Article Impact Level: HIGH Data Quality: STRONG Summary of Journal of Geriatric Psychiatry https://doi.org/10.1016/j.jagp.2025.12.012 Dr. Helen H. Kyomen et al.
Points
- The Humane Intelligence framework establishes a patient centered and ethically attuned relational model for designing and monitoring artificial intelligence tools specifically within the specialized field of older adult mental health care.
- Clinical teams can utilize the Moral Grid Operational Index to link abstract ethical pillars to observable point of care behaviors that ensure transparency and reciprocity during high stakes medical decision making.
- This framework explicitly prohibits fully automated clinical actions by requiring human in the loop safeguards and clinical grade standards to protect vulnerable patients from potential algorithmic bias or documented medical harms.
- Performance evaluations within this model prioritize geriatric specific outcomes such as patient function and caregiver burden while maintaining strict alignment with current global regulatory standards and food and drug administration guidelines.
- Implementing these testable patient centered routines allows healthcare systems to procure and monitor artificial intelligence technologies that effectively augment clinical wisdom without displacing the essential human elements of geriatric psychiatric care.
Summary
This article proposes a novel “Humane Intelligence” (HI) framework designed to address the unique ethical and clinical challenges of integrating artificial intelligence into geriatric psychiatry. While current AI governance often prioritizes algorithmic infrastructure, the HI framework shifts the focus toward the lived experiences of older adults and their caregivers. It is structured around four foundational pillars: Relational Intelligence, Transparency with Care, Reciprocity and Consent, and Ethical Governance in Strategic Regions. By applying these principles from point-of-care interactions to system-level policies, the framework seeks to ensure that AI applications augment rather than displace essential human-centered care.
Implementation is facilitated through the Moral Grid Operational Index, which operationalizes the four pillars into observable point-of-care behaviors and verifiable evidence. The framework utilizes a structured clinical sequence to signal potential problems, initiate interventions, and verify subsequent benefits or harms. To maintain safety, the guidelines establish a firm boundary against fully automated clinical actions, recommending “human-in-the-loop” safeguards and clinical-grade standards for patient-facing programs. Assessment metrics are specifically tailored to geriatric outcomes, prioritizing patient function, psychological distress, caregiver burden, and health equity over purely technical performance data.
The blueprint aligns with contemporary global standards, including the 2025 JAMA Summit on AI priorities and World Health Organization guidance for large multimodal models. It further incorporates U.S. Food and Drug Administration recommendations for change control plans and ONC frameworks for decision-support interventions. By translating abstract ethical principles into testable patient-centered routines, the HI framework provides a scalable methodology for healthcare systems to evaluate, procure, and monitor AI tools. This approach ensures that technological advancements in older-adult care remain ethically attuned, clinically grounded, and transparently governed.
Link to the article: https://www.ajgponline.org/article/S1064-7481(25)00567-6/abstract
References
Kyomen, H. H. (2025). Humane intelligence in geropsychiatric care: Relational artificial intelligence, clinical wisdom, and the moral grid operational index. The American Journal of Geriatric Psychiatry, S1064748125005676. https://doi.org/10.1016/j.jagp.2025.12.012
