Exploring and Extending Human-Centered Design to Develop AI-Enabled Wellbeing Technology in Healthcare

In Europe, the working conditions in healthcare are undergoing major changes as digital transformation extends more and more widely to different job descriptions. The use of mobile technologies, telemedicine and other digital tools intended to support clinical decisions has improved health workers’ performance and mental health, as well as their competencies [1]. Digitalisation also requires new ethical reflection skills from healthcare professionals to understand factors of guiding and promoting ethical approaches [2]. There is an urgent need for effective digital tools and technologies and an unprecedented rush to implement eHealth services, including telemedicine consultation and digital contact tracing, in countries across the WHO Region. Strategic alignments are made to support this change [1, 3]. The need for digitalisation is due to the ageing of the population in Europe [4], the shortage of healthcare professionals in all occupational groups [5, 6] and the aftermath of the COVID-19 pandemic [1]. Different kinds of AI applications are predicted to have a growing role in healthcare and wellbeing devices and services [7]. The recent European Artificial Intelligence Act is going to change the use of AI within the EU region in the upcoming years [8].

The maturity of healthcare information management varies from country to country. Also, citizens’ skills in the use of digital technology vary in different countries. In Finland, almost 80% of adult age of citizens have at least a low level of digital skills [7].

There is also variation in digitalisation and informatics skills in different countries and between professions. The introduction and smooth use of new technology require expertise from the individual, but the maturity level of digitalisation in society and organizations is also important [9]. In healthcare, devices must meet the regulation for a medical device [10, 11], but wellbeing technology is also widely used by citizens, and they utilize the information they produce to maintain their own well-being [12].

Digitalisation is most welcome, but there are risks involved, for example, in terms of human aspects of care and undermining of fundamental rights [7]. Multiple health-related institutions and stakeholders, including World Health Organization (WHO), are promoting the adoption and scale-up of digital health technologies (DHT) innovations worldwide. These promotional initiatives aim to translate scientific research into action and enhance knowledge through scientific engagement, assessing and linking geographical needs with innovation pipelines, and implementing practical approaches that balance the benefits and risks of DHTs [13].

Studies have shown that mobile applications (apps) can effectively support lifestyle-related health. Demographic and personal factors of the target group should be considered when developing health apps. The inclusion of appropriate functionalities and their personalization can ensure a high uptake of health apps in workplaces [14].

Conversational AI and chatbots have been used in the last decade to improve access to mental health services [15‐21]. Chatbots are automated systems which replicate users’ behavior on one side of the chat communication. They are mimic systems which imitate the conversations between two individuals [22]. Chatbots can facilitate interactions with those who are reluctant to seek mental health advice due to stigmatization and allow more conversational flexibility [15, 16]. Threats to the chatbot include the cost of cloud services, the still-developing field of AI, and the unethical over-imitation of a human therapist or its replacement [16, 21].

When designing medical devices and wellness technology, it is crucial to consider evidence-based design and experience-based approaches in service design [23]. The ethical perspective is also essential [2], particularly during the design process, which involves transdisciplinary cooperation with professionals from various sectors [24].

Service design can offer a method to research and develop AI-enabled solutions in the complex healthcare sector. These approaches can inspire and support individuals to participate in the development process [25‐27]. HCD is the design approach that centers people and their needs, motivations, emotions, behavior, and perspective in the development of a design. Both users and service provider stakeholders are involved in design activities during and potentially after the service design process, supporting the change that co-design brings [27]. Multiprofessional, cross-sectoral healthcare involving multiple care system levels is a design context that requires context-specific knowledge, such as evidence-based care and specific design competencies, to include the perspectives of diverse actors in design processes [26].

Service design can promote transdisciplinary and cross-disciplinary processes encompassing numerous stakeholders, scientific orientations, and perspectives. It can clarify how to work together to ensure all aspects are considered when innovating new or developing existing technologies [24].

This study aims to elucidate the application and insights gained from the Service design process with HCD approach while developing an AI-enabled empowering solution, Voima-chatbot. The paper provides a detailed account of the development phases, using the case of Voima-chatbot as an exemplar. The objective is to enhance understanding of the feasibility of the service design process with HCD approach in developing AI-enabled technological solutions and point out future research insights.

The research question is:

What are the implications and results of applying Human-Centric design in developing an AI-enabled technological solution?

This study was conducted as research through design, meaning that design was an integral part of the process, providing both the data for the research and the practical artefacts from the workshops and other interventions, such as the ideation questionnaire. By using research through design [28, 29], a dual benefit was obtained: the design process with HCD approach, along with the co-designed artefacts, helps to better understand both the factors affecting the development of an AI-enabled chatbot and gaining understanding of the healthcare workers occupational well-being.

The development process of Voima-chatbot was part of a wider project that focused on the well-being of healthcare workers with pre-existing system stresses relating to resource constraints, crisis management, growing demand, recruitment, and retention. The project involved several phases from 2021 to 2023, with different methods, participants, and results (see Table 2).

Voima-chatbot was developed to build a scalable digital well-being service (non-medical device) that supports the well-being of healthcare professionals with empowering and solution-focused methods. Voima-chatbot utilizes an asset-based approach, which means that it works within the individual’s own world of meanings, supporting a functional interpersonal relationship and pursuits towards activating one’s own abilities, skills, strengths, and assets to enable a positive change [46, 47].

Voima-chatbot is an AI-enabled technological solution that uses a conversational AI platform to operate. Conversational AI is based on several advanced technological components, such as Natural Language Processing (NLP), Machine Learning, intent recognition, entity extraction and speech-to-text converters [48]. Voima-chatbot is not a medical device or therapy [11]. Its area of use is in the early prevention level as a well-being device.

Table 1 provides a comprehensive overview of the development process in the project, detailing the various phases, methods used, participants involved, and the deliverables at each stage. This chapter highlights the current Service design process for AI-enabled services adhered to an HCD approach, but also the problems that were resolved or not resolved with HCD approach and what are the findings for future development of service design process for AI-enabled services with HCD approach in healthcare. This model is illustrated earlier in Fig. 1.

Next, this article presents and evaluates the phases seen in Table 1. Based on the Service Design process with Human-Centered approaches and HCAI principles.

The objective of this research was to augment the comprehension of the feasibility of the HCD approach in creating AI-enabled technological solutions. This was accomplished during the development process of an AI-enabled solution, Voima-chatbot. The application and insights derived from the service design process, which utilized HCD design approach for AI-enabled services with 10 phases [41, 42] is presented in Fig. 1. The model incorporated three levels: technology, business, and design.

In addressing the research question - ‘What are the implications and results of applying Human-Centric design in the development of an AI-enabled technological solution?’ - a methodology of research through design was employed.

These development process phases produced information analyzed through HCD and HCAI point of views. Results emphasized the importance of observing transdisciplinary and cross-disciplinary processes, which involve numerous stakeholders and scientific orientations when working in the healthcare context.

This study has identified various strategies that merit integration into the model delineated by Jylkäs et al. [41, 42]. The healthcare context [23, 26], where understanding the various factors at play is crucial, but also research and evidence-based research when applied in healthcare [23]. It is mandatory when there is a development process for medical devices [10, 11].

Ethics and sustainability [2, 35, 38] factors guiding and promoting ethical activities contribute to the realization of the reflective process of ethical and sustainable activities. In ethical problem-solving, professionals base their judgement on legislation and ethical guidelines as well as on the ethical basis of social and healthcare. Ethical activities are promoted and facilitated, for example, through ethical management, organizational structures, and operational culture [2]. Sustainability also refers to Future insights into AI-enabled services. AI-enabled technological solutions are constantly developing [59]. Information safety and security, regulations, laws, EU data interoperability and policies [7, 8, 11]. It is one of the fundamental elements to decide if the application is a medical device or not. The categorization as a medical device brings a plethora of laws and regulations that necessitate careful consideration throughout the process [11]. Throughout this process, deliberations were held regarding the scope of the device and the possibilities of technical solutions. At the inception of this process, a decision was made to engineer a device aimed at promoting well-being.

Enhanced understanding of the development process, user needs, and expectations during the development process was cultivated through the application of Human-Centered design methodologies employed in the developmental stages (refer to Table 2). This research posits that the design process can foster and elucidate collaborative efforts when stakeholders are acknowledged during the process. This ensures comprehensive consideration of all facets when innovating new technologies or refining existing ones [25, 26], for example, Technology Readiness Levels (TRL) [45].

User-testing technique yielded data indicating a need for more understanding of human behavior, interactions, and the human-machine relationship. It was also acknowledged that conventional approaches, such as questionnaires, might be too simplistic, with the small number of participants skewed for obtaining this information. It would be beneficial to enhance understanding of different facets of human-AI interaction, such as emotions, cognition, assets, mutual learning, or failure/success. Interfaces such as chatbots are shared boundaries between the sociotechnical systems of computers, connecting hardware, software, and human users. Ethnographical research can incorporate technical walkthroughs and interfaces to complement participant observations in local settings where interfaces are accessed. Interface ethnography can be utilized with multi-sited fieldwork designs since interfaces are components of transnational networks and mediate between different actors [60].

Pervasive themes for all stages of the development process were comprehending the context where the user and the intended technological solution are situated. Collecting user insights and genuine material for testing from diverse range of people. Humans can provide training data for machine learning applications and directly perform tasks that are challenging for computers in the pipeline with the assistance of machine-based approaches. This is a way to avoid biases [38, 61]. Developing the AI solution with transdisciplinary team is essential. It was learned that when healthcare professionals are involved in the process understanding on AI-enabled services was built among the participants [9, 13, 26].

Accompanying the findings of this research to the existing Jylkäs et al. model, it was contemplated that a broader perspective and implications of implementing Life-Centered design ought to be addressed incorporating Human-Centered design [35, 62]. This research could inform the development of similar AI-enabled health technologies by presenting a need for updating a new model and methods for the design process with a Life-centered design approach of AI-enabled solutions that could be prototyped and tested in the following research. Life-centered design, for example, expands human-centered design to include consideration for nature and vulnerable humans by merging practices such as circular design, biomimicry, systems thinking, and futuring, and aligning designers with global goals, such as the United Nation’s Sustainable Development Goals [62‐64]. As life-centered design is still emerging, its practices vary and is practiced by only a few. More research is needed to evaluate Life-centered design in digital services that utilize AI.