From RECIST to AI: longitudinal imaging models and the risk prediction vs kinetic diagnosis distinction

Current AI imaging models for immunotherapy response are risk-stratification tools, not diagnostic classifiers of hyperprogressive disease (HPD). HPD diagnosis requires at least three imaging time points to capture growth acceleration relative to baseline trajectory. Most eviden…

• The review traces evolution from RECIST 1.1 through immune-adapted criteria (iRECIST) to functional/molecular imaging and AI-driven longitudinal models.
• Diagnosis of hyperprogressive disease is kinetic, requiring pre-baseline, baseline, and on-treatment scans; no single-time-point AI model can reconstruct that trajectory.
• Persistent obstacles to clinical adoption include heterogeneous acquisition protocols, missing pre-baseline scans, weak external validation, and limited biological interpretability.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.