A Bayesian decision framework for F1–F2 MASLD follow-up under adherence uncertainty
9 affiliations
- 1Siemens Medical Solutions USA Inc.
- 2Northwell Health Feinstein Institutes for Medical Research
- 3Donald and Barbara Zucker School of Medicine at Hofstra/Northwell
- 4Siemens Healthcare Diagnostics Inc.
- 5Medical Affairs, Universitat de Barcelona
- 6Instituto de Salud Global de Barcelona
- 7CUNY Graduate School of Public Health & Health Policy
- 8Division of Gastroenterology, Virginia Commonwealth University School of Medicine
- 9Thomas Jefferson University College of Population Health
Abstract
Introduction: Metabolic dysfunction-associated steatotic liver disease (MASLD) affects over one-third of the global population, yet follow-up frequency for patients with mild-to-moderate fibrosis (F1–F2) remains undefined. Variable adherence to lifestyle modification and to pharmacotherapies such as resmetirom and semaglutide further complicates management, risking both resource overuse and undetected disease progression.
Methods: We developed a Bayesian-informed expected net monetary benefit (NMB) decision framework comparing three strategies: aggressive follow-up (AF, visits every 6 months), conservative follow-up (CF, annual visits), and CF with mandatory between-visit Enhanced Liver Fibrosis testing (CF + ELF). Game theory concepts of adverse selection, moral hazard, and signaling motivated the model structure. Payoffs used published fibrosis transition probabilities, US EQ-5D-5L utilities, and Medicare costs at a willingness-to-pay threshold of $100,000/QALY. Adherence probability p was calibrated to therapy-specific proxy estimates. Sensitivity analyses varied the fibrosis improvement probability, the early-detection benefit of AF (0–75%), and the accountability effect of mandatory ELF (δ).
Results: AF was preferred over CF when p < 0.559, and CF + ELF was preferred over AF when p > 0.787. For lifestyle modification ( p = 0.30), AF was model-preferred. For semaglutide ( p = 0.835) and resmetirom ( p = 0.855), CF + ELF was model-preferred if mandatory ELF reduced non-adherent progression by at least 68% and 77%, respectively, and CF otherwise. CF was model-preferred in the intermediate zone (0.559 < p < 0.787), which is relevant to future therapies. Thresholds were invariant to the fibrosis improvement probability, and therapy-specific recommendations held when AF reduced non-adherent progression by approximately 31–63%. Patients weakly preferred AF on health outcome grounds, strictly so when non-adherent, suggesting that observed preferences for CF reflect burden. In a worked Bayesian example, two consecutive worsening ELF results lowered a resmetirom patient’s adherence probability from 0.855 to 0.506, making escalation to AF model-preferred.
Discussion: Follow-up intensity in F1–F2 MASLD may be better calibrated to treatment context and adherence probability than to uniform schedules. A hybrid approach combining patient signaling with physician oversight, informed by non-invasive testing and social determinants of health, offers an adaptable framework across lifestyle and pharmacological treatment settings. Prospective estimation of the early-detection benefit of AF and the accountability effect of mandatory ELF is needed before clinical implementation.
Related
- Adherence to vaccination schedules in patients under active oncological treatment and follow-up: protocol for a systematic review and meta-analysis.
- Lung-Reporting and Data System Category (Lung-RADS) and Adherence to Low-Dose Computed Tomography Screening and Recommended Follow-Up: A Multicenter Cohort Study
- Redefining Non Invasive Post Transplant Surveillance: A Bayesian Meta Analysis and Decision Curve Framework for Donor Derived Cell Free DNA in Heart Transplantation
- From CPAP Prescription to Long-Term Follow-Up: Care Pathway, Adherence, and Multidomain Treatment Outcomes in Obstructive Sleep Apnea
- High-intensity interval training in cancer patients: a bayesian meta-analysis of effects on VO2max, dose–response, and adherence
BibTeX
@article{boltyenkov2026bayesian,
title = {{A Bayesian decision framework for F1–F2 MASLD follow-up under adherence uncertainty}},
author = {Artem T. Boltyenkov and Michael Eiswerth and Pina C. Sanelli and Jeffrey V. Lazarus and Jason J. Wang and Wanyi Chen and Szu-Yu Kao and Donald B. Chalfin and Kinpritma Sangha and Arun J. Sanyal},
journal = {Frontiers in Public Health},
year = {2026},
month = oct,
doi = {10.3389/fpubh.2026.1930169},
url = {https://www.frontiersin.org/articles/10.3389/fpubh.2026.1930169/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - A Bayesian decision framework for F1–F2 MASLD follow-up under adherence uncertainty
AU - Artem T. Boltyenkov
AU - Michael Eiswerth
AU - Pina C. Sanelli
AU - Jeffrey V. Lazarus
AU - Jason J. Wang
AU - Wanyi Chen
AU - Szu-Yu Kao
AU - Donald B. Chalfin
AU - Kinpritma Sangha
AU - Arun J. Sanyal
PY - 2026
DA - 2026/10/26
JO - Frontiers in Public Health
DO - 10.3389/fpubh.2026.1930169
UR - https://www.frontiersin.org/articles/10.3389/fpubh.2026.1930169/full
AB - Introduction: Metabolic dysfunction-associated steatotic liver disease (MASLD) affects over one-third of the global population, yet follow-up frequency for patients with mild-to-moderate fibrosis (F1–F2) remains undefined. Variable adherence to lifestyle modification and to pharmacotherapies such as resmetirom and semaglutide further complicates management, risking both resource overuse and undetected disease progression. Methods: We developed a Bayesian-informed expected net monetary benefit (NMB) decision framework comparing three strategies: aggressive follow-up (AF, visits every 6 months), conservative follow-up (CF, annual visits), and CF with mandatory between-visit Enhanced Liver Fibrosis testing (CF + ELF). Game theory concepts of adverse selection, moral hazard, and signaling motivated the model structure. Payoffs used published fibrosis transition probabilities, US EQ-5D-5L utilities, and Medicare costs at a willingness-to-pay threshold of $100,000/QALY. Adherence probability p was calibrated to therapy-specific proxy estimates. Sensitivity analyses varied the fibrosis improvement probability, the early-detection benefit of AF (0–75%), and the accountability effect of mandatory ELF (δ). Results: AF was preferred over CF when p < 0.559, and CF + ELF was preferred over AF when p > 0.787. For lifestyle modification ( p = 0.30), AF was model-preferred. For semaglutide ( p = 0.835) and resmetirom ( p = 0.855), CF + ELF was model-preferred if mandatory ELF reduced non-adherent progression by at least 68% and 77%, respectively, and CF otherwise. CF was model-preferred in the intermediate zone (0.559 < p < 0.787), which is relevant to future therapies. Thresholds were invariant to the fibrosis improvement probability, and therapy-specific recommendations held when AF reduced non-adherent progression by approximately 31–63%. Patients weakly preferred AF on health outcome grounds, strictly so when non-adherent, suggesting that observed preferences for CF reflect burden. In a worked Bayesian example, two consecutive worsening ELF results lowered a resmetirom patient’s adherence probability from 0.855 to 0.506, making escalation to AF model-preferred. Discussion: Follow-up intensity in F1–F2 MASLD may be better calibrated to treatment context and adherence probability than to uniform schedules. A hybrid approach combining patient signaling with physician oversight, informed by non-invasive testing and social determinants of health, offers an adaptable framework across lifestyle and pharmacological treatment settings. Prospective estimation of the early-detection benefit of AF and the accountability effect of mandatory ELF is needed before clinical implementation.
SN - 2296-2565
ER -