Periodic Reporting for period 3 - CO-ADAPT (CO-ADAPT: Adaptive Environments and Conversational Agent Based approaches for Healthy Ageing and Work Ability)
Période du rapport: 2021-06-01 au 2022-05-31
Ageing citizen face particular difficulties in maintaining an active work and life-style originating from at least from two sources 1) reduced capabilities due to age-related and health risks conditions 2) societal transformations that revolve around digitalization. Ageing users are less computer literate than younger generations and have in addition higher usability requirements. Conversely Artificial Intelligence, sensors and interface technologies such as conversational agents provide opportunities to support work ability and wellbeing in particular for an ageing population.
The aim of CO-ADAPT is to provide a framework (OBJ 1) that will serve as a change-enabler identifying key aspects that support a two way adaptation, supporting ageing citizens to adapt to changed conditions and providing necessary smart adaptations from systems in work and personal life. The OBJ 2 focuses on a personalised conversational application with comprehensive change support for ageing citizens. The OBJ 3 is to develop secure and GDPR compliant AI techniques capable of contextually recommending change programs. And the OBJ 4 investigates three types of smart adaptations in work systems to age in smartshift scheduling, collaborative robots in an assembly station, and task based information entity recommendations, aiming at 2 digits percent improvements in task performance and or satisfaction.
The aim of CO-ADAPT is to provide a framework (OBJ 1) that will serve as a change-enabler identifying key aspects that support a two way adaptation, supporting ageing citizens to adapt to changed conditions and providing necessary smart adaptations from systems in work and personal life. The OBJ 2 focuses on a personalised conversational application with comprehensive change support for ageing citizens. The OBJ 3 is to develop secure and GDPR compliant AI techniques capable of contextually recommending change programs. And the OBJ 4 investigates three types of smart adaptations in work systems to age in smartshift scheduling, collaborative robots in an assembly station, and task based information entity recommendations, aiming at 2 digits percent improvements in task performance and or satisfaction.
The CO-ADAPT project summarised all lessons learnt in a consolidated framework to be submitted for publication. The framework is based on the foundational concepts of healthy ageing and workability, it motivates the human centred development of digital transformations in particular considering ethical aspects of AI of transparency, accountability and explainability. The framework also contains four case studies demonstrating AI based digital transformation in support of workability and wellbeing.
The framework is supported by catalogues of Active ageing principles for technology design and recommendations and change strategies for active ageing. The project also collected and formulated ethical guidelines in particular considering adaptive and intelligent technologies and ageing workers. The development of all technologies in the project was characterised by human centredness, participatory design, and iterative development with stakeholders.
The project implemented successfully three adaptive systems at work. This includes guidelines for adopting smart shift scheduling, an Adaptive Assembly Workstation (AAW) equipped with a cobot that is capable of adapting to the operator's internal state during the work cycle, a task based recommender system to support workers at the computer with useful information based on context published open source.
CO-ADAPT also delivered in open source a privacy preserving and secure platform to collect at the study sites participant data. An analytics portal app offers a front-end to the platform for managing the organisational and user entities involved in studies, as well as for offering the analytics visualisation. The project developed a conversational system with Deep Neural Network dialogue models.
The project organised successfully evaluation trials for four different adaptive technologies meeting improvement close to or above 2 digit percentage over baselines without the systems. The Smart Shift Scheduling (particpant n= 5 519) show beneficial effects on the primary outcome, sickness absence, with a reduction of nearly 10% in absences among hospital employees, as well as positive results for cost saving using participatory scheduling and wellbeing. The adaptive assembly station resulted in a 2-digit improvement in performance (i.e. a reduction of the time on task) and a 2-digit decrease in stress and mental load. An overall positive subjective experience was reported by all participants.
We completed the evaluation and development of a novel entity recommender system demonstrating target 2 digit improvement was met considering task performance and operationalised through proportion of relevant information found.
The project also evaluated the impact of a conversational application in a real case of psychological support. Analysis conducted on psychological test results, administered before the intervention and at the end of the same, showed improvements in well-being and motivation for the group of participants that used a conversational application for which a 2-digit percent increase was recorded. The project also organised a high level workshop inviting all related project from the same call to make sense of the emerging results from all the projects. We took part also in co-authoring a jointly a white paper and a research paper with most other projects including participation in a health panel. Other dissemination tools helped reach a steady growth of followers. The project agreed on exploitation plan with a primary plan based on the framework connected to secondary outcomes which include the conversational application, smarshift scheduling, the adaptive assembly station and the entity recommender.
The framework is supported by catalogues of Active ageing principles for technology design and recommendations and change strategies for active ageing. The project also collected and formulated ethical guidelines in particular considering adaptive and intelligent technologies and ageing workers. The development of all technologies in the project was characterised by human centredness, participatory design, and iterative development with stakeholders.
The project implemented successfully three adaptive systems at work. This includes guidelines for adopting smart shift scheduling, an Adaptive Assembly Workstation (AAW) equipped with a cobot that is capable of adapting to the operator's internal state during the work cycle, a task based recommender system to support workers at the computer with useful information based on context published open source.
CO-ADAPT also delivered in open source a privacy preserving and secure platform to collect at the study sites participant data. An analytics portal app offers a front-end to the platform for managing the organisational and user entities involved in studies, as well as for offering the analytics visualisation. The project developed a conversational system with Deep Neural Network dialogue models.
The project organised successfully evaluation trials for four different adaptive technologies meeting improvement close to or above 2 digit percentage over baselines without the systems. The Smart Shift Scheduling (particpant n= 5 519) show beneficial effects on the primary outcome, sickness absence, with a reduction of nearly 10% in absences among hospital employees, as well as positive results for cost saving using participatory scheduling and wellbeing. The adaptive assembly station resulted in a 2-digit improvement in performance (i.e. a reduction of the time on task) and a 2-digit decrease in stress and mental load. An overall positive subjective experience was reported by all participants.
We completed the evaluation and development of a novel entity recommender system demonstrating target 2 digit improvement was met considering task performance and operationalised through proportion of relevant information found.
The project also evaluated the impact of a conversational application in a real case of psychological support. Analysis conducted on psychological test results, administered before the intervention and at the end of the same, showed improvements in well-being and motivation for the group of participants that used a conversational application for which a 2-digit percent increase was recorded. The project also organised a high level workshop inviting all related project from the same call to make sense of the emerging results from all the projects. We took part also in co-authoring a jointly a white paper and a research paper with most other projects including participation in a health panel. Other dissemination tools helped reach a steady growth of followers. The project agreed on exploitation plan with a primary plan based on the framework connected to secondary outcomes which include the conversational application, smarshift scheduling, the adaptive assembly station and the entity recommender.
The CO-ADAPT project contributed to demonstrate how Smart Shift scheduling can save costs and improve wellbeing of shift workers, this is a important innovation with striking benefits for the European society considering the proportion of shift workers in Europe is around 18%.
CO-ADAPT developed and evaluated collaborative robotics in an adaptive assembly station, as one of the first studies of such technologies focusing on ageing workers.
The project developed a digital assistant that for the first time works across systems and applications and is able to recommend information based on the specific tasks of a user. We demonstrated how this is useful for all users but in particular for ageing computer users.
CO-ADAPT developed a Conversational Agent which goes beyond the state of the art for its ability to understand the user input in natural language and customise dialogs, but also for the presence of different types of personalised contents, contexualized with respect to the model’s comprehension of the users’ input during the interactions.
The developed technologies were all evaluated and demonstrated improvements in key outcomes connected to the work activity close or over 2 digit percentage. This is a very important indication to consider how smart and participatory work organization, digital assistants, and cobots can bring important benefit for all workers but in particular for ageing workers.
CO-ADAPT developed and evaluated collaborative robotics in an adaptive assembly station, as one of the first studies of such technologies focusing on ageing workers.
The project developed a digital assistant that for the first time works across systems and applications and is able to recommend information based on the specific tasks of a user. We demonstrated how this is useful for all users but in particular for ageing computer users.
CO-ADAPT developed a Conversational Agent which goes beyond the state of the art for its ability to understand the user input in natural language and customise dialogs, but also for the presence of different types of personalised contents, contexualized with respect to the model’s comprehension of the users’ input during the interactions.
The developed technologies were all evaluated and demonstrated improvements in key outcomes connected to the work activity close or over 2 digit percentage. This is a very important indication to consider how smart and participatory work organization, digital assistants, and cobots can bring important benefit for all workers but in particular for ageing workers.