LAI-PrEP Bridge Period Decision Support Tool v4.1.0: Code, Configuration, and Supplementary Materials

Zenodo v4.1.0 Update - Complete Documentation Update Date: December 12, 2025Previous Version DOI: 10.5281/zenodo.17727117 (v2.1.0)NEW Version DOI: 10.5281/zenodo.17873201 (v4.1.0)Full URL: https://zenodo.org/record/17873201 What's New in v4.1.0 Manuscripts Uploaded (Viruses Journal Submission) lai_tool_final.tex - LAI-PrEP Computational Validation Manuscript Complete computational validation at UNAIDS global scale (21.2M patients) Progressive validation across 4 scales (1K, 1M, 10M, 21.2M) Comprehensive edge case testing (18/18 pass rate, 100%) Full Discussion section addressing AI suitability in healthcare Supplementary_File_S3_AI_Readiness_Healthcare.tex - Framework for Responsible AI Deployment Critical examination of computational vs. clinical validity Evidence quality assessment (Tier 1-3 classification) Interpretability and algorithmic transparency analysis Equity and health disparities considerations Benefit-risk calculus and staged implementation framework Supplementary Files Supplementary Materials (S1-S4) S1: Machine-readable configuration files (JSON) S2: Complete 21-intervention library with evidence synthesis S3: AI Readiness framework (as detailed above) S4: Code repository documentation and reproducibility instructions Documentation Updates README with reproducibility instructions Updated LICENSE files Citation metadata (CITATION.cff) Version History Version Date DOI Focus Status v2.1.0 Oct 24, 2025 10.5281/zenodo.17727117 Initial code release Superseded v4.1.0 Dec 12, 2025 10.5281/zenodo.17873201 Viruses manuscript submission CURRENT Key Metrics from v4.1.0 Manuscripts Computational Validation Results Sample scales: 1,000 → 1,000,000 → 10,000,000 → 21,200,000 patients Algorithmic precision: ±0.018 percentage points (95% CI) at 21.2M scale Precision improvement: 144-fold increase from 1K scale Test pass rate: 18/18 edge cases (100%) Convergence: Mean success rates stabilized by 1M patients Primary Findings Baseline bridge period success: 23.96% (95% CI: 23.94–23.98%) With interventions: 43.50% (95% CI: 43.48–43.52%) Relative improvement: 81.6% Global impact: 4.1 million additional successful transitions Population Disparities PWID baseline: 10.36% (highest need) MSM baseline: 33.11% (lowest need) Disparity gap: 22.75 percentage points PWID intervention benefit: +265% relative improvement Adolescent benefit: +147% relative improvement Regional Analysis (UNAIDS Global Scale) Sub-Saharan Africa: 62% of global patients, 21.69% baseline success Europe/Central Asia: 6% of patients, 29.33% baseline success Regional equity gap: 7.64 percentage points SSA relative improvement with interventions: +91.2% Economic Projections Annual HIV infections prevented: ~80,000–200,000 (midpoint: 100,000) Lifetime treatment costs saved: $40 billion Implementation cost: $19.1 billion Annual ROI: 2.1:1 5-year cumulative ROI: 10.5:1 Complete Manuscript Contents Main Manuscript Structure Title: Computational Validation of a Clinical Decision Support Algorithm for LAI-PrEP Bridge Period Navigation at UNAIDS Global Target Scale Journal: Viruses (MDPI) Article Type: Original research manuscript Word Count: ~52,000 (including supplementary materials) Tables: 16 total (main + supplementary) Figures: 8 total (including supplementary) References: 87 citations Section Breakdown Main Manuscript (lai_tool_final.tex): Introduction LAI-PrEP promise and implementation challenges Bridge period attrition crisis (47% failure rate) Need for computational decision support Study objectives and distinction between computational vs. clinical validity Materials and Methods Evidence synthesis from >15,000 clinical trial participants Algorithm development with three-tier evidence classification Population-specific baseline rates 21 structural barriers with quantified impacts 21 evidence-based interventions with effect sizes Configuration-driven software architecture Mechanism diversity scoring algorithm Synthetic population generation procedures Intervention combination models (two-stage approach) Progressive validation study design (4 tiers) Comprehensive edge case testing (18 scenarios) Outcome measures and statistical analysis Software availability and data sharing Results Progressive validation: Convergence and precision analysis Unit test results across all validation tiers Comprehensive edge case testing (100% pass rate) Population-specific predictions vs. published trials Population-specific intervention effects Regional analysis at UNAIDS global scale Barrier impact analysis with dose-response relationship Risk stratification distribution Global impact projections Discussion Principal findings and contributions Computational precision vs. clinical uncertainty Framework for prospective clinical validation Contextualization of findings Strengths and limitations AI Suitability for Healthcare: 5 Critical Questions External validity and false confidence Evidence quality and extrapolated parameters Interpretability and clinical oversight Equity and population heterogeneity Benefit-risk calculus and staged implementation Limitations of computational validation vs. real-world performance Future directions Conclusions Summary of computational validation achievements Distinction between algorithmic and clinical readiness Call for prospective validation and equity-focused implementation Commitment to responsible AI deployment Supplementary File S3: AI Readiness in Healthcare Comprehensive 45-page framework addressing: External Validity Computational precision ≠ clinical certainty Mathematical vs. external vs. prospective validity Synthetic data limitations Staged implementation approach Evidence Quality Tier 1 (direct LAI-PrEP): 8 interventions Tier 2 (HIV prevention analogs): 9 interventions Tier 3 (cross-field extrapolation): 4 interventions Parameter uncertainty vs. computational precision Dynamic evidence integration strategies Interpretability Algorithmic transparency: How calculations work Mechanistic reasoning: Why recommendations are made Uncertainty quantification: Confidence intervals Population-specific baselines Interpretability paradox and error detection Supporting clinical judgment over algorithmic certainty Equity and Heterogeneity Aggregation bias in healthcare AI Multi-dimensional stratification approach Individual barrier assessment (13 barriers) Algorithmic fairness considerations Within-population heterogeneity recognition Distributional impact assessment Benefit-Risk Calculus Projected benefits: 4.1M transitions, 100K infections prevented Implementation risks: Resource misallocation, false confidence, equity harm Staged implementation framework: Phase 1: Pilot validation (2-3 sites, 50-100 patients) Phase 2: Multi-site validation (10-15 sites, 500-1000 patients) Phase 3: Scaled implementation with continuous monitoring Limitations of Computational Validation Simulation vs. reality differences Parameter uncertainty vs. computational precision Context-specificity of parameters Path forward: Prospective validation and continuous refinement 🔗 Updated Citations for Manuscripts For Data Availability Statements APA Format: Demidont, A. C. LAI-PrEP bridge period decision support tool: Computational validation at UNAIDS global scale [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.17873201 BibTeX Format: @software{demidont2025laiprep, author = {Demidont, Adrian C}, title = {LAI-PrEP Bridge Period Decision Support Tool}, subtitle = {Computational Validation at UNAIDS Global Scale}, version = {4.1.0}, year = 2025, publisher = {Zenodo}, doi = {10.5281/zenodo.17873201}, url = {https://doi.org/10.5281/zenodo.17873201} } Chicago Manual of Style (Notes-Bibliography): Demidont, Adrian C. "LAI-PrEP Bridge Period Decision Support Tool: Computational Validation at UNAIDS Global Scale." Zenodo. December 12, 2025. https://doi.org/10.5281/zenodo.17873201. Vancouver Format: Demidont AC. LAI-PrEP bridge period decision support tool: Computational validation at UNAIDS global scale [computer software]. Zenodo. 2025. https://doi.org/10.5281/zenodo.17873201 Nature Format: Demidont, A. C LAI-PrEP bridge period decision support tool: Computational validation at UNAIDS global scale. Zenodo https://doi.org/10.5281/zenodo.17873201 (2025). MLA Format (9th Edition): Demidont, Adrian C. "LAI-PrEP Bridge Period Decision Support Tool: Computational Validation at UNAIDS Global Scale." Version 4.1.0, Zenodo, 12 Dec. 2025, doi.org/10.5281/zenodo.17873201. 📄 Manuscript LaTeX References Updated In Data Availability Statement: \section*{Data Availability Statement} All code, configuration files, validation datasets, and supplementary materials are publicly available on Zenodo (DOI: \url{https://zenodo.org/record/17873201}) and GitHub Repository \url{https://github.com/Nyx-Dynamics/lai-prep-bridge-tool-pub} (release v4.1.0, commit: [current-commit-hash]). In GitHub Badge Section (if applicable): \includegraphics[alt={DOI}]{https://zenodo.org/badge/DOI/10.5281/zenodo.17873201.svg} Files in v4.1.0 Zenodo Record Manuscripts lai_tool_final.tex (52KB) Supplementary_File_S3_AI_Readiness_Healthcare.tex (45KB) [Additional supplementary files S1, S2, S4 as applicable] Software lai_prep_decision_tool_v2_1.py (850 lines) lai_prep_config.json (configuration with 21 interventions) test_edge_cases.py (18 test scenarios) Validation result JSONs (1K, 1M, 10M, 21.2M scales) Documentation README.md (installation, usage, reproducibility) LICENSE.md (MIT + CC-BY 4.0) CITATION.cff (machine-readable citations) CONTRIBUTING.md (if applicable) Contact & Citation Information For Questions Abo

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.