BACKGROUND:Denmark's publicly funded routine HPV vaccination program has included boys born from 2005 onward, leaving earlier birth cohorts of young men potentially unprotected. METHODS:A published deterministic dynamic transmission metapopulation model was adapted to evaluate the impacts of a 3-year male catch-up vaccination program on the cases, deaths, and costs of HPV-associated diseases in Denmark over a 100-year time horizon. Routine gender-neutral HPV vaccination of adolescents with a nonavalent vaccine was modeled with and without a male catch-up program, at 4 catch-up vaccination coverage rates (VCRs) from 40% to 70%. RESULTS:Adding a temporary catch-up program for men born in 1997-2005 was projected to avert 253 HPV-associated cancer cases and 89 deaths at a VCR of 40%. Increasing coverage to 70% was estimated to avert 359 cases and 128 deaths. Catch-up vaccination may be considered cost-effective at all modeled VCRs, with incremental cost-effectiveness ratios of €35,584-35,755 per quality-adjusted life year compared to routine adolescent vaccination alone. CONCLUSIONS:Expanding Denmark's male catch-up HPV vaccination program to include all men born in 1997-2005 would reduce the burden of HPV-associated cancers and diseases and may represent a cost-effective public health strategy.
Background & Aims: Intra- and inter-reader variability complicates liver biopsy scoring for metabolic dysfunction-associated steatohepatitis (MASH). AI-powered digital pathology (AI-DP) is emerging as a tool to assist pathologists, with the promise of improving pathologist concordance. This study evaluates how second harmonic generation (SHG)-based AI-DP influences pathologist decision-making for fibrosis scoring and provides guidance for integrating SHG-based AI assistance into pathology workflows. Methods: Four pathologists reviewed 120 MASH cases with and without the assistance tool. We assessed changes in concordance, scoring, and pathologist perceptions via surveys and detailed case reviews. A qualitative analysis was performed on 22 selected cases (18% of all MASH cases reviewed), including cases with reduced or increased inter-pathologist variability and pathologist-AI discrepancies. Results: Pathologists reported the assistance tool as significantly helpful in 57% of cases, prompted conscious score changes in a further 12%, and not useful in 30%. Notably, score shifts between cases with and without assistance occurred even when the assistance tool was deemed unhelpful, indicating a subtle influence on scoring. The assistance tool reduced discrepancies in 65% (33/51) of previously discrepant cases, but increased discrepancies in 27% (15/66 with 1-score discrepancy and 3/66 with >1-score discrepancy) of cases initially without discrepancies. This reflects inherent pathologist variability and differing thresholds and confidence in challenging or borderline cases. Analysis of concordance changes with the assistance tool provided insights into pathologists’ fibrosis staging considerations. Conclusion: SHG-based AI-DP tools are promising adjuncts for fibrosis scoring in MASH. Our findings provide practical insights and recommendations for effective AI integration and highlight areas for future AI development. Collaborative human-AI approaches will be key to advancing precision histopathology and improving clinical trial outcomes in MASH. Impact and implications: AI-powered digital pathology (AI-DP) as a pathologist’s assistance tool can be utilized to improve pathologist concordance. However, pathologist judgement remains critical. In this study, we evaluated how pathologists used the assistance tool and how the platform impacted their decision-making. The findings are significant for pathologists and clinical researchers as they provide guidance for the well-informed use of SHG-based AI-DP as a tool to aid pathologists in fibrosis scoring in metabolic dysfunction-associated steatohepatitis. The insights gained can further inform policymakers about the integration of AI technologies in clinical trials or pathology workflows in clinical practice, promoting standardized practices.
Optimally sequencing experimental assays in drug discovery is a high-stakes planning problem under severe uncertainty and resource constraints. A primary obstacle for standard reinforcement learning (RL) is the absence of an explicit environment simulator or transition data $(s, a, s')$; planning must rely solely on a static database of historical outcomes. We introduce the Implicit Bayesian Markov Decision Process (IBMDP), a model-based RL framework designed for such simulator-free settings. IBMDP constructs a case-guided implicit model of transition dynamics by forming a nonparametric belief distribution using similar historical outcomes. This mechanism enables Bayesian belief updating as evidence accumulates and employs ensemble MCTS planning to generate stable policies that balance information gain toward desired outcomes with resource efficiency. We validate IBMDP through comprehensive experiments. On a real-world central nervous system (CNS) drug discovery task, IBMDP reduced resource consumption by up to 92\% compared to established heuristics while maintaining decision confidence. To rigorously assess decision quality, we also benchmarked IBMDP in a synthetic environment with a computable optimal policy. Our framework achieves significantly higher alignment with this optimal policy than a deterministic value iteration alternative that uses the same similarity-based model, demonstrating the superiority of our ensemble planner. IBMDP offers a practical solution for sequential experimental design in data-rich but simulator-poor domains.
