by Michael Salina research manager from Finnish Institute of Occupational Health and after that we will have four presentations from the other uh for some other method work Partners nixu Corporation healthware Corporation Polytechnic University of Porto and vtt at the end we will have a q a session and for that I will ask the audience to write your questions uh to the chat so that we can pick them up from there at the Q a the presentation time slots are a bit tight so it may not be feasible to have questions in there so with that out of the way let's go to the matted work so my network is an international project it's established under the idea framework and funded nationally there are project Partners mainly from Europe but also from Republic of Korea and uh the Inception of the project was the observation that the mental health problems are rapidly becoming the main case of lost work days in knowledge work so as you can see nearly 50 percent of all lost working days have links to the work stress and the cost is quite significant not just to the individual of course but also to the organizations and to the society at Large so the project set out to have the uh or aiming at facilitating prevention of these kind of cases and helping maintaining workplace well-being so as I said five countries have provided Partners 17 Partners total this is a 9.4 million euro budget that has been running for three and a half years now and will end around November or in the end of November this year the consortia are cross-disciplined we have partners that are looking from the health angle we have Insurance Partners we have partners that are experts in regulations and privacy issues and then of course uh quite a few technology partners and the project has carried out several seven real world longitudinal Pilots ranging from 1.5 months to up to six months uh focusing on either distress as experienced by knowledge workers or the environmental factors that may be reasons for stress or at least the working Comfort is an issue there and uh the project had few goals there we wanted to develop methods to assess the mental conditions using sensor data so that it would be unobstrusive and uh also something that can be utilized on long term in real life use the other goal is there to develop novel tools for HR management organizations and then also to the individuals to support uh managing their own well-being at work and then looking very importantly looking at the Privacy safeguards uh because if you measure something at the workplace and especially if you measure workers at the workplace that that requires a quite strict privacy uh uh sort of taking care of the privacy issues there so that's the project overall and these are the sort of targets or create tools that we wanted to create so individual support tools you can see the organizations that have focusing there the organizational tools and then looking after the environmental comfort so that was or is mad at work uh in a nutshell and uh with that we are then moving on to the keynote and Mikhail if you please take the screen yes wait a minute [Music] so hello everyone I guess you can hear hear me and see my slide yes great so maybe a few words about myself first so as what they use it my work as a research manager to finish Institute occupational health and I have been working for this organization for more than 25 years now so quite a long career within one Institute and my academic background is in psychology and I have been studying mostly working hours health and safety in different occupational settings and today I will talk about the Digital Data methods and solutions to promote well-being in knowledge work and I would like to thank the organizers for this invitation and opportunity to present here and I have divided my talk into sort of three sections first I will say something about knowledge work itself then something about well-being at work and finally I will discuss so-called data-driven management of well-being at knowledge work so let's start with this term knowledge work uh I think this term has been characterized very aptly and briefly as thinking for a living whereas the term knowledge worker has been characterized in a much longer way for example already in 1950s Peter trucker defined that these workers or high level workers who apply theoretical and analytical knowledge acquired through formal training to develop and to develop products and services and later on in in the 1990s he foresaw that the most valuable asset of 21st century institution will be its knowledge workers and their productivity and perhaps what we are missing in this sentence is this well-being and that's the topic we have today uh when we look at the current EU statistics it seems that this forecast by Peter trucker hit quite right or at least not entirely wrong as today some 75 million people work in so-called knowledge intensive services in the EU accounting for around 40 percent of total employment in in the EU and if in next little bit discuss this other term important term well-being at work I would say that it's even perhaps more complex than this term knowledge work actually in 2010 the then director general of the Finnish Institute of Occupational Health I Revenue organized an International Conference around this topic and in that conference well-being was seen as the subjective state of being healthy happy contented comfortable and satisfied with one's life so quite positive definition and well-being at work it was seen as a situation where employees flourished and achieve their full potential for both their own benefit and that of the organization so again very positive definition of this well-being at work this term well-being at work and in that particular conference it was also discussed that to achieve this kind of very positive situation and working conditions we need to take into account multiple individual work and Society related factors such as individuals health the workplace as a whole management processes that kind of things and also environmental factors and I think that despite this complexity of this term well-being at work there are at least two clear conclusions when we think about these digital methods and solutions and promotion of well-being in knowledge work first I think that this kind of definition it offers us multiple Avenues or factors we may Target with our methods and solutions and secondly it seems seems to be quite clear that we may also use these methods and solutions not only to manage risks but also strengthened kind of positive elements of work our resources and one very useful model to be used when we apply our methods and solutions is this job demands resources model developed by Becker demeruti and so Philly more than 20 years ago here you can see a very uh I mean to a great extent simplified version of the model based on the publication of rapidly and her co-authors and as you can see it it has two uh Pathways one is called pathogenic pathway and the other one called is called cytologenic pathway and the format begins as you can see with chop demands such as work interruptions and time pressure and this pathway it ends with stress related symptoms and conditions like insomnia exhaustion and psychosomatic disorders whereas the letter letter this pathway it begins with for example manager and peer support and appreciation and it ends with very positive symptoms if you like like job satisfaction effective commitment job related enthusiasm and work engagement and I think that in everyday life also in knowledge we work we have both of these Pathways present and that's why I think it's important to take them both into account when making decisions what to measure by our digital methods and what to develop by our digital solutions to promote well-being at work of course one clear so-called precondition for using our digital methods and solutions to promote well-being at work is digitization and digitization of workplaces themselves here I have picked up one figure from a very recent EU level report that's very nicely describes the main elements needed to really be able to implement so-called data-driven management model in the workplace and when we think of workplaces of knowledge intensive Services I think I think it's quite clear that there are many elements already in place to use the this so-called data-driven management model and and this digital uh methods and solutions to promote well-being at work so maybe we can ask why haven't we already successfully implemented this model is it so that we are still a little bit missing effective and feasible digital methods and solutions for this particular purpose and to answer this question I actually conducted let's call it a mini literature search to see what level of evidence we have that for example mobile and wearable trackers promote employee health and well-being an unlimited my search the so-called systematic reviews and meta-analysis because they summarize very nicely the results of the best original studies conducted until then and secondly I focused only on those systematic reviews that addressed either physical activity sleep or stress and let's next shortly review the main conclusions from these high quality high quality systematic reviews listed here I know that this slide is extremely busy and my purpose is not to go through all details I have written on this slide so let's try to pick up the most uh relevant messages [Music] to the physical trackers I I think that the current evidence suggests that they are they at least have potential to increase physical activity which is a well-known fact to underlying health and well-being however when we look at the conclusions drawn from sleep and stress trackers it seems that they are still mostly under validation but on the other hand it seems that there are opportunities in the future to use these devices as at least as as part of health and well-being interventions and secondly I did also a guide of mini literature shirts uh to identify those e-health so digital solutions that have been used in occupational settings and especially to evaluate the effectiveness of this digital Solutions and these five systematic reviews I have listed they cover those digital solutions that provide for example guided or unguided cognitive behavioral therapy mindfulness-based therapy Stress Management training and various psycho-educative programs and if we next look at again this busy slide and try to pick up the main conclusions and messages I think that this these systematic reviews based on based on them it seems that these digital Solutions actually are quite promising at least in terms of the efficiency in improving mental ability foreign of interest is that there seems to be very little Improvement in the effectiveness over the past decade despite all technological advancements within the same period of time and one reason for this mining binding may be that the use of these Solutions is not optimally timed due to a lack of valid up-to-date data on employees well-being and I think that in this our project not at work we have made several attempts to develop so-called New Generation methods to provide work organizations with the possibility to collect collect up-to-date data on employees well-being and thus help these organizations time their health and well-being interventions optimally one examples one example of these methods is based on data collect date from mouse keyboard and application usage that's a part of normal working day and one clear advantage of this kind of data is that it could be collected without causing any or very little burden on employees themselves another example is so-called organizational organization parameter and with this method it will be possible to collect relevant data for this for instance self-reported productivity and stress and also about factors underlying these experiences on a weekly basis and also at the organizational level and I think that we will hear more about this matter method later on in this webinar and in this married work project we have been also validating these new generation methods against data collected by so-called gold standard methods and we have for example collected saliva samples to deter physiological stress levels and we have his diary to collect data about self-reported work and sleep hours as well as objective data on community performance and our preliminary results they suggest that there are significant associations between for example self-reported stress and productivity collected by the organization parameter and cortisol levels sleep duration and cognitive flexibility observed in our intensive measurements by our core standard methods and I think that in all these preliminary findings they suggests that it will be possible to collect valid data and employee developing quite frequently with the organizational level without causing no or very little burden on employees themselves in the near future another reason sorry yes so another reason for not seeing so much of an improvement in the effectiveness of this digital Solutions may be that these Solutions have not been implemented in an optimal manner and to tackle this challenge I have listed here six main features of health and well-being in the Renaissance that are known to to determine to a large extent how successful these interventions actually are and if you look at these features it's easy to see that features such as having or or relevant bodies on board having a possibility to personalize interventions and also scale them according to the context at hand are among are amongst perhaps the most important features So based on current research it can be assumed that those digital methods and solutions that succeed to incorporate these particular features will prove to be the most successful in the future and to conclude I think that it's fair to say based on our research evidence until now that there is a lot of evidence to support this idea that these digital methods they they carry and they have potential to promote employee health and well-being in knowledge work but of course we are not yet there so to say but we need to take some more steps to really be helpful for organizations and and help them to to improve their employees well-being and health and I think that these steps I have listed here are amongst perhaps the most important ones so promoting user acceptance ensuring the validity integrity and protection of especially personal data increasing the effectiveness of digital solutions to promote employee health and well-being and also making these Solutions feasible for those workplaces that do not have so much resources and know how how to use these methods in practice so thank you very much for your attense thank you and uh on the second bullet which was the uh privacy and uh data security uh we are then segwaying it to the next presentation which is by miksu Corporation so on if you can take the screen thank you Cooper thank you very much now let's see I don't have my camera on now you should be able to see me and also see my presentation right yes great um so my name is I'm a data protection lawyer at Nixon next was a cyber security services company assisting our clients in Daily cyber security matters and we help our clients ensure business resilience um and as part of our service offering we also provide data protection and legal advice relating to different security and privacy questions and that's where I come in so um I advise our clients in a range of legal questions regarding data use adoption or development of new technologies data sharing matters or data sharing Arrangements and I also assist our clients in developing different compliance processes and governance models that fit their needs for digitalization and data use and today I'll give you I'll give you a short presentation on ensuring privacy in Innovative use of well-being data I have also divided my presentation into three parts so first we will look at what personal data is after that we will discuss embedding privacy into the design of Technology and the third and final part is about proportionality and how to ensure proportionality when we use well-being data it's quite a short presentation but packed with a lot of information so I hope you will find this useful then let's dive right into it so what is personal data um personal data is really any information that relates to an identified or identifiable individual it's a very very broad term legally so this identified or identified identifiable individual can really be anyone any living being it can be a consumer it can be an employee a student or any other living person um that we collect personal data about usually when we talk about personal data people tend to think of personal data being a name social security number passport maybe personal address but it's so much more than that it's really any data set that can be combined with some other data and then reveal something about an individual IP addresses can in certain contacts be considered personal data keystrokes on a laptop habits that you can derive from different wearables for example heartbeats stress levels measured from wearable devices or anything else that directly or indirectly relates to an individual is personal data now our personal data is not regulated in the same way some personal data is more sensitive than than others um and so we have certain personal data sets that can be categorized as sensitive or even something that we call special category personal data special category personal data if we start with this one these are data that in the past and unfortunately still today have been used for discriminating against individuals special category personal data this is Health Data data relating to someone's religion political affiliation sexual orientation and others like this it's a set list of personal data categories listed in in the general data protection regulation also known as gdpr um now when it comes to well-being data there is no one definition for well-being data especially not a legal definition that we can find in in data Protection Law um but Health Data is really any data that concerns Health uh any data pertaining to the health status of an individual which reveals information about physical or mental health um so there we can see that well-being data is probably pretty often categorized as special category data because it reveals something about someone's mental or physical health but then as we saw in the previous presentation we'll be well-being data can be so much more it doesn't necessarily directly relate to somebody's health but in any case we should always treat it as something that is more sensitive than a name for example now the more sensitive data is the more we need to do to ensure that what we do with it is legal so collection and use of more sensitive data will always require careful planning and really knowing your legal framework uh that regulates the use of that data so the more sensitive data the more we need to do to make sure that what we're doing is legally okay and correct what do we need to consider then when we want to use technology for for um processing well-being data personal data must always be collected and processed in accordance with a set of key principles key gdpr principles and I will show them on the slide I'm not going to go through all of them in detail but um let's have a look at what let's discuss some of the most important ones for for well-being data here you can see six key gdpr principles that we always need to abide by when we process any personal data including well-being data lawfulness is all about making sure that we know our legal framework and that we have a legal ground for processing data including collecting data we need to make sure that we obtain consent if we need consent we need to make sure that we abide by national legislation um regulating use of data for example this is particularly relevant in the employment setting because we have lots of National Employment laws that we need to abide by lawfulness and fairness is also very much about ensuring that the individual whose data we collect knows what's going on and that it's ethically correct to do what we're doing with the data and this is particularly relevant for for well-being data purpose limitation that's all about defining what you're going to do with the data before you collect it and you need to communicate this purpose to the individual whose data you're collecting and using purpose limitation is also all about making sure that we don't use data in a way that would be surprising for the individual whose data we collect but we need to make sure that the purpose is always intact and um something that the person can can expect accuracy that was mentioned in their previous presentation quality of data is absolutely crucial we need to make sure that the quality is top-notch especially if we're deriving new information about an individual based on the well-being data that we're collecting from them this is particularly relevant for wearable devices where we know that there is a very different framework than for medical devices for example medical devices have different standards for making sure that data really is accurate and and up to date we don't have similar standards for for wearable well-being devices for example and integrity and confidentiality that is absolutely crucial this is all about information security data security making sure that we protect the well-being data that we collect from individuals we need to make sure that only the people who really need to access this type of data and use this type of data get access to it that there aren't data breaches and that we don't reveal the data that individual individuals have entrusted us with that we don't reveal it to people who don't have a right to know it and this is absolutely crucial for well-being data because of its sensitive nature we need to pay particular attention to today's security then in addition to these six core principles we need to respect something that is called the principle of proportionality now proportionality is a very legal term um but it's all about finding solutions that are necessary appropriate and that pursue legitimate objectives when we design new technologies when we want to collect and use well-being data in new ways we need to make sure that everything that we're doing is necessary appropriate and that we pursue legitimate objectives it's really about balancing the intensity of the potential harms that we provide or to to individuals when we collect data about them harms or interferences is this the right legal term so it's all about balancing the intensity of the interference with individuals with the importance or legitimacy of the objectives that we're trying to pursue when we want to collect and use data so it's a balancing exercise and the more sensitive some data is the more we need to do to make sure that there is an appropriate balance between our objective and the potential interference with individuals sometimes an objective can be very legitimate it can be very desirable and it can be very much right and fair but the means to obtain that objective they may be so invasive that they just don't justify the snowball aim that we're trying to to achieve so we need to make sure that there always is a proper balance between the aim that we're trying to achieve and the means that we use to achieve this aim so when you think back to this presentation and if you remember the word proportionality think about balancing exercise now what are some of the key themes or tools that we can consider when trying to make sure that what we're doing is proportionate and okay when we want to use well-being data in in new technologies I've listed uh some things here on this slide um the first one is all about confidentiality I mentioned it already make sure that you keep data in a confidential and secure because well-being data will always be more sensitive to to a person so make sure that you pay particular attention to to his confidentiality and really respect local employment laws when processing employee data some employment laws might be very strict they might tell you what you can or even what you cannot do with the employee data so make sure that you know what the local employment laws say about processing employee data when exploring This legal National legal framework pay particular attention to what's the division of responsibilities between Occupational Health Service and employer responsibilities for example because that would be relevant for for well-being and Health Data plays individuals in control of their data this is very much about giving power to individuals this is something that if we talk about sensitive data that says something about an individual so always put the individual in control of their data ensure that individuals know what you're doing with data what you want what you collected and what you want to do with it and make sure that they actually understand what it is that you're telling them so make sure data use is transparent and understandable here's something to explore is um for example centralized employer-controlled Solutions um or versus decentralized individual control Solutions so when do we use technology solutions that are controlled by an employer and when do we use Solutions technologies that are decentralized and controlled by the individual whose data we collect and use consider consent based approaches to data use obtaining consent can be tricky in an employment context but due to the sensitivity of well-being data it might pretty often be be something that we need to do it will obviously depend on the context um but do consider consent based approaches when you want to use well-being data and then finally because of the sensitivity of well-being data anonymize it as much as possible and make sure that you understand what anonymization means it's both a legal and Technical concept legally it has quite a high bar so really make sure that you know what anonymization means and that it really isn't possible to to re-identify somebody from a set that you have anonymized and then if we look at the blue box always conduct something that we call data protection impact assessments when you want to use data in a new way or develop a new tool new technologies this is an um an assessment regulate or coming directly from the gdpr it's all about describing data use and about identifying risks data protection risks associated with data use and because of the sensitive nature of well-being data do include ethical considerations in your assessment as well right a um heavy set of three slides my key message is really placed individual in the middle of what it is that you're doing and always know your your legal framework and do remember to conduct these data protection and ethical impact assessments thank you very much for your time thank you that was shedding light to uh often somewhat murky depths of of regulations an excellent presentation there and uh next it will be a different kind of lighting discussion by Hillary Corporation and Omar Negan Henry please take the floor hello yes uh just a moment all right um hi my name is Omar NASA and this is my colleague Negan Karimi we are data scientists at helbard so a little bit about help our itself we are a lighting control company mainly based in Finland but we also have offices in UK Sweden as well as having presence in more than 70 other countries now um the main focus of helwar is lighting control systems and what we mean by lighting Control Systems is you can imagine a building with hundreds and thousands of Luminaires uh and then we have a network of sensors so we kind of have this this centralized intelligence which we which we use uh to smartly control the lights and now the the obvious benefit of course is sustainability we want to conserve energy um but at the same time we also want to provide the correct uh lightning environment uh to the occupants of the indoor space so when we talk about lighting intelligence you can think of it this way that we have a network of sensors and dominators which are then providing us iot data and we can use this data to to create an understanding of how the buildings are being used so my colleague and I we are working at a team called future Lighting in halwa in our primary responsibility is to collect this data and analyze this data and basically understand like I said how the buildings are being used but in the context of mad at work uh we wanted to utilize the knowledge of lighting intelligence and then see can we combine it with other data sources and also try to understand uh how we can improve well-being in indoor spaces so this presentation basically contains two projects uh which we utilizes this multi-modal approach where we combine the data from the lighting intelligence systems and also the data from other buildings systems uh to understand and see if we can improve the well-being at workplace so one example of that is indoor air quality um why is air quality important it's because um you can find any study online which says that yeah we spend over 90 percent of our time uh indoors so because we spent so much time indoors ventilation is quite crucial because if the spaces are not properly ventilated the poor indoor air quality can lead to severe negative health effects and probably the most important indicator of this is the carbon dioxide levels so if the CO2 levels are very high they can have adverse impacts on the human health uh for example increasing blood pressure raising heart rates and then it also has uh an inverse relationship with productivity and tiredness levels so one common indoor air quality investigation we wanted to do was to utilize this CO2 data you can see uh rnet for CO2 sensors it's uh it's uh we have partnered with them and uh we are using their sensors in our sensing platform but the main point to take away here is that they provide information on how you can read these CO2 levels so as we all know 420pvm that corresponds to Fresh outdoor air but we would like to keep the indoor CO2 levels uh below 1000. um then they can of course exceed 1000 BPM and above 1400 BPM there are studies which show that brain cognitive function can decrease by 50 percent and that can have a negative impact on um occupants well-being as well as their productivity so like I mentioned before lighting intelligence um is quite related to sensors and one of the most important sensors we have uh is a motion sensor so in a building we we utilize these motion sensors to see or to measure what is the occupancy in a certain space and use that information to control the Luminaires but when it comes to air quality we can we we want to see what is the relationship between um between the changes in indoor air quality and the data we are getting from these occupancy sensors this uh is kind of advantageous because we have already established a network of occupancy sensors and we can see from the data that what would be the impact on on air quality if there is prolonged occupancy so this is Quantified by measuring the density of CO2 and also the historical occupancy data because you can see that what times for example a certain meeting room is occupied uh when do you get peace and occupancy and then how does it correlate with the with air quality so if we take a moment and look at the graph here I have intentionally showed you two different uh indoor spaces and the bottom indoor space is from an area which is quite well ventilated so even though you can see there is consistent occupancy there are people in that space the CO2 levels never really exceed or they never really increase but in the top part we have the same data from a very small meeting room and there you can see that when there is occupancy the CO2 levels show a significant change now um we we also know that occupancy is a leading indicator so you will first have occupancy in a room and then the air quality will degrade uh in the space if the space is not properly ventilated and this is quite important because uh we want to differentiate between a poorly ventilated and a properly ventilated space so we ran a virana pilot in our office and We examined various meeting rooms with the capacity of one person and we wanted to analyze that what is the impact of occupancy on the CO2 data so by by analyzing the the duration of occupancy and how much the peak is with the CO2 we can come up with some values uh that how long that room should be used before it starts to have a negative impact on the occupants well-being um so the goal was to summarize these uh these this information concisely and display in each room and you can see some of the examples here so one example is that yeah the air quality remains uh at a good level for 13 minutes but then after 20 minutes it will be it will degrade uh and then we were also trying to give some information to the users that if that happens or if you have been in this room for 30 minutes please open the open the door open the windows to allow it to properly ventilate now one thing to remember is that this is not going to happen in every building on every room because it is majorly dependent on how the ventilation system is configured so what was the user feedback um we let this pilot run for approximately uh eight weeks during which interviews were conducted at different uh different periods of time and uh the the one thing that I want to talk about here is personal action so like I mentioned that we were also giving this information to the user that for X minutes it's okay to be there and then it the air quality can degrade but the overall rate of personal action was quite low meaning that the number of people who were sufficiently concerned about poor air quality they didn't really take any action uh and we found that users might not fully read the information or might not act on it which is kind of understandable is that if you have a primary purpose in that room which is to you know be in the meeting then a secondary uh in piece of information uh that that might not get the user's uh Focus or attention so the key takeaways is that yes people understand when the air quality is degraded that it has an impact on their health but they are generally not aware so maybe education and awareness is quite important uh the intervention should be done automatically that the ventilation system should proactively react to this uh to the worsening air quality uh of course we try to present the information as minimally as possible but then again text is always uh it's a bit hard to read all the text that you provide to the users or maybe a more impactful system could be just a blinking system or a simple lighting system traffic light system for example and also provide a clear plan of action like open the door or the window with that having said I will now hand over to my colleague who can then go through the second project uh that we did yes I will present our results for the analyzed off the HVAC performance for improving indoor air quality and Energy Efficiency systems are one of the largest consumer of energy in a building and energy consumption through the heating ventilation and air conditioning represent a significant person of energy usage and here we have focused on the ventilation components of the HVAC system one of the challenge here could be over ventilation which is expensive and consumes more air conditioner energy than necessary and on the other hand in proper ventilation leads to a decreased indoor air quality the main idea here is to design a recommendation system that suggests optimal times for adjusting the ventilation rate aiming to improve the air quality while optimizing for energy saving simultaneously and we collaborated with the facility manager at the Nova hospital and we have employed CO2 data Supply and Xbox air volume Supply airflow is the uh air flow from the outside to inside and the exhaust Air volume is the airflow in the reverse direction from inside to the outside and as well as we have used the occupancy data from the hellvar's occupancy sensor and we analyze the relationship between the occupancy data and the CO2 and CO2 data and present the result of this relationship by the visualization method for example in this plot you can see the average of the occupancy data by the green line and the CO2 data by the gray lines during two weeks in the first step we check this data just in two weeks and we can see the CO2 levels I have a tendency to go higher than 1000 BPM frequency and it means that this space needs to improve the air quality and we suggest that probably they can consider improving the air quality there and then they change the airflow they increase the airflow and again we checked the CO2 data and occupants the data after the modification in this table we have the result of the CO2 data as a percentage of the value before modification and after that it is understandable that just two percentage of data is more than 1000 before the modification it's not a significant value but anyway hospitals requirements and even value more than 800 people PPM would be a high value foreign occupancy data and uh there in these parts of the hospital over ventilation happens since most of the CO2 data are are between the 400 and 600 so we suggest saving more energy in such a area and then we checked the effect of the dropping up in such area and we can see that how CO2 data increase after the modification but still air quality is in the safety level after the modification and here I can mention some challenge for example almost the HVAC system or elegancy and do not have the cloud apis or easily accessible inference and then we have experimented with the real-time predictive system which can produce recommendation on how to modify the insulation rate based on the current succupancy conditions however running a practical experiment remains a challenge due to the Lake of the availability of such system and indoor air quality is a significantly overlooked area of well-wing and productivity decrease given nature of the Legacy HVAC systems many Market Solutions provide the ability to friends and make inform decisions based on their current conditions but some watches are still needing in developing fully autonomous self-learning system that require no input from user and finally even a straightforward data analyze can provide significantly actionable insights as the demonstrated with this pilot uh yeah that or or analyze related to my network projects thank you so using sensors and then utilizing the existing sensors in the buildings to guide other systems yes thank you okay thank you thank you and uh then the next one is is from Polytechnic University of Porto Sima please take the floor so greetings everyone and first of all I would like to express our gratitude for the opportunity to share our work among the presents and this goes to vtt in the managing team behind this webinar today and of course we would like to extend our gratitude and greetings to everyone present and listening to these presentations for the participants that don't already know me my name is simofred I'm a PhD candidate in health data science and I'm working in Porto School of wealth specifically in the center for translational health and medical biotechnology research alongside Professor Matilda Rodriguez and Professor noon very every day strive to make data-driven decisions support and recommendation to pursue the best outcomes in almost every branch of health related research and here specifically in stress detection and mitigation this will be a short presentation about our Portuguese solution and the project made it work and to give you a quick perspective here's our participating Partners in the National Consortium we have two research and development centers and universities for the school of wealth and ezep with Professor Fatima specifically made this with Enrique Nunu and Clint with Rita and biometric with Lewis Here we see a span of what is matter networks various models and solutions to improve employees well-being engagement and performance and we position ourselves in the category of the individual support tools alongside and now we will focus a little bit about our specific solution we developed an integrated stress detection and recommendation system that is based on a totally unobtrusive video platysmography system and to be clear we use the webcam as a remote PPG sensor our solution is capable of predicting stress on set and also address trends of long-term stress we adapted and develop an elf toolkit with eight Dimensions to establish other recommendations would be presented to every participant and since our recommendation system is ranking based from one to five but it's also content based and it builds upon this ground of the health toolkit we will further have the opportunity to see some dimensions of our recommendations but this is an evolving system with the capability of implementing new recommendations and rules for presenting the recommendations when we want with very little entropy in the future but how did we got here so if you're here present today you probably know that stress detection is a sensible yet a very hard variable to address we started by building a theoretical framework for all the blocks that we we needed gathered evidence from the literature and needs from end users regarding stress monitoring with some focus groups and perceptions from a transcultural survey that have more than 600 participants despite being quite long and informative for our video platismography system we did a laboratory of validation with ECG different lighting scenarios and workload as well to validate this totally unobtrusive physiological data collection solution we develop another systematic review to collect and identify the most effective and commonly used interventions for stress mitigation being on set and also long stress and we build our model with a recall of approximately 95 percent where we identify the appropriate data collection variables lastly we compiled our evidence-based recommendations in a library in developed biofeedback based interventions for the immediate stress recommendations that are quite lacking in the literature to finish our roadmap we designed the data collection in laboratorial and in real life settings environments and did a very effective data management between every member of the Portuguese Consortium in order to maintain anonymity for all the participants engaging the solutions satisfaction and development questionnaires did several reassessments for the data collection software and now we are happy that we adapted our protocol for office and also for home desk setups being able to collect data everywhere and every time possible this is our digital codes diagram and the rationale for the stress detection where we can see the physiological data driving the motion towards the recommendations and this is something that we wanted to address from day one so this three possible data paths for the stress detection we can have onset detection of stress of single stress Peaks where there is a trigger in the alarm system we can have a detection of long-term stress where we must address real changes in lifestyle like presented here sleep exercise weight management optimism and Good Deeds social relationships amongst others and finally but not least prevention this digital codes can not only function as being a supportive tool for stress but also a prevention kit this is a look at the architecture of the data that we gathered and how does it transmit to everyone so the data is collected and passes through the classification model and we have two possibilities now the personal data and recommendation and in the near future aggregated data and visualization of teams and their well-being going deeper into our system it is a video platismography system remote video platismography PPG whatever you want to call it but it's based on the hand PPG method with fine tuning we also have analyxnet emotion recognition system with seven outputs being neutral angry discussed fear episode and surprise we have ocular data as well being blinking analysis duration of blink blinking frequency and an alertness variable called betus this is a look on how the recommendation system is designed and we have some example of work related recommendations here such as work life balance finding meaning at work amongst others in this slide we can see part of our evidence-based recommendations Library starting from the top with the planned models for the software and the main themes for recommendations going from task management healthy sleep habits structured pauses mindfulness promotion of active and Healthy Lifestyles the possibility to have this customized intervention and others that we are still developing like biofeedback training for unstressed or onset stress mitigation this is a really quick look to our best results and the needed attributes to have this prediction model we found that the random Forest is the algorithm that fits the best in our solution having the extracted extracted HRV features of avnn RM SSD And the emotions angry and fear the variable purpose the base weight of the person and type of sleeping schedules bringing our results to almost 95 recall and 88 F1 score everything all of this in mind we have achieved quite interesting results through this project starting with the contributions to the field of remote video platismography stress detection and video feed data management bringing new evidence and sharing some light on the possibilities of using these tools for managing personal well-being and performance we also build some ground in developing not only a data focused solution but listening to the end users and their experiences in order to ensure Comfort safety and performance we are quite happy with our evaluation methods being compiled with gold standard scales adapted health toolkits for the initial Health assessment and being able to to build from the ground up through this basis we've developed an evidence-based recommendation library that is focused on stress mitigation and targets important issues in the workplace and specifically for knowledge workers we will be implementing in the future biofeedback principles and interventions that focus on stress mitigation on set since this is what the visitor is lacking a bit in the moment and now we can see some of our Technical and scientific discriminations to the moment and we have quite a few more in the in the near future to end I will just give you a quick demonstration of of our solution starting with with myself so this is just a video of how it works in in our pilot so I'm at home Gathering some physiological data as you can see we have the emotions the heart rate the blinking analysis as well this is fulfilling two questionnaires that we have on a daily basis we also had four random labels of stress throughout the day starting with 0 to 10 this is what we see now and here we see our recommendation system that is uh already running this is a video for it review so we are doing [Music] um an account for ITA review and you can see that it's it's working now so we must just wait a little bit yes and you can see now they have toolkit that I was mentioning before that has a lot of questions but they can be fulfilled from time to time from the participant doesn't have to be a one-time thing just so that the participant is not overwhelmed by the questionnaire and here we can see a demonstration of how the notifications come just for demonstration purposes we are doing a ping of 30 seconds for every new recommendation this will not be the final solution as you can see and then the person can just check the recommendation give it a rank from one to five and in the no more section they can see the evidence-based recommendations and where this recommendation is coming from and lastly we just wanted to show you uh a sample from the same kind of of data that we have from PPG and also from the polo rage 10. our laboratorial validation is with ECG the biopac mp36 but this is just for visualization and for demonstration purposes and that's also a paper stating that the polar h10 is 99.5 0.4 percent accurate and we can just see that the data is really close to the polar h10 just for this visualization purposes this is both real-time data in our solution and also in the polar h10 and it's that's it from our part and thank you for being present and listening to our work development we will be more than happy to have a discussion and answers in the in the question and answer session and thank you once again thank you and uh then last presentation for today is by vtt and Julia Elena please take the floor okay thank you um can you see my screen yes okay so um I'm the president today Elena will help me in particular if they will be uh difficult and interesting questions and I'm using your researcher of Robitussin nature technical Research Center of Finland and uh my background is on data driving services and application and information design information visualization sales and technology and I'm presenting here actually the work of the group of researchers from BTT you can see the names here our background is very different here but very suitable to address the challenges of stress detection we have our background of stress detection Technologies on science and technology on stress data related analytics AI machine learnings so and also software developer here so we have the very very very good and enthusiastic group to address the challenges that have been presented uh to you with reverse speakers uh so my last presentation is dedicated in particular to yet another piece of let's say technology and developments related organization uh well-being and organization barometer in particular uh so if we look uh like how the organizational well-being is uh tackled now in many organization uh of course everyone will uh know that currently um this well-being is assessed by low questionaris uh they are infrequent and they might be affected by by current situation there is also a time lag or between collection data answering the question processing present in data decision making improving and and then then the time for action will come uh then again you are surveying questionnaire will arrive so naturally some continuous measurements and capturing instant feedback on presenting uh well-being data for exploration can certainly help and that's why we decided to take this Challenge and better to work project and so we decided to do at least to implement to research Implement sub-concepts related organization barometer we call it uh in a way like continuously updating data-driven well-being map uh the aim was from the beginning to collect um uh well-being stress data using various means unobtrusively uh collected from sensors at a convenient easy to use um easy to feel self-reports um and uh also to integrate these using Innovative visualization means information flow means uh to the these concept and prototype of organizational barometer and uh yes as also mentioned earlier we in particular this project we target our knowledge intensive professions here so to compile uh this to you in let's say uh two main research questions that we uh vtt researchers were targeting here this project or we were thinking about uh what kind of technological means can support our prolonged stress and uh stress stressors detector detection of the requirements was to use on the protossive semi semi-unsupervised uh means and of course the accuracy should be acceptable and then the second research question we said for us uh what kind of information design would support individual by but also organizations organizational reflective thinking and action taken to improve well-being at work on again individual and organizational level so my presentation or it contains or I would say two parts um first of all I quickly um explain you how how we approach to these two research questions and what kind of methods we have been using what kind of sensors we we we decided to use and why and then the second part of my presentation is I will hopefully try to show you some designs that we we did and also some prototypes that we developed okay so as you uh may know of course the stress detection is not a very new area but still it is a a new in the way an immature research area because most of the stress detection studies and data analysis has happened in other lab studies so there is uh very little or published data sets that would allow to to analyze and detect the stress but um they are many sensing technology is available or you can see here by slide some examples um uh one can measure the physiological data or it's considered the most objective approach uh to to measure the stress using the heart rate our abilities skin conductivity or stress or can be also measured using behavioral data like data collected from mobile phones movements detections computer data usage then of course other sensors can be uh in principle usually like a death camera face expression and speech or the speech recognition and um there is of course uh sensors that can be detected some environmental conditions or in in the fails and also of course they are commercial solutions for personal use available both in the in the market but then one or think about the selection of particular Technologies uh uh it is usually the trade of of obtrusion is accuracy and privacy yeah and the context of usage but then of course our independent of how are the data well-being data is collected um the data is um is not meaningful if uh the user is not able to understate it and most importantly to act on it and um if you look on what is available around uh they are available some research prototypes some of them you can see here I collected for you some examples some some are targeting or individual use to present or stress uh and uh well-being data and less less let's let's say research but they are all research prototypes yeah and less research prototypes are targeting Collective group group data visualizations actually you can see from the examples here that fish based visualization is is very popular um these Solutions employ various let's say visualization techniques or some of the most simple one use timelines calendars chats um less solutions that use some uh let's say advanced technology like visual metaphors analogous um so uh in in fact metaphors is a promising and we see it's a relatively new approach uh to to visualized well-being data especially at our real condition at work environment they are also proposed actually many system for other um domains like to address this Collective awareness uh like for example uh the election debates uh have been Advanced visualized but uh well-being of data for group usage at work environment is is a steel research area uh so how how we started with this um of course or very very quickly to give you just clumps of impression how we did this of course as uh previously early presented by our colleague or from Finnish uh Occupational Health Institute there is a stress and stress research is not the new area or job demands resource research has been actively uh active research area in 80s uh started already in 80s um the the so we use this of course as a background for our designs and developments we also use some psychological well-known theories like for example um uh self-determination Theory which say which says about the psychological needs or that the worker may have at work we also use some other theories of uh like flow theories of the psychology of optimal experience and happiness which also said us and say us something about the importance of what people do job content what what how they use or able to use uh the the skills and resources they have uh we also have been talking with end users or we of course we leverage vtt as a test bait uh more than 2000 researchers so we run various polls to understand mainly uh how how what kind of language uh people use then they talk about their challenges then they talk about um what kind of content or that they do at work and we also always working with Finnish companies to see uh what kind of uh reason for stress or they they mentioned more often uh what kind of um information and uh content they would like or comfortable to share with their colleagues and what kind of information they would be comfortable to share for example with line managers um it's not that uh let's say new information it's River confirmation on the literature research and as I say the most important for us because we really wanted to measure something we wanted to implement these Concepts in the in a reporting application for us it was very important to understand what kind of language uh how how how people are Express these concerns in all language to make this design understandable also for them yeah in addition we also run cross-cultural survey uh from uh using uh for uh in four five European countries um 700 responders to understand uh

2023-07-13 01:46

Other news