%0 Journal Article
%T From Observable Behavior to Inferred Suffering: A Conceptual Framework for Defining the Limits of Artificial Intelligence in Symptom Assessment among Nonverbal Adults Receiving Palliative Care
%A Giulia Romano
%A Francesco Esposito
%J Journal of Integrative Nursing and Palliative Care
%@ 3006-5550
%D 2026
%V 7
%N 2
%R 10.51847/xMuOMrgYIE
%P 146-156
%X To define the epistemic limits of artificial intelligence (AI) when symptom-related information from nonverbal adults receiving palliative care is transformed from observable behavior into claims about subjective suffering. Patients who cannot reliably communicate may express distress through facial activity, movement, vocalization, physiological change and contextual alterations recognized by caregivers or clinicians. Structured observational instruments and emerging AI systems can improve detection, consistency and information synthesis. Yet observed signals, proxy interpretations, algorithmic classifications and lived suffering are not interchangeable forms of knowledge. Original non-empirical conceptual boundary analysis informed by critically selected literature on nonverbal symptom assessment, observational pain measures, proxy reporting, machine-learning inference, clinical decision support, nursing judgment and responsible healthcare AI. The analysis distinguishes five epistemically different positions: observable evidence, human or proxy interpretation, model classification, probabilistic clinical inference and unobserved subjective suffering. AI may strengthen detection and support clinical reasoning when its inputs, target construct, validation population and uncertainty are explicit. Its output does not independently establish the content, intensity or meaning of an individual patient's subjective suffering. Human interpretation is also fallible and should not be treated as an infallible comparator. Responsible use of AI in nonverbal palliative care requires calibrated claims rather than either technological overconfidence or categorical rejection. The strength of any statement about suffering should remain proportionate to the evidence source, model target, validation context and unresolved uncertainty, while nursing accountability and relational interpretation remain clinically consequential.
%U https://journalinpc.com/article/from-observable-behavior-to-inferred-suffering-a-conceptual-framework-for-defining-the-limits-of-ar-zd3dja9169wzy7q