Stakeholder Perspectives on Trustworthy AI for Parkinson Disease Management Using a Cocreation Approach: Qualitative...

Area of Science:

• Neuroscience
• Medical Informatics
• Bioethics

Background:

• Parkinson disease (PD) is the fastest-growing neurodegenerative disorder globally, presenting significant healthcare challenges.
• Artificial intelligence (AI) and wearable sensors offer potential for PD diagnosis, monitoring, and prognosis.
• Ethical AI adoption necessitates principles like user trust, transparency, fairness, and human oversight.

Purpose of the Study:

• To explore diverse stakeholder perspectives on AI in PD care.
• To guide the development of ethical AI-driven digital health solutions for PD.
• To emphasize transparency, data security, fairness, bias mitigation, and human oversight in AI applications for PD, as part of the AI-PROGNOSIS project.

Main Methods:

• An exploratory qualitative approach using cocreation workshops with key stakeholders.
• Involved 24 participants: individuals with PD, healthcare professionals, AI experts, and bioethicists.
• Semistructured discussions focused on trust, fairness, explainability, autonomy, and psychological impact of AI in PD care.

Main Results:

• Five main themes emerged: AI trust and security, AI transparency and education, AI bias and fairness, human oversight, and AI's psychological impact.
• Key findings highlighted the need for data safety, system accuracy, clear explanations of AI technologies, and equitable access.
• Emphasis was placed on AI-human collaboration and the essential role of human review in AI processes.

Conclusions:

• Implementing robust security, transparent and explainable AI models, and bias mitigation strategies is crucial.
• Integrating human oversight and considering the psychological impact of AI are essential for effective PD care.
• Actionable guidance is provided for developing trustworthy and effective AI-driven digital solutions for PD diagnosis and management.