INTRODUCTION
AI aims to mimic human cognitive functions and allows the collection and analysis of large amounts of data (Jiang et al., 2017). For this reason, it is increasingly widespread in healthcare (Masucci et al., 2021), in activities such as anamnesis, diagnosis support, treatment planning, monitoring of the pati- ent’s condition (Reddy et al., 2019), offering clinicians the opportunity to tailor early interventions to each individual (Subramanian et al., 2020). AI can also help detect early disease risks as the system uses past and present information about health problems and can identify whether the patient is at risk of contracting a disease or not (Becker, 2019). Al could lead to greater empowerment of the patient (Polese and Carrubbo, 2016) and improve the bond of trust between doctor and patient. AI is expanding into areas previ- ously only managed by human experts (Yu et al., 2018), but machines can only support and improve human roles (Barile et al., 2021).
© 2022. Published by AHFE Open Access. All rights reserved. 210
Augmented Decision Making Model for Responsible Actors in Healthcare 211
However, many AI algorithms are difficult to understand or explain and actors fear mistakes in diagnosis and treatments (Davenport and Kalakota, 2019), with negative effects on actors’ trust in AI systems (Das, 2020). Ensu- ring transparency, explainability and intelligibility in AI systems in healthcare is one of the six principles identified by the WHO (WHO, 2021) to under- stand how to provide health systems with high quality, fair and reliable decisions (de Fine Licht and de Fine Licht, 2020). The role played by transpa- rency in making AI systems effective decision support systems is particularly interesting.
For this reason, this work tries to answer the following research question:
- R.Q.: is the use of AI-based tools sufficient to enable augmented decision- making processes for responsible health actors?
Following the ‘theory synthesis’ (Jaakkola, 2020) in intending how to give back drivers and suggestions to researchers and practitioners in such a way, for properly understanding the design and use of AI, in order to pur- sue win-win conditions for the benefit of all parties involved to co-create value (Polese et al., 2017), this conceptual paper starts with the description of the theoretical background. In the next section, the challenges posed by AI systems in healthcare are described to frame an augmented decision-making model based on AI. However, many doubts are relieved in terms of tran- sparency so the co-design of AI systems for healthcare (Share-to-Care) is proposed as solution, to encourage a great acceptance of new technologies in healthcare thanks to reasoned transparency and to allow an augmented decision-making process through actors’ capabilities co-elevation. In the last part, non-conclusive considerations are presented.
THEORETICAL BACKGROUND Service Innovation and Data-Driven Decision-Making
Today innovation is increasingly associated with the function of encouraging companies in pursuing and maintaining their survival, through the achie- vement of a competitive advantage (Damanpour, 2010). Service innovation is achieved through a fruitful combination of technological elements, social relations, organizational adjustments and commercial interactions, based on a human-centered perspective (Yu and Sangiorgi, 2018). Service innovation occurs within networks (Vargo et al., 2015) when existing value propositions are modified through a process of integrating existing resources or by inven- ting new resources (Åkesson et al., 2016) and involves the creation, renewal and transformation of pre-existing knowledge. Innovation cannot be exclu- sively linked to the use of new technologies (Koskela-Huotari et al., 2016); they have the “task” of providing data. The use of data and technologies is closely linked to human behavior (Spohrer and Maglio, 2010). Innovation implies that through technologies responsible entities may improve them- selves (Spohrer, 2021), and, as service systems are centered on people, it occurs when people, thanks to technologies, are able to optimize service systems operation (Maglio, 2015). Data-driven is a mindset that sees data
as functional strategic resources for increasingly effective and timely decisi- ons (Provost and Fawcett, 2013), rather than as mere outputs of intuitions and experiences (Polese et al., 2018), first in Healthcare (Capunzo et al., 2013; Carrubbo et al., 2013). Data-driven decision-making can improve the competitiveness of companies and enable them to pursue their marketing objectives (Troisi et al., 2020; Carrubbo et al., 2021). In the health sector, a complex context increasingly centered on the patient, the collection and analysis of patient data could allow doctors to intervene in an increasin- gly timely and precise manner (Sakr and Elgammal, 2016), with an overall improvement in general performance (McColl-Kennedy et al., 2012) and response times (Dautov et al., 2019). Data-driven decision-making involves the use of advanced analytics tools, such as artificial intelligence solutions and data management systems. The ability to acquire and analyze data in real-time thanks to the application of artificial intelligence algorithms allows companies to comply in a better way with companies’ goals and initiatives (Brynjolfsson and McElheran, 2016).
TOWARDS AN AUGMENTED DECISION-MAKING MODEL BASED ON AI SYSTEMS From Artificial Intelligence (AI) to Intelligence Augmentation (IA)
AI is the combination of software techniques and IT infrastructures, allo- wing performances comparable to human intelligence (Topol, 2019) and data-driven decision-making processes. AI is a widely debated topic in the literature: some believe that over time it can exceed human performance (Grace et al., 2018) and others suppose that Yulya, a humanoid, could win the Nobel Prize for medicine in 2036. AI researchers aim to understand how to make people wiser by enhancing their decision-making and skills, provi- ding them with otherwise inaccessible information based on data (Zhou et al., 2021) thanks to AI systems. IA implies a mixture of AI technologies and peo- ple: through technologies, people may amplify their capabilities (Barile et al., 2018) and optimize their thought structure by improving their interpretative schemas (Barile et al., 2020). IA implies a perfect human-machine interaction in which machines do what they do best to help humans do what humans do best (thinking) (Zhou et al., 2021). IA can mitigate fears regarding the use of AI systems (Mohanty and Vyas, 2018). IA is intelligence that derives from a collaborative and effective human-machine interaction and integration; it allows systems to evolve into more intelligent and wiser configurations, based on rational and emotional components (Barile et al., 2021; Badr et al., 2021).
From Automated Decision to Augmented Decision There are doubts about the adequacy of the information released by the AI systems and fears in terms of privacy protection, reliability, accuracy and security of the tool (Lytras and Visvizi, 2021) that may invalidate its use by health professionals (Shinners et al., 2020) and patients (Reddy et al., 2020). It implies that the data available to health decision-makers are often partial, fragmented and unstructured, with doubts about their quality (de Fine Licht
Augmented Decision Making Model for Responsible Actors in Healthcare 213
and de Fine Licht, 2020) and fears that the automated decision may be ina- dequate. Greater transparency about the functioning of such systems could lead to a more profitable collaborative relationship between the parties, sti- mulating an environment able to encourage phenomena of value co-creation (Mabillard et al., 2021). The augmented decision may be the solution: an automated decision enhanced by decision-makers’ knowledge. The reaso- ned transparency can allow a decipherable and intelligible functioning of AI systems’ black box (Wischmeyer, 2020). More understandable, user-friendly, usable, explainable and effective tools would favor decision-making proces- ses more clear, reasoned and based on objective evidence. The augmented decision is not just an automated decision, which frightens and invalidates the use (or even only the correct use) of these tools by users, but it is enhanced and increased thanks to the reasoned transparency.
From Data-Driven Decision-Making to Augmented Decision-Making AI may allow a strengthened decision-making process (Raisch and Krakowski, 2021) based on an augmented decision-making process that implies a perfect human-machine interaction in which both two entities can exceed their limitations (Keding and Meissner, 2021). The augmented decision-making process, as understood in this work, is not only enhanced thanks to the use of new technologies that provide decision-makers with data and sophisticated calculation tools but is a decision-making process enabled and improved by the impact that new technologies can have on decision- makers’ cognitive functions. The cognitive and deductive potential of the individual may be amplified thanks to good quality data and interpreta- tive schemes (improved by technology) and information units (increased by technology) integrated with decision-makers’ value categories (Barile et al. 2021). For this, it is necessary to understand how individuals in healthcare are willing to use AI and how to act to improve this acceptance degree.
REASONED TRANSPARENCY FOR AUGMENTED DECISION-MAKING IN HEALTHCARE Reasoned transparency would incentivize patients to release their data and enable healthcare professionals/decision-makers to make more reliable deci- sions. To pursue an augmented decision-making model, it is necessary to consider users’ responsibility.
“Share-to-Care”: AI Systems Co-Design to Further a Great Acceptance of New Technologies in Healthcare An augmented decision in healthcare is possible if doctors and support staff understand the decision support system used functioning and handle its results if necessary. Good quality and not fragmented data are issued by patients aware of being active parts of a process aimed to guarantee their well-being. This may be possible thanks to the implementation of peer-to-peer design and production platforms, where patients, specialists, designers and manufacturers can collaborate to maximize health activities and actors’ learning and awareness. The involvement of all users may be
useful to understand the problem. For this reason, new technology early adopters will have to help the designers to define the problem and to look for a solution (e.g. a new digital tool). This would contribute to the reaso- ned transparency that would mitigate patients’ fears in terms of AI systems transparency. Felzmann et al. (2020) proposed the Transparency by Design model to support organizations in proposing transparent AI systems. The first phase, considered in this work, concerns the transparency in the design of AI systems; its indicators are the proactivity (technologies have to embody communities’ values), integration (companies have to explain transparency measures and standards), and the focus on the audience, (decisions have to be accessible to people). These indicators have to prevent distrust and suspi- cion and to guarantee the public’s perception of reliability. Bianchini et al. (2020), in a study carried out by the Department of Design of the Polite- cnico di Milano together with the pharmaceutical company Sanofi, propose the approach “Wear-to-Care” to design and materialize smart open-source wearables, DermAware, for atopic dermatitis patients, based on open and bottom-up innovation. The present study proposes to generalize the model proposed by Bianchini et al. (2020) to any other tool based on AI systems for any other patients’ disease.
Phase 1: Problematizing starting from patients and doctors’ points of view This phase aims to understand patients’ needs and behaviors, so the com- pany has to search for a sample of patients and doctors through the databases of health facilities, to carry out interviews and favor the emergence of relevant factors for them. To give the research greater robustness, online questionnai- res to citizens will also be administered to confirm or disavow what emerged from the interviews.
Phase 2: Designing and prototyping a new product-service concept With the data analysis thus collected in the first phase, it may be pos- sible to define the idea thanks to the joint definition of the problem to be solved. In this phase, there is the design of the tool, based on the specific medical needs, relevant categories that emerged from the previous investiga- tion (phase 1) and the technical potential of the technology. The following step is the production of the prototype.
Phase 3: Testing the prototype The last phase involves testing the prototype by the sample of doctors and patients interviewed in phase 1 to assess whether a revision is necessary for the prototype design phase or the new technological solution can be launched on the market.
The bottom-up design would allow to respect the functional dimension of the tool, (possible also with a top-down approach) but also to make the patient trigger the whole innovation process.
Enabling Augmented Decision-Making Through Actors’ Capabilities Co-Elevation and Upskilling Share-to-Care may enable augmented decisions, but it’s not sufficient the use new technologies, it’s necessary that data used for decisions, provided by new technologies, are completed and reliable (R.Q.). The introduction of new
Augmented Decision Making Model for Responsible Actors in Healthcare 215
co-designed technological solutions, which can be well accepted and effecti- vely used by users/tools co-creators, may enhance the healthcare offer and may increase patient trust and confidence in AI systems. Thanks to Share- to-Care, actors become more responsible and a better version of themselves (Spohrer, 2021); their capabilities may co-evolve and upskill, which means that AI designers are perfectly capable of launching tools that are not meant to die in their introduction phase because patients are willing and able to use them and thus doctors may optimize the results of their decisions. The increased patients’ knowledge about the tools with which health professio- nals develop diagnoses and treatments could improve their involvement and willingness to release resources (time, money, trust, data). The co-design of AI systems for healthcare would make the tool more understandable for users and more user-friendly (proactivity). From the interviews and questionnai- res, it could be possible to understand patients’ functional and social needs, useful to establish the standards on which the automated decision is based (integration). The bottom-up approach to innovation allows a better predi- sposition of patients to accept technologies, improves their skills, provides health decision-makers with data useful for a democratization of the deci- sion, and makes the new technologies easier to use and more understandable for actors. The new technology can be customized according to users’ expe- ctations, needs and characteristics (focus on the audience). This also lets to more personalize health services and make them more precise and timely, thanks to qualitatively more reliable data thanks to a better degree of trust, confidence and knowledge of users, data-driven decision-making processes, based on augmented decisions, are possible.
CONCLUDING REMARKS AND PRACTICAL IMPLICATIONS This work suggests that the impact of AI systems in healthcare has to be tre- ated with a multidisciplinary interpretation (medical, IT, management) and proposes an update in terms of augmented decision-making: new technolo- gies can enhance people cognitive functions, thanks to an information flow which would otherwise be difficult to process, but this information flow have to derive from good quality data for which a good predisposition to release them and a good degree of technology acceptance by users are necessary. Fol- lowing a phenomenological approach, “Share-to-Care” framework, for the co-design of AI systems deemed potentially acceptable by end-users, is pro- posed. The Share-to-Care framework proposes a user-centered approach to innovation in healthcare and it can have significant opportunities in terms of value co-creation. The main limitation of this work lies in its conceptual nature. A case study could be useful to validate the effectiveness of this insight and the framework proposed and to understand better if it’s can be appropri- ate for each health context and each disease. for example, patients suffering from neurodegenerative diseases, may be patients not able to be active, in participating in their care, as well as in the co-design of tools that allow moni- toring of their conditions. In this case, it should be understood which other actors should be considered and involved (for example their families) and how, the fact that they are not directly the patients, can affect their degree of
acceptance of the technology and their predisposition to use it. These aspects have to be explored in further research. Furthermore, the main implication lies in the multidisciplinary approach used to design AI systems in healthcare, usually affected by transparency fears which may undermine the perceived quality of service thus supplied. An in-depth study of AI systems that can be well accepted by end-users, because clearer and more explainable, may be useful to support the decision-making process in healthcare.
REFERENCES
Åkesson, M., Skålén, P., Edvardsson, B., Stålhammar, A. (2016). Value proposition test-driving for service innovation: How frontline employees innovate value propositions, Journal of Service Theory and Practice, volume 26, No. 3, 338–362. Badr, N.G., Carrubbo, L., Ruberto, M. (2021). Responding to COVID-19: Potential
Hospital-at-Home Solutions to Re-configure the Healthcare Service Ecosystem. In HEALTHINF, 344–351. Barile, S., Bassano, C., Lettieri, M., Piciocchi, P., Saviano, M. (2020) “Intelligence augmentation (IA) in complex decision making: a new view of the VSA concept of relevance”, in International Conference on Applied Human Factors and Ergonomics, pp. 251–258. Springer, Cham. Barile, S., Bassano, C., Piciocchi, P., Saviano, M., Spohrer, J. C. (2021). Empowering
value co-creation in the digital age, Journal of Business & Industrial Marketing. Barile, S., Piciocchi, P., Bassano, C., Spohrer, J., Pietronudo, M.C. (2018) “Redefining the role of artificial intelligence (AI) in wiser service systems!, in Inter- national Conference on Applied Human Factors and Ergonomics, pp. 159–170. Springer, Cham. Becker, A. (2019). AI in medicine: What is it doing for us today?, Health Policy and
Technology, Volume 8, No. 2, 198–205. Bianchini, M., Bolzan, P., Parini, B., Maffei, S., Filippo, C. (2021) “Wear-to-Care.
Co-designing the Next Wave of Open Wearables in the Healthcare Sector”, in International Conference on Wearables in Healthcare, pp. 122–138, Springer, Cham. Brynjolfsson, E., McElheran, K. (2016). The rapid adoption of data-driven decisionmaking, American Economic Review, Volume 106, No. 5, 133–39. Burton, J. W., Stein, M. K., Jensen, T. B. (2020). A systematic review of algorithm
aversion in augmented decision making, Journal of Behavioral Decision Making, Volume 33, No.2, 220–239. Capunzo, M., F. Polese, G. Boccia, L. Carrubbo, F. Clarizia, F. De Caro (2013), “Advances in service research for the understanding and the management of service in healthcare networks,” Gummesson, E., Mele, C., Polese, F. (eds), System Theory and Service Science: Integrating three perspectives in a new service agenda, Giannini, Napoli, pp. 14–17. Carrubbo, L., Clarizia, F., Hysa, X., Bilotta, A. (2013). New’smarter’solutions for
the healthcare complex service system. Gummesson, E., Mele, C., Polese, F. (eds), System Theory and Service Science: Integrating Three Perspectives In a New Service Agenda, Giannini, Napoli, 14–17. Carrubbo, L., Megaro, A., Notari, F. (2021). Data-Driven Decisions for a Smarter and Resilient Healthcare Service Ecosystem. In The International Research & Innovation Forum (pp. 69–77). Springer, Cham. Damanpour, F. (2010). An integration of research findings of effects of firm size
and market competition on product and process innovations, British Journal of Management, Volume 21, No. 4, 996–1010.
Augmented Decision Making Model for Responsible Actors in Healthcare 217
Das, A. (2020). Trust in “trust-free” digital networks: How inter-firm algorithmic relationships embed the cardinal principles of value co-creation, AIS Transactions on Human-Computer Interaction, Volume 12, No. 4, 228–252. Dautov, R., Distefano, S., Buyya, R. (2019). Hierarchical data fusion for Smart
Healthcare, Journal of Big Data, Volume 6, No. 1, 1–23. Davenport, T., Kalakota, R. (2019). The potential for AI in healthcare, Future healthcare journal, Volume 6, No. 2, 94. de Fine Licht, K., de Fine Licht, J. (2020). Artificial intelligence, transparency, and
public decision-making, AI & society, Volume 35, No. 4, 917–926. Felzmann, H., Fosch-Villaronga, E., Lutz, C., Tamò-Larrieux, A. (2020). Towards transparency by design for AI, Science and Engineering Ethics, Volume 26, No. 6, 3333–3361. Grace, K., Salvatier, J., Dafoe, A., Zhang, B., Evans, O. (2018). When will AI exceed
human performance? Evidence from AI experts, Journal of Artificial Intelligence Research, Volume 62, 729–754. Jaakkola, E. (2020). Designing conceptual articles: four approaches, AMS review, Volume 10, No. 1, 18–26. Jiang, F., Jiang, Y., Zhi, H., Dong, Y., Li, H., Ma, S., ... Wang, Y. (2017). AI in
healthcare: past, present and future, Stroke and vascular neurology, Volume 2, No. 4. Keding, C., & Meissner, P. (2021). Managerial overreliance on AI-augmented decision-making processes: How the use of AI-based advisory systems shapes choice behavior in R&D investment decisions. Technological Forecasting and Social Change, 171, 120970. Koskela-Huotari, K., Edvardsson, B., Jonas, J. M., Sörhammar, D., Witell, L. (2016).
Innovation in service ecosystems - Breaking, making, and maintaining instituti- onalized rules of resource integration, Journal of Business Research, Volume 69, No. 8, 2964–2971. Lytras, M. D., Visvizi, A. (2021). Artificial intelligence and cognitive computing: Methods, technologies, systems, applications and policy making, Sustainability, Volume 13, No. 7, 3598. Mabillard, V., Demartines, N., Joliat, G. R. (2021). How Can Reasoned Transpa-
rency Enhance Co-Creation in Health Care and Remedy the Pitfalls of Digitiza- tion in Doctor-Patient Relationships?, International journal of health policy and management, 1–5. Maglio, P. P. (2015). Smart Service Systems, Human-Centered Service Systems, and the Mission of Service Science, Service Science, Volume 7, No. 2, ii-iii. Masucci, A., Megaro, A. and Sirianni, C.A. (2021). The Role of New Technologies in
Value Co-Creation Processes: Healthcare Management and the National Health System as a System of Services, Journal of Service Science and Management, Volume 14, No. 2, 189–212. McColl-Kennedy, J. R., Vargo, S. L., Dagger, T. S., Sweeney, J. C., Kasteren, Y. V.
(2012). Health care customer value cocreation practice styles, Journal of service research, Volume 15, No. 4, 370–389. Mohanty, S., Vyas, S. (2018) “Putting it all together: Toward a human-machine collaborative ecosystem”, in How to Compete in the Age of Artificial Intelligence, pp. 215–229. Apress, Berkeley, CA. Polese, F., Carrubbo, L. (2016). Eco-sistemi di servizio in sanità. G Giappichelli
Editore. Polese, F., Carrubbo, L., Bruni, R., Maione, G. (2017). The viable system perspective of actors in eco-systems, The TQM Journal, Volume 29, No. 6, 783–799.
Polese, F., Carrubbo, L., Caputo, F., Sarno, D. (2018). Managing H-SESs: Abstracting a sustainability-based view from hospitalization at home (HaH) practices, Sustainability, Volume 10, No. 11, p. 3951. Provost, F., Fawcett, T. (2013). Data science and its relationship to big data and
data-driven decision making, Big data, Volume 1, No. 1, 51-59. Raisch, S., Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox, Academy of Management Review, Volume 46, No. 1, 192–210. Reddy, S., Fox, J., Purohit, M.P. (2019). Artificial intelligence-enabled healthcare
delivery, Journal of the Royal Society of Medicine, Volume 112, No. 1, 22–28. Sakr, S., Elgammal, A. (2016). Towards a comprehensive data analytics framework for smart healthcare services, Big Data Research, Volume 4, 44-58. Shinners, L., Aggar, C., Grace, S., Smith, S. (2020). Exploring healthcare profes-
sionals’ understanding and experiences of AI technology use in the delivery of healthcare: an integrative review, Health informatics journal, Volume 26, No. 2, 1225–1236. Spohrer, J. (2021) “Service innovation roadmaps and responsible entities learning”, in ITM Web of Conferences (Vol. 38). EDP Sciences. Spohrer, J. C., Maglio, P. P. (2010) “Toward a science of service systems”, in
Handbook of service science, pp. 157–194. Springer, Boston, MA. Subramanian, M., Wojtusciszyn, A., Favre, L., Boughorbel, S., Shan, J., Letaief, K.
B., ... Chouchane, L. (2020). Precision medicine in the era of AI: implications in chronic disease management, Journal of translational medicine, Volume 18, No. 1, 1–12. Topol, E. J. (2019). High-performance medicine: the convergence of human and artificial intelligence, Nature medicine, Volume 25, No. 1, 44–56. Troisi, O., Maione, G., Grimaldi, M., Loia, F. (2020). Growth hacking: Insights on
data-driven decision-making from three firms, Industrial Marketing Management, Volume 90, 538–557. Vargo, S. L., Wieland, H., Akaka, M. A. (2015). Innovation through institutionalization: A service ecosystems perspective, Industrial Marketing Management, Volume 44, 63–72. Wischmeyer, T. (2020) “Artificial intelligence and transparency: opening the black
box”, in Regulating artificial intelligence, pp. 75-101. Springer, Cham. World Health Organization. (2021), Ethics and governance of AI for health: WHO guidance. Yu, E., Sangiorgi, D. (2018). Service design as an approach to implement the value
cocreation perspective in new service development, Journal of Service Research, Volume 21, No. 1, 40–58. Zhou, L., Paul, S., Demirkan, H., Yuan, L., Spohrer, J., Zhou, M., Basu, J. (2021).
Intelligence Augmentation: Towards Building Human-Machine Symbiotic Rela- tionship, AIS Transactions on Human-Computer Interaction, Volume 13, No. 2, 243–264.