Generative AI Solutions in Healthcare: Potential and Current Obstacles

The healthcare sector needs a digital transformation due to population growth, changing disease patterns, and rising costs. Generative AI models, such as GPT, offer great potential to improve the efficiency and effectiveness of healthcare, particularly through the automation of administrative processes and support in diagnosis. However, the implementation of these technologies requires modernizations of the digital infrastructure, especially improved interoperability and adaptable legal frameworks.

Noah Boenninghausen, M.Sc., is a Senior Business Analyst and part of the Healthcare team at BearingPoint. His focus is on researching and developing future-proof solutions in the healthcare sector.

Prof. Dr. Volker Nürnberg teaches at the Allensbach University of Applied Sciences Business Administration, is Head of Healthcare at BearingPoint and is a member of the Ethics Commission at the Federal Ministry of Health.

Healthcare is undergoing a digital transformation. The necessity for such a restructuring becomes clear when considering various developments: The rapid growth of the world's population presents the existing medical infrastructure with a highly asymmetrical supply-demand ratio (Pfannstiel et al., 2020). A changing spectrum of diseases, with an increase in multimorbidity and chronic conditions, results in a growing need for interdisciplinary and trans-sectoral collaboration (Ortloff and Schmitz, 2024). Furthermore, the emerging inflation of demands poses a cost explosion for healthcare due to constantly rising costs for the latest treatment services and innovative medical technology (Zhu and others., (2019). Challenges in the healthcare system are already pushing numerous clinics to the brink of bankruptcy. If we do not succeed in such a transformation, the provision of healthcare services will have to be intensified in the future using the RRP paradigm (rationing, rationalization, and prioritization), which could severely endanger the guarantee of the fundamental principle of solidarity in healthcare (Henke et al., 2022).

Consequently, alternative solutions are needed that replace outdated process, communication, and administrative structures with innovative, collaborative, and affordable solution offerings. Artificial intelligence (AI), big data, and wearable computer technologies, in conjunction with innovative management approaches, offer significant innovation potential to find answers to the increasing scarcity of resources in healthcare (Zhang and Kamel Boulos, 2023). It is therefore observed how scientists, experts, and entrepreneurs are adopting such innovations and reshaping the healthcare market through technological progress. A central component in such a restructuring is the ubiquitous use of Artificial Intelligence (AI). Among the rapidly developing AI technologies, generative AI models, such as the Generative Pre-trained Transformer (GPT) models developed by OpenAI, show remarkable potential for the existing challenges in healthcare (Zhang and Kamel Boulos, 2023). By utilizing large quantities of vital data and knowledge, GPT models can transform various aspects of healthcare and usher in a new era of clinical decision support, patient communication, and data management (Chui and others., 2023). Their potential to process and interpret complex medical information, thereby automating comprehensive process structures, has sparked remarkable optimism regarding their transformative impact on healthcare practice (Nova, 2023).

But where will developments like those of generative AI models lead? Will generative AI truly take a central role in our healthcare system in the future, enabling the revolution of complex interaction processes and thus realizing necessary quantum leaps in efficiency improvements?

Methodical approach

The objective of the investigation is to examine the potential and central challenges of generative AI solutions in healthcare using the following research questions:

• Formula 1To what extent can the efficiency be increased by implementing generative AI solutions

and the effectiveness of the supply service can be improved?

• F2What adjustments in the digital infrastructure are necessary to realize the potential

to leverage generative AI solutions?

To answer the research questions, ten experts were interviewed using a standardized questionnaire. The qualitative content analysis was evaluated according to Mayring's (2015) procedure. The experts' answers were inductively assigned to categories and form the central findings of the work, building upon the literature review conducted beforehand.

From Documentation to Diagnosis: Application Fields of Generative AI Models in Healthcare

According to the interviewed experts, the use cases for generative AI solutions are diverse. However, the primary focus of language model solutions addresses improved human-machine interaction by enabling more natural and efficient communication and decision-making pathways through speech recognition systems and chatbots. Consequently, it can be expected that generative AI models will be introduced into the healthcare market in the future in the form of efficiency-boosting service solutions, thereby rationalizing central administration processes as well as fundamentally restructuring existing care models.

Generative AI models are primarily intended to function as assistance systems, supporting both physicians and patients in their daily lives. Currently, generative AI solutions can generate the greatest benefit in the area of clinical administration processes due to lower regulatory hurdles. By automatically capturing, structuring, and storing relevant content from ongoing doctor-patient interactions, it becomes possible to shift the focus of care to patients and automate the documentation burden for physicians and nursing staff. This not only makes processes more efficient but also more effective and people-centered. Furthermore, generative AI applications will also find their way into highly regulated application areas. Consequently, it is expected that physicians will increasingly be supported by generative AI solutions in the form of chatbots for the early detection of multimorbid diseases and the derivation of individual action options. In contrast, in the unregulated sports and wellness market, generative AI solutions gain particular importance when language models are implemented in wearable technologies. Thus, in addition to permanent and location-independent vital data collection, they enable users to engage in an interactive dialogue for the evaluation and interpretation of sensitive vital data. Consequently, the niche movement „Quantified-Self“ becomes suitable for the masses through the user-friendly evaluation of generated raw data using chatbots. Therefore, the increasingly changing understanding of the patient's role must be incorporated into the development of future care pathways, given the availability and resulting potential of digital and intelligent assistance systems.

Why we are failing to implement innovative AI solutions and what adjustments are needed for this

Despite the enormous potential of generative AI solutions in healthcare, their widespread implementation has so far been lacking. According to the experts surveyed, the reasons for this are multifaceted. Central to the development and introduction of corresponding solutions is the modernization of the existing digital infrastructure, particularly through improved interoperability of heterogeneous system landscapes. Accordingly, compatible data architectures must be created within the framework of platform solutions, so that interdisciplinary and trans-sectoral interaction results in an open data exchange between doctors, patients, researchers, and developers. Data access becomes particularly relevant when heterogeneous innovation clusters are to develop competitive, European models and make them available for application in highly sensitive healthcare areas. However, in order to drive innovation-promoting market dynamics in Europe, market entry for start-ups must be facilitated – for example, through access to relevant data and the promotion of pilot projects within the framework of cooperation opportunities. Adjustments to legal frameworks are also central to the implementation process of generative AI solutions. While the safety precautions for AI applications in health-related fields of application must be urgently observed, future regulations must pay close attention to the innovation-hindering nature of regulations when designing them. This increasingly creates the danger that orthodox hierarchies, rigid performance authorities, and a rigid market regulatory framework will significantly hinder relevant innovation drivers from entering the market. However, in order to meet the challenges in healthcare with innovative answers, a market regulatory framework that allows for them is needed.

In conclusion, it must be stated that the results from the expert discussions confirm the previously identified potentials and existing obstacles from the literature review. The potential of generative AI solutions is significant. However, to fully exploit this potential, a rethinking of incompatible system structures is necessary, which have so far prevented any development of central innovation potentials.

Bibliography:

Bejjani, M., Göcke, L., and Menter, M. (2023) ‘Digital entrepreneurial ecosystems: A systematic literature review’, Technological Forecasting and Social Change, 189. Available at: https://doi.org/10.1016/j.techfore.2023.122372.

Chakraborty, I., Edirippulige, S. and Ilavarasan, P.V. (2023) ‘What's Next in Health Technology Startups? Insights and Best Practices’, Digital Health, 9. Available at: https://doi.org/10.1177/20552076231178435.

Chui, M. and others. (2023) The Economic Potential of Generative AI: The Next Productivity Frontier.

Daumann, F. and others. (2020FOR-MED Journal for Healthcare Management Imprint Publisher: Reviewer Board: Editor-in-Chief: Copyright, FOR-MED Magazine [Preprint].

Haring, R. (2022) Health Sciences. Available at: www.springerreference.de.

Henke, K.-D. and others. (2015) Future Model for an Efficient Healthcare System in Germany Evidence-Based Medicine View project Polypill in Secondary Prevention of Myocardial Infarction View project Future Model for an Efficient Healthcare System in Germany. Available at: https://www.researchgate.net/publication/267259465.

Hiltawsky, K. and Boll, S. (2022) AI Business Models for Healthcare.

Löser, A. and others. (2023) Developing and applying large language models. Approaches for sovereign action.

Ortloff, JH., Schmitz, D. (2024). Multimorbidity and Chronic Diseases. In: Schmitz, D., Fiedler, M., Becker, H., Hatebur, S., Ortloff, JH. (eds) Chronic Care – Science and Practice. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-68415-3_9

Nova, K. (2023) Generative AI in Healthcare: Advancements in Electronic Health Records, Facilitating Medical Languages, and Personalized Patient Care.

Pfannstiel, M., Kassel, K., and Rasche, C. (2020) Innovations and Innovation Management in Healthcare.

Pfannstiel, M.A., Jaeckel Roger, and Da-Cruz, P. (2019) Market Access in Healthcare.

Porsche Consulting (2021) Lessons learned with mySugr: Launching a successful digital health startup in Europe.

Rasche, C., Reinecke, A.A. and Margaria, T. (2022) ‘Artificial Intelligence in Healthcare as a Core Competence? Status Quo, Development Trends, and Disruptive Potential’, in Artificial Intelligence in Healthcare. Springer Fachmedien Wiesbaden, pp. 49-79. Available at: https://doi.org/10.1007/978-3-658-33597-7_2.

Second, G. and others. (2019) ‘Knowledge transfer in open innovation: A classification framework for healthcare ecosystems’, Business Process Management Journal, 25(1), pp. 144–163. Available at: https://doi.org/10.1108/BPMJ-06-2017-0173.

Spitzer, S.G. (2020) ‘Bridging sectoral barriers with integrative care, Internal Medicine Doctor. Springer Medicine, pp. 903–911. Available at: https://doi.org/10.1007/s00108-020-00843-8.

Stachwitz, P. and Debatin, J.F. (2023) ‘Digitalization in Healthcare: Today and in the Future, Federal Health Gazette – Health Research – Health Protection. Springer Science and Business Media Deutschland GmbH, pp. 105–113. Available at: https://doi.org/10.1007/s00103-022-03642-8.

Stummeyer, C. and Behm, M.E. (2023) Platform Economy in Healthcare: Health as a Service - Digital Business Models for Better Treatment Quality and Patient Experience.

TechNavio (2023) Digital Health Market Analysis North America, Europe, Asia, Rest of World (ROW) – US, China, UK, Germany, France – Size and Forecast 2023-2027.

Viola Henke and others. (2022) Digital strategy in hospitals, Digital strategy in hospitals. Springer Fachmedien Wiesbaden. Available at: https://doi.org/10.1007/978-3-658-36226-3.

WEF (2024) Patient-First Health with Generative AI: Reshaping the Care Experience.

Zhang, P. and Kamel Boulos, M.N. (2023) ‘Generative AI in Medicine and Healthcare: Promises, Opportunities, and Challenges’, Future Internet. Multidisciplinary Digital Publishing Institute (MDPI). Available at: https://doi.org/10.3390/fi15090286.

Zhu, H. and others. (2019) ‘Smart Healthcare in the Era of the Internet of Things, IEEE Consumer Electronics Magazine, 8(5), pp. 26–30. Available at: https://doi.org/10.1109/MCE.2019.2923929.

Kaiser, Robert (2014): Qualitative Expert Interviews: Conceptual Foundations and Practical Implementation. Springer-Verlag.