Artificial Intelligence-Enhanced Pathology

Researchers develop artificial intelligence (AI) algorithms via machine learning analysis of scanned biopsy slide images. They can identify patterns associated with prostate cancer in its various
microscopic patterns. These algorithms are trained progressively based on classifying cases and controls regarding their clinical outcomes. Some companies are currently marketing AI pathology tests for clinical decision- making.

Recent studies have led to the development and validation of the first prognostic AI pathology tool in the published National Prostate Cancer Network guidelines,1 and the first validated predictive tool for identifying which patients are more likely to benefit from adding hormonal therapy to radiotherapy.1-3 In a subsequent study that evaluated algorithmic fairness between races, these PAI approaches performed comparably in Black patients compared to White patients.2 In another study of patients on active surveillance, the AI pathology algorithm was more accurate than the urological pathologist in predicting tumor upgrading.4 Thus, AI pathology offers the promise of accuracy comparable or superior to standard clinical pathology, possibly with less variability.
1. Esteva A et al Npj digital Medicine 2022; 5:71
2. Roach M et al J Clin Oncol 2022; 40:108 3. Tward JD et al JCO Precis Oncol 2024;8
doi:10.1200/po.24.00145
4. Ding CC et al J Natl Cancer Inst 2024;
116:1683-1686