Shaping artificial intelligence – an example from nursing care

The development of artificial intelligence (AI) as a tool to analyse and understand data has gained momentum. AI began as early as the 1950s. However, AI is only really unfolding and improving its performance in recent years. The reason for this is improved computing power, access to large amounts of data, high storage capacity and more advanced statistical analysis methods.

AI has many definitions. The term intelligence is still unclear today, but is used nonetheless. A suitable definition is: “Artificial intelligence is an umbrella term for methods, algorithms and systems to implement (apparently) intelligent behaviour in a computer system.” (Auer et al., 2019, p.35). It is therefore about procedures that computers are taught to perceive, communicate, think, reason and make decisions similarly or better than humans (Castellanos, 2018). AI is particularly relevant for healthcare and especially nursing, the largest professional group in healthcare. This is because nurses generate and collect a large amount of data in patient care, which provides information on patient outcomes, nursing interventions, healthcare services used, as well as key figures for management and administration. Moreover, nurses can be found in all care settings, i.e. from the acute, long-term and rehabilitation to the home setting.

Possible benefit

Why should the AI be used by nurses? The aim of AI is to provide nurses with the right recommendations at the right time in order to make the right decisions that have the best possible impact on patient outcomes. AI assists in the processing of data and information and provides insights and correlations. It draws on large amounts of data, learns from a multitude of cases to provide the best possible recommendations for individual situations. Based on these recommendations, it is then the task of the nurse or the interdisciplinary team to make the treatment and care decision with the involvement of the patient. The AI recommendation should guide but not determine the decision, as contextual knowledge, expertise, patient preference, scientific evidence and available resources must also be taken into account.

Current state of research

A literature review shows that the number of scientific publications on AI for the field of care has jumped since 2019 (O’Connor et al., submitted). Most of the studies conducted are based on hospital data and use a retrospective or observational design. Frequently used AI approaches are: supervised and unsupervised methods (which include various methods such as neural networks, random forest, support vector machine, logistic regression, decision tree or gradient boosting) and natural language processing. AI has so far been developed mainly for direct patient care, i.e. for intermediate care or intensive care units, for wound care, delirium or discharge management as well as for the care of elderly people in the area of falls, infection and health emergency situations. Other AI application areas include nursing documentation, nursing language, management (staff), administration and education (skills) (O’Connor et al., 2021). Most AI developments in the nursing/healthcare sector have taken place in the USA and just under 15% in Europe.

Knowledge gap

This look at the literature revealed two striking limitations of previous research: First, the nursing profession is still hardly involved in the development, testing and application of AI solutions. That is, in 33% of the studies analysed, nurses led an AI project. In all other studies, it was other disciplines, e.g. computer scientists or physicians, who developed AI solutions that were applied in the field of nursing and other health professionals. Secondly, there are still few studies that show implementation successes of AI solutions in clinical practice, such as improved clinical decision-making and improved patient outcomes, but this may be related to the current state of use of AI solutions in clinical practice. AI currently focuses on data-based development and testing of AI solutions without implementing AI systems in daily clinical practice for the use of nurses, physicians or other healthcare professionals. This state of research suggests that competencies needed for AI development are not yet widespread among nurses, especially in Europe. Such competencies include data processing, representation and communication interdisciplinary with computer scientists or technology developers, among others. Further competences, i.e. data-based implementation of measures and their evaluation, changed interaction between care and patient (virtual contacts, monitoring) or the optimisation of work processes (standardisation, automation), would enable nurses to lead AI projects.

Measures needed

In order to promote the development of AI solutions for everyday clinical practice, nurses need to be empowered (transferable to other health professionals). This means they need competencies to drive the development, testing, implementation and evaluation of AI applications. The aim is for nurses to acquire competencies that enable them to assess the validity of data used for AI development and training. They should be able to ensure and assess to what extent the results generated by AI make sense, whether they are trustworthy enough and how they can be used in patient care. To achieve this goal, the following steps are useful:

  • Bachelor’s and Master’s degree programmes should integrate basic knowledge of AI into curricula.
  • Specialised further education courses, e.g. Certificate of Advances Studies (CAS) or Master of Advanced Studies (MAS), should enable the basic and profession-specific application of AI. It is conceivable that students from both informatics or data science and health professions are trained together, with corresponding specialisations. In this way, a common language can be established, from which both professions can benefit by learning to understand the “glasses on the world” of the other profession.
  • In clinical practice, an innovation-oriented culture should be promoted by leaders and the necessary resources (knowledge, technology, prioritisation, evaluation, outcome evaluation) should be provided.
  • Clinical practice teams should be sensitised to the fact that AI is another way of analysing and using data and information for patient care.
  • Staff especially qualified in data use should be integrated into clinical departments, either nurses specialised in AI or data scientists willing to understand the language of the nursing profession, in order to (further) develop and implement meaningful and practice-relevant AI applications in close cooperation.
  • The start of the development and testing of AI solutions should always address a concrete problem that can be objectified in clinical practice, in which there is great interest and which is therefore given priority in terms of time. It is important that all stakeholders of the problem are involved so that it can be comprehensively understood and addressed by an AI solution.
  • Risks of AI applications should be taken seriously, e.g. risks of discriminatory or erroneous decisions, but also lack of data quality or lack of access to data, because these factors negatively influence the validity of AI recommendations.
  • Research and clinical practice should consistently concretise and demonstrate the benefits and added value of AI solutions in a situation-specific manner.
  • Research and clinical practice should systematically clarify where AI applications can optimally contribute to process control.
  • One requirement for AI solutions should be that a recognisable added value for patient care is defined, which, for example, “frees up” time for direct patient contact.
  • The AI applications should be designed to be easy to use; in particular, the AI recommendations should be intuitively understandable.
  • A discourse is needed on the factors that promote the acceptance of AI solutions in clinical practice.
  • Positive application examples that are spontaneously plausible and can be passed on verbally (storytelling) should be used for communication in order to promote acceptance.
  • Good practice examples from other countries should be collected and analysed to better understand how AI can be successfully introduced in clinical practice

Conclusion and outlook

Future-oriented care should address the AI issue if it wants to remain or be competent, person-centred, efficient, effective, of high quality and economical. AI offers the possibility to relieve caregivers through automation and to take empirical data into account when making decisions, as a supplement to their personal assessment. However, especially in the case of the use of AI, informational self-determination and adequate human interaction should be maintained. For a successful use of AI in care, however, there are still technological challenges to be overcome, including data quality, practical data management and tools for data interpretation, e.g. the visualisation of data. In addition, a societal discourse and political framework conditions are needed to define needs, quality assurances and standardisations for the AI ecosystem.


  1. Buchanan C, Howitt ML, Wilson R, Booth RG, Risling T, Bamford M, Predicted Influences of Artificial Intelligence on the Domains of Nursing: scoping review, JMIR Nursing 2020;3(1):e23939. doi: 10.2196/23939
  2. Boden, M.A., Artificial Intelligence: A Very Short Introduction. 2018, Oxford: Oxford 20 University Press.
  3. Castellanos S. What exactly is artificial intelligence. The Wall Street Journal. 2018.
  4. Auer C, Hollenstein N, Reumann M (2019) Artificial intelligence in healthcare. In: Health digital. Springer, Berlin, Heidelberg, pp 33-46. artificial intelligence in healthcare | SpringerLink
  5. Mcafee A., and Brynjolfsson, E., Machine Platform Crowd. How to make the most of our digital future. 2018, Kulmbach: Stock Exchange Media.
  6. O’Connor, S. (2021). Artificial intelligence and predictive analytics in nursing education. Nurse Education in Practice, 56, 103224-103224.
  7. O’Connor, S., Yan, Y., Thilo, F., Felzmann, H., Dowding, D., Lee, J.J., 2021. artificial intelligence in nursing: a literature review. Intell. Med. submitted for publication.
  8. Riedl R, Situational quality management – Countering the discomfort with artificial intelligence, Psychiatric Nursing (accepted)
  9. Robert N. (2019). How artificial intelligence is changing nursing. Nursing management, 50(9), 30-39.
Creative Commons Licence

AUTHOR: Friederike J. S. Thilo

Prof. Dr Friederike Thilo is Head of Innovation Field "Digital Health", aF&E Nursing, BFH Health. Her research focuses are: Design collaboration human and machine; technology acceptance; need-driven development, testing and evaluation technologies in the context of health/disease; data-based care (artificial intelligence).

AUTHOR: Reinhard Riedl

Prof. Dr Reinhard Riedl is a lecturer at the Institute of Digital Technology Management at BFH Wirtschaft. He is involved in many organisations and is, among other things, Vice-President of the Swiss E-Government Symposium and a member of the steering committee of TA-Swiss. He is also a board member of, Praevenire - Verein zur Optimierung der solidarischen Gesundheitsversorgung (Austria) and, among others.

AUTHOR: Siobhan O'Connor

Dr Siobhan O'Connor is a Lecturer at the School of Nursing and Midwifery, National University of Ireland Galway and an Adjunct Assistant Professor at Western University, Canada. She has a multidisciplinary background in both nursing and information systems, and her research focuses on the design, implement, and use of technologies for patient self-management.

Create PDF

Related Posts

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *