Artificial Intelligence (AI) in Performing Landscape Review and Linguistic Analysis for Curative Intent in Prostate Cancer (PC)

OBJECTIVES: Emerging PC treatments have increased the possibility of achieving cure, even in high-risk disease. Stakeholders may define and perceive cure differently. We describe a landscape review and linguistic methodology investigating the concept of cure in PC.

METHODS: The methodology involved subject-matter experts (SMEs) and AI-powered tools: 1) SMEs selected cure-related keywords; 2) SMEs used Elicit, a semantic search tool, to identify contextual terms frequently used with keywords; 3) Keyword search extracted hits from 4 stakeholder platforms: published literature in MEDLINE (5 years), HCPs in Sermo (2 years), general public in Twitter/forums (27 months), and policymakers in Overton (5 years). NetBase Quid, a natural language processing tool, was used to perform linguistic analysis of hits, metadata, and sentiment about cure; 4) SMEs selected data subsets for each keyword and reviewed them for insights. Data were summarized quantitatively and qualitatively.

RESULTS: Estimated numbers of hits that 7 identified keywords returned across platforms were 12,429 (Cure), 6063 (Survivor), 1904 (Remission), 1179 (Survivorship), 432 (Curative Intent), 381 (No Evidence of Disease), 83 (Complete Remission). Most common keywords were Cure among general public (11,815 hits) and HCPs (224), Survivorship in literature (378), and Survivor among policymakers (378). All stakeholders discussed Cure primarily in early-stage PC. Insights from all 7 subsets showed stakeholders utilized various terms to describe the concept of cure: disease progression measurements (eg, prostate-specific antigen) were used along with Cure in literature, Cure and Survivor among general public, Cure Rates among HCPs, and Potential Cure and Survivor/Survivorship among policymakers when discussing curative-intent treatment. In the Cure-related subset, SMEs reviewed 2452 (general public), 232 (literature), 206 (HCPs), and 153 (policymakers) hits. Cure, Curative Intent, Survivorship, Remission, and Survivor were associated with positive sentiments.

CONCLUSIONS: Combined AI-assisted and human-led large-scale landscape review revealed different linguistic and contextual preferences across stakeholders in communicating about cure in early-stage PC.