Ethical Reflections on Integrating Artificial Intelligence in Care Practices

Abstract

Background

The integration of artificial intelligence (AI) and robotics into disability care presents transformative opportunities while simultaneously raising pressing ethical concerns. Issues related to autonomy, human dignity, and equitable access require careful consideration, particularly as these technologies reshape the dynamics of care delivery and clinical relationships.

Purpose

Drawing on an interdisciplinary approach that synthesizes insights from bioethical literature, illustrative case studies, and expert perspectives from healthcare, law, and technology, this reflection examines the ethical landscape of AI-supported rehabilitation and assistance. Particular attention is given to risks such as algorithmic bias, over-reliance on automation, and the potential erosion of the human dimension in care. A biopsychosocial model serves as a guiding framework to analyze how technological systems intersect with the lived experiences of individuals with disabilities. Ethical tensions emerge around personalized care, transparency in decision-making, and the inclusivity of data and design processes.

Conclusions

The analysis emphasizes the need for governance models that embed ethical safeguards and promote fairness, while also encouraging participatory design involving patients, caregivers, and healthcare professionals. By situating technological developments within broader socio-political and clinical contexts, this reflection identifies pathways toward a more equitable and human-centered integration of AI. Recommendations include investment in inclusive datasets, the development of fairness-aware algorithms, and the establishment of regulatory mechanisms that align innovation with fundamental rights and principles of social justice in healthcare.

Keywords

artificial intelligence disability equity rehabilitation health care

Introduction

The advent of AI and robotics has opened new possibilities for supporting individuals with disabilities, promoting independence, and improving social integration.¹ However, this advancement raises critical bioethical questions that intersect with anthropology, law, and socio-political considerations. Central to this discourse is the biopsychosocial model, which views disability as a dynamic interplay of biological, psychological, and social factors.^2,3 This paper examines the ethical implications of implementing AI in rehabilitation, focusing on autonomy, dignity, equity, and the evolving relationships between patients and healthcare professionals.

To set the stage for current advancements, it is crucial to understand the historical evolution of AI and robotics in healthcare.⁴ Early developments in assistive technologies laid the groundwork for today’s sophisticated systems, which integrate AI and robotics to deliver personalized and adaptive care.^5,6 This evolution has prompted an interdisciplinary effort involving ethicists, engineers, and clinicians to ensure responsible innovation and deployment.⁷

This article intends to offer a comprehensive and critical overview of the ethical implications arising from the integration of artificial intelligence into disability care. By examining key principles such as autonomy, dignity, and equity, it aims to contribute to the ongoing discourse by advancing ethically grounded and human-centered proposals that support an inclusive, transparent, and socially responsible deployment of these technologies within contemporary healthcare systems.

Ethical Foundations

Autonomy and Dignity as Central Ethical Principles is the principle of respecting the autonomy and intrinsic dignity of individuals with disabilities.⁸ AI technologies must empower users, ensuring their participation in decision-making processes. The United Nations Convention on the Rights of Persons with Disabilities (UNCRPD) and guidelines from the European Commission advocate for AI systems that enhance human capabilities without compromising autonomy.^9,10 Challenges such as the “black box” nature of AI systems, algorithmic discrimination, and limited explainability require robust regulatory oversight to preserve patient rights and foster trust in technology.¹¹

Case studies of successful AI implementations highlight both opportunities and challenges in preserving dignity and autonomy.¹² For example, AI-driven communication aids have transformed the lives of individuals with speech impairments, enabling them to express their preferences and participate actively in social and professional contexts.¹³ However, well-documented instances of biased algorithms underscore the need for heightened vigilance. For example, Ferrara (2024)¹⁴ outlines how AI systems used in healthcare, credit, and criminal justice have repeatedly produced discriminatory outcomes due to biased training data or opaque decision processes. Similar risks are evident in AI-driven recruitment systems, where algorithms have been shown to disadvantage candidates based on gender, ethnicity, or personality traits.¹⁵ These cases reveal how structural biases embedded in data and model design can perpetuate or even amplify social inequities if not carefully mitigated.

Technological Integration in Disability Care

AI and robotics have revolutionized rehabilitation, offering tools like robot-assisted gait training and therapeutic devices for motor skill recovery.¹⁶ These systems provide feedback, facilitate repetitive movements, and integrate with home-based care. Socially assistive robots (SARs) support patients with dementia, autism, and motor disabilities, enhancing therapeutic outcomes and social engagement.¹⁷ Despite their promise, these technologies risk creating psychological dependencies and emotional attachments, necessitating a balanced approach that prioritizes human dignity and psychosocial well-being.¹⁸

Global disparities in access to these technologies reveal significant ethical challenges. Marginalized communities often lack resources to implement or benefit from AI and robotics, exacerbating existing inequalities. Strategies such as subsidies, international cooperation, and inclusive design principles are essential to address these inequities.¹⁹

Impact on Healthcare Relationships

The use of AI and robotics transforms the dynamic between patients and healthcare workers (HWs).²⁰ While these technologies can enhance efficiency and diagnostic precision, they may also lead to over-reliance, diminishing the human connection integral to care. The European Parliament’s guidelines stress that AI should complement, not replace, the physician-patient relationship.²¹ By addressing automation bias and fostering collaboration between clinicians and engineers, healthcare systems can mitigate risks and ensure technology serves as an extension of human expertise.²²

Emerging trends, such as brain-computer interfaces and real-time analytics, have the potential to further transform these relationships. These innovations require healthcare professionals to acquire new skills and adapt their practices to integrate AI seamlessly into patient care.^23,24

Disparities of Care

Disabled individuals often face significant disparities in healthcare, including limited access to services, lack of proper accommodations, and biased treatment.²⁵

While AI holds the potential to improve healthcare accessibility and outcomes for disabled populations, it can also exacerbate existing inequities if not designed inclusively.^15,25,26 AI systems, such as diagnostic tools or treatment algorithms, may not always account for the specific needs of disabled individuals, leading to misdiagnoses or inadequate care.

Moreover, as emphasized by Mougin et al. (2022)²⁷ and Wald (2020),²⁸ training datasets frequently lack adequate representation of disabled individuals, leading to models that fail to accommodate their specific needs. For instance, clinical decision-support tools and diagnostic systems may produce inaccurate outcomes for users with rare or complex disabilities due to insufficient or unbalanced training data.^26,28 To address such disparities, Ferrara (2024)¹⁴ recommends integrating fairness-aware learning methods and ensuring that disability-inclusive data is prioritized in the design and validation of AI systems. Only through such deliberate strategies can AI technologies be safely and equitably integrated into healthcare frameworks.²⁷

In addition to well-documented structural disparities, a growing body of evidence highlights how AI systems may introduce or exacerbate systemic biases against individuals with disabilities. These biases often stem from non-representative training datasets, opaque algorithmic processes, and generalized performance metrics that fail to accommodate atypical clinical profiles.

For instance, studies by Li et al. (2024)²⁵ and Ferrara (2024)¹⁴ reveal that AI-based diagnostic tools exhibit reduced accuracy when applied to individuals with complex or rare disabilities, increasing the risk of misdiagnosis or delayed treatment. Chen et al. (2023)¹⁵ further report that predictive models used for triage or risk stratification often underestimate the severity of symptoms in disabled patients, particularly when those symptoms fall outside of the standardized patterns learned during training.

Recent statistics indicate that less than 5% of medical training datasets include annotated data from patients with motor or cognitive impairments (Wald, 2020; Mougin et al., 2022),^27,28 and fewer than 1% account for individuals with multiple disabilities as a distinct category. This underrepresentation contributes directly to erroneous or inequitable clinical decisions.

Proposed Solutions:

• Development of inclusive datasets that represent diverse types and severities of disabilities. Pilot projects such as MIMIC-Disability aim to ethically collect and integrate such data.²⁹

• Implementation of fairness-aware learning techniques that embed equity criteria into training objectives, as recommended by Ferrara (2024).¹⁴

• Active participation of disability communities in the design and validation of AI tools to ensure these technologies reflect lived experiences and meet specific clinical needs.

These strategies are essential to prevent AI—originally intended to promote accessibility—from becoming a driver of exclusion and marginalization for disabled individuals.

Stakeholders’ Perspectives

Healthcare professionals, as key stakeholders in the implementation of AI in care settings, express both cautious optimism and important reservations.³⁰ Studies reveal that medical practitioners generally welcome AI tools that assist with triage, diagnosis, or workflow streamlining—particularly in settings characterized by high patient volume and resource constraints. For example, Townsend et al. (2023)^31,32 found that NHS emergency department clinicians perceived the proposed DAISY system (a Diagnostic AI for triage) as a valuable support tool to reduce wait times, standardize initial assessments, and prioritize urgent cases. However, their support was contingent on the system being used as an aid—not a replacement—for human judgment. Concerns included the system’s inability to interpret non-verbal cues, maintain empathetic interactions, and adapt to nuanced clinical contexts, which practitioners consider central to quality care.

Similarly, Moschogianis et al. (2025)³³ highlighted that while primary care staff recognized the potential of AI to reduce administrative burden and improve patient response times during electronic consultations (eVisits), they emphasized the importance of preserving clinical oversight. Staff and patients alike expressed that AI should function as a complement to, rather than a substitute for, direct human engagement. Key conditions for acceptance included transparency, accuracy, and reassurance that AI would not erode the therapeutic relationship.

Further, studies involving medical educators and students in Pakistan³⁴ found a broadly positive attitude towards AI integration, especially its ability to enhance learning and research. Faculty, however, raised concerns about ethical issues such as data privacy and warned against over-reliance on automated systems in clinical judgment. The lack of institutional support and training were highlighted as significant implementation barriers, especially in low-resource settings.

In a complementary study conducted in India (Jackson et al., 2024),³⁵ a majority of medical students perceived AI as a useful assistive tool capable of reducing medical errors and enhancing diagnostic accuracy. Nevertheless, over two-thirds feared it would erode the humanistic dimension of medicine, disrupt patient-physician trust, and raise ethical concerns such as confidentiality breaches. These findings underscore the dual outlook of current and future healthcare professionals: while they acknowledge AI’s promise, they advocate for its careful and ethically guided implementation, preferably accompanied by structured educational support.

In the authors’ view, the perceptions of both current and future healthcare professionals reveal that the integration of AI technologies into clinical practice remains in a nascent phase. This evolving landscape suggests that the attitudes of healthcare stakeholders are still being shaped by preliminary experiences and hypothetical scenarios rather than by systematic, long-term exposure. It would therefore be valuable to investigate how these attitudes might evolve over time, particularly in response to two pivotal developments: a tangible erosion of human clinical competencies due to automation, or a shift in the perceived balance between AI’s risks and its benefits. Whether the opinions of healthcare professionals will carry sufficient weight to influence policymaking—potentially leading to restrictions on the deployment of AI, or, though less likely, to its limitation—remains an open and pressing question.

From the patients’ perspective, the incorporation of AI into rehabilitation practices offers both opportunities and challenges. Many patients recognize the potential of AI-supported interventions to enhance accessibility, particularly in home-based settings, where travel or mobility limitations might otherwise impede participation. For example, vision-based sensors and AI-driven feedback systems can enable patients to perform physical exercises independently while receiving real-time assessments and guidance, addressing the shortage of healthcare professionals and promoting adherence to therapy regimens.³⁶ During the COVID-19 pandemic, patients reported that tele-rehabilitation not only ensured continuity of care but also fostered a renewed sense of engagement and self-management, particularly when platforms were perceived as intuitive and supportive of individualized therapeutic goals.³⁷ Similar outcomes may also be hypothesized for remotely operated artificial intelligence systems, provided they are designed with comparable levels of usability and personalization.

However, concerns remain prevalent. Some individuals report difficulties engaging with digital platforms due to low technological literacy, especially among older adults.^36,38 Despite recognizing the potential of AI to enhance healthcare efficiency, many patients—particularly older adults—express concerns about its limited capacity to provide empathetic, personalized care. The perceived absence of human warmth and emotional understanding in AI-driven interactions has led to reported feelings of discomfort and disconnection, with a strong preference for maintaining human involvement in care delivery.³⁸ Furthermore, ethical apprehensions related to informed consent, privacy, and data security are especially pronounced among older patients, who often exhibit limited digital literacy and a trust deficit in AI systems.^38,39 As such, while AI-supported rehabilitation is generally valued for its flexibility and personalization, patients consistently advocate for these technologies to serve as supportive tools that respect autonomy and reinforce—rather than replace—the human dimension of care.^38–40

At present, authors have also cautioned against the risk that AI systems, if not carefully implemented, may contribute to detrimental forms of social isolation, particularly in vulnerable populations where sustained human interaction remains essential to ensuring psychosocial wellbeing and therapeutic engagement.^41,42

Challenges and Recommendations

To effectively guide the ethical integration of AI in disability care, it is helpful to think in terms of short-term and long-term priorities. In the short term, interdisciplinary training programs should be developed to help healthcare workers build competence in AI literacy, ethical awareness, and digital accessibility. This foundational knowledge can foster more informed, confident use of technology in clinical settings. At the same time, developers must focus on creating explainable and transparent AI systems that allow users to understand how decisions are made—building trust and ensuring accountability.^27,28

Equally important is the adoption of a human-centered design approach, which involves individuals with disabilities directly in the development of AI tools.⁴³ By drawing on their lived experiences, these systems can better reflect real needs and psychosocial contexts. Public awareness campaigns should also be launched to promote digital inclusion and help patients and caregivers feel more comfortable and informed about AI-supported rehabilitation. Complementing these efforts, pilot projects—such as MIMIC-Disability—should be initiated to improve the representativeness of datasets and ensure diverse types of disabilities are ethically and accurately included in system training.²⁹

Over the longer term, more structural measures must be pursued. National and international regulatory bodies should establish clear ethical guidelines and legal frameworks to govern how AI is implemented in care environments, ensuring compliance with human rights principles. Healthcare policies should be updated to embed fairness, equity, and autonomy as guiding values in AI integration. Furthermore, public and private investment in infrastructure is critical to bridging the digital divide and expanding access to assistive technologies in underserved communities. Finally, sustainability must be built into the process through long-term monitoring systems that assess the clinical and ethical impact of AI tools, making sure they continue to serve patients effectively and justly.^12,14,29

Conclusion

The integration of AI and robotics into disability care requires an ethically grounded, phased approach that balances technological innovation with human values. Prioritizing short-term actions—such as workforce training, explainability, and inclusive design—can lay the groundwork for safe and equitable deployment. Over time, structural reforms including regulatory oversight and policy transformation will be essential to sustain ethical integrity and social inclusion.⁴⁴

To ensure that AI serves as a tool for empowerment rather than exclusion, stakeholders must commit to a biopsychosocial and human-centered model. We call on policymakers to enact legislation that embeds ethical safeguards and accessibility standards into AI regulation. Researchers should be supported through targeted funding initiatives to develop inclusive datasets and fairness-aware algorithms. At the same time, AI developers must engage directly with disability communities to co-design tools that address real-world needs and reflect lived experiences.

Only through sustained collaboration among these actors—combined with robust governance mechanisms—can we ensure that technological progress results in tangible, equitable improvements in the lives of individuals with disabilities.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Giulia Ricchezze

Author Biographies

Giulia Ricchezze is a MS who is currently pursuing a PhD.

Luca Tomassini is an MD who is currently pursuing a PhD.

Mauro Bacci is a former Full Professor of Forensic Medicine.

References

United Nations . Building an accessible future for all: AI and the inclusion of persons with disabilities. United Nations Western Europe, 2024. https://unric.org/en/building-an-accessible-future-for-all-ai-and-the-inclusion-of-persons-with-disabilities/

De Micco

Scendoni

. Three different currents of thought to conceive justice: legal, and medical ethics reflections. Philosophie 2024; 9(3): 61.

Scendoni

Fedeli

Cingolani

. The state of play on COVID-19 vaccination in pregnant and breastfeeding women: recommendations, legal protection, ethical issues and controversies in Italy. Healthcare 2023; 11(3): 328.

Dwivedi

Sharma

Rana

, et al. Evolution of artificial intelligence research in technological forecasting and social change: research topics, trends, and future directions. Technol Forecast Soc Change 2023; 192: 122579.

. Rethinking artificial intelligence from the perspective of interdisciplinary knowledge production. AI Soc 2024; 39(6): 3059–3060.

Alum

Ugwu

OPC

. Artificial intelligence in personalized medicine: transforming diagnosis and treatment. Discov Appl Sci 2025; 7(3): 193.

Patel

Esper

, et al. The crucial role of interdisciplinary conferences in advancing explainable AI in healthcare. BioMedInformatics 2024; 4(2): 1363–1383.

UN General Assembly . Convention on the Rights of Persons with Disabilities. Article 3 – General Principles . January 2007. https://www.un.org/development/desa/disabilities/convention-on-the-rights-of-persons-with-disabilities/article-3-general-principles.html

UN General Assembly . Convention on the Rights of Persons with Disabilities. January 2007; A/RES/61/106,: 24. https://social.desa.un.org/issues/disability/crpd/convention-on-the-rights-of-persons-with-disabilities-crpd.

10.

European Commission . United Nations convention on the Rights of Persons with Disabilities. 2025. https://commission.europa.eu/strategy-and-policy/policies/justice-and-fundamental-rights/disability/united-nations-convention-rights-persons-disabilities_en

11.

Gallistl

Banday

MUL

Berridge

, et al. Addressing the black box of AI-A model and research agenda on the Co-constitution of aging and artificial intelligence. Gerontol 2024; 64(6): gnae039.

12.

Adegbesan

Akingbola

Ojo

, et al. Ethical challenges in the integration of artificial intelligence in palliative care. J Med Surg Public Health 2024; 4: 100158.

13.

Griffiths

Broomfield

Hrastelj

, et al. AI, communication aids and the challenge of authentic authorship – whose line is it anyway. J Enabling Technol 2025; 19: 102–112.

14.

Ferrara

. Fairness and bias in artificial intelligence: a brief survey of sources, impacts, and mitigation strategies. Sci 2024; 6(1): 3.

15.

Chen

Clayton

Novak

, et al. Human-centered design to address biases in artificial intelligence. J Med Internet Res 2023; 25: e43251.

16.

Alshami

Nashwan

AlDardour

, et al. Artificial intelligence in rehabilitation: a narrative review on advancing patient care. Rehabilitacion 2025; 59(2): 100911.

17.

Kriegel

Grabner

Tuttle-Weidinger

, et al. Socially assistive robots (SAR) in In-Patient care for the elderly. Stud Health Technol Inf 2019; 260: 178–185.

18.

Manzoor

Vimarlund

. Digital technologies for social inclusion of individuals with disabilities. Health Technol 2018; 8(5): 377–390.

19.

Byrne

. How artificial intelligence is helping to identify global inequalities. Nature, 2024. https://www.nature.com/articles/d41586-024-01534-2

20.

Koski

Murphy

. AI in healthcare. Stud Health Technol Inf 2021; 284: 295–299.

21.

European Parliament . Directorate general for parliamentary research services. Artificial intelligence in healthcare: applications, risks, and ethical and societal impacts. Publications Office, 2022.

22.

Qudah

Luetsch

. The influence of mobile health applications on patient - healthcare provider relationships: a systematic, narrative review. Patient Educ Counsel 2019; 102(6): 1080–1089.

23.

Shih

Krusienski

Wolpaw

. Brain-computer interfaces in medicine. Mayo Clin Proc 2012; 87(3): 268–279.

24.

Padulo

. Sport and health science: interdisciplinary approaches to modern challenges. Br Med Bull 2025; 155(1): ldaf007.

25.

Moon

Shankar

, et al. Health inequities, bias, and artificial intelligence. Tech Vasc Interv Rad 2024; 27(3). https://www.techvir.com/article/S1089-2516(24)00046-5/fulltext.

26.

Charow

Jeyakumar

Younus

, et al. Artificial intelligence education programs for health care professionals: scoping review. JMIR Med Educ 2021; 7(4): e31043.

27.

Mougin

Hollis

Soualmia

. Inclusive digital health. Yearb Med Inform 2022; 31(1): 2–6.

28.

Wald

. AI data-driven personalisation and disability inclusion. Front Artif Intell 2020; 3: 571955.

29.

Leu

Gerfin

Spycher

. The validity of the MIMIC (Multiple Indicators/MultIple Causes) health index--some empirical evidence. Dev Health Econ Publ Pol 1992; 1: 109–142.

30.

Castagno

Khalifa

. Perceptions of artificial intelligence among healthcare staff: a qualitative survey study. Front Artif Intell 2020; 3: 578983.

31.

Townsend

Plant

Hodge

, et al. Medical practitioner perspectives on AI in emergency triage. Front Digit Health 2023; 5: 1297073.

32.

Wang

Fei

Wei

, et al. Knowledge and attitudes toward artificial intelligence in nursing among various categories of professionals in China: a cross-sectional study. Front Public Health 2024; 12: 1433252.

33.

Moschogianis

Darley

Coulson

, et al. Seven opportunities for artificial intelligence in primary care electronic visits: qualitative study of staff and patient views. Ann Fam Med 2025; 23(3): 214–222.

34.

Naseer

Saeed

Afzal

, et al. Navigating the integration of artificial intelligence in the medical education curriculum: a mixed-methods study exploring the perspectives of medical students and faculty in Pakistan. BMC Med Educ 2025; 25: 273.

35.

Jackson

Ponath Sukumaran

Babu

, et al. Artificial intelligence in medical education - perception among medical students. BMC Med Educ 2024; 24(1): 804.

36.

Sardari

Sharifzadeh

Daneshkhah

, et al. Artificial intelligence for skeleton-based physical rehabilitation action evaluation: a systematic review. Comput Biol Med 2023; 158: 106835.

37.

Ding

Shang

Xie

, et al. Artificial intelligence in the COVID-19 pandemic: balancing benefits and ethical challenges in China’s response. Humanit Soc Sci Commun 2025; 12(1): 245–319.

38.

Rony

MKK

Deb

Khatun

, et al. Attitudes of older patients toward artificial intelligence in decision-making in healthcare. J Med Surg Public Health 2025; 6: 100193.

39.

Cecconi

Adams

Cardone

, et al. Generational differences in healthcare: the role of technology in the path forward. Front Public Health 2025; 13: 1546317.

40.

Reddy

. Generative AI in healthcare: an implementation science informed translational path on application, integration and governance. Implement Sci 2024; 19(1): 27.

41.

Gurevich

El Hassan

El Morr

. Equity within AI systems: what can health leaders expect? Health Manag Forum 2023; 36(2): 119–124.

42.

Cheong

. Transparency and accountability in AI systems: safeguarding wellbeing in the age of algorithmic decision-making. Soc Sci Res Netw 2024; 6. https://papers.ssrn.com/abstract=4961260.

43.

Allen

Yawson

Haggenmüller

, et al. Human-centered AI as a framework guiding the development of image-based diagnostic tools in oncology: a systematic review. ESMO Real World Data and Digital Oncology 2024; 6: 100077.

44.

De Micco

Tambone

Frati

, et al. Disability 4.0: bioethical considerations on the use of embodied artificial intelligence. Front Med 2024; 11: 1437280.