Translational Gaps in Immuno-AI: From algorithmic accuracy to clinical trust.

Oisakede, Emmanuel O; Ayo Daniel, Raphael Igbarumah; Olawuyi, Olabanke Florence; Alabi, John Oluwatosin; Analikwu, Claret Chinenyenwa; Olawade, David

Translational Gaps in Immuno-AI: From algorithmic accuracy to clinical trust.

Oisakede, Emmanuel O, Ayo Daniel, Raphael Igbarumah, Olawuyi, Olabanke Florence, Alabi, John Oluwatosin, Analikwu, Claret Chinenyenwa and Olawade, David ORCID: https://orcid.org/0000-0003-0188-9836 (2026) Translational Gaps in Immuno-AI: From algorithmic accuracy to clinical trust. Human immunology, 87 (8). p. 111774.

[thumbnail of 1-s2.0-S0198885926001205-main.pdf]

Preview

Text
1-s2.0-S0198885926001205-main.pdf - Published Version
Available under License Creative Commons Attribution.
| Preview

Official URL: https://doi.org/10.1016/j.humimm.2026.111774

Abstract

Artificial intelligence has shown remarkable promise in predicting patient responses to immune checkpoint inhibitors across cancers. However, despite high statistical performance, clinical translation remains minimal. This disconnect between algorithmic accuracy and clinical adoption, termed the translational gap, reflects unresolved challenges in validation, interpretability, and regulatory integration. This review critically examines key barriers preventing translation of Immuno-AI systems from research prototypes to clinically trusted decision-support tools. It analyzes methodological, regulatory, ethical, and infrastructural factors limiting implementation and proposes strategies for developing clinically trustworthy AI in immuno-oncology. A structured literature search was conducted across PubMed, Embase, Scopus, and Web of Science for studies published 2018-2025 reporting AI or machine learning models predicting ICI response or toxicity in human cohorts. Narrative synthesis was applied, focusing on translational bottlenecks. Three dominant factors underpin the translational gap: (1) insufficient external and prospective validation, leading to overestimation of model performance; (2) limited interpretability and absence of explainable frameworks suitable for clinical use; and (3) regulatory and infrastructural immaturity, including lack of harmonised standards for adaptive AI systems. These limitations contribute to absence of clinician confidence and hinder regulatory approval. Bridging the translational gap in Immuno-AI requires a shift from model-centric optimisation to system-level accountability. Clinically trustworthy AI must be validated across institutions, designed for interpretability, and governed by transparent, ethical frameworks. Collaborative efforts among researchers, clinicians, and regulators are essential to ensure future Immuno-AI systems achieve algorithmic excellence, clinical credibility, and social legitimacy. [Abstract copyright: Copyright © 2026 The Author(s). Published by Elsevier Inc. All rights reserved.]