Validation of an AI-based end-to-end model for prostate pathology using long-term archived routine samples

2026-05-04

AI Summary

This study addresses the limited generalizability of AI models across long-term, multicenter, and heterogeneous prostate histopathology datasets by proposing GleasonAI—an end-to-end multiple instance learning model incorporating attention mechanisms for ISUP grading. Trained and independently validated on 10,366 archived biopsy cores collected between 1998 and 2015, GleasonAI achieves a quadratic weighted kappa of 0.86, matching the performance of experienced pathologists while demonstrating consistent stability across temporal and geographic dimensions. As the first AI system validated on large-scale, real-world data spanning nearly two decades, GleasonAI not only surpasses existing foundation models in performance but also shows significant association between its grading outputs and prostate cancer–specific mortality, underscoring its clear prognostic value.

Abstract

Artificial intelligence (AI) is becoming a clinical tool for prostate pathology, but generalization across variations in sample preparation and preservation over prolonged time periods remains poorly understood. We evaluated GleasonAI, an end-to-end attention-based multiple instance learning model, on an independent validation cohort comprising 10,366 biopsy cores from 1,028 patients across 14 Swedish regions, using archival diagnostic specimens from the ProMort cohorts collected between 1998-2015. The model achieved an overall quadratic-weighted kappa of 0.86 for core-level ISUP grading, comparable to several experienced pathologists and consistent across geographic regions. Notably, performance remained stable across the 17-year collection period, demonstrating robustness to time-related variation in archival material, a property not consistently observed with foundation model-based approaches, with exploratory analysis demonstrating a significant prognostic gradient across AI-assigned grade groups for prostate cancer-specific mortality. These findings support the generalizability of the AI grading model and demonstrate the potential of pathology archives as a large-scale resource for AI development, validation, and retrospective prognostic research.

Problem

• AI generalization
• prostate pathology
• archival samples
• time-related variation
• sample preservation

Innovation

• end-to-end AI model
• multiple instance learning
• archival robustness
• prostate cancer grading
• long-term generalizability