Union-FAST: An Intelligent-Agent Intervention to Increase Antiviral Treatment Uptake in Diagnosed-but-Untreated Hepatitis B Patients

1. Study Overview 1.1 Study Title An Intelligent-Agent Intervention (Union-FAST) to increase antiviral treatment uptake in diagnosed-but-untreated hepatitis B patients: a multicenter prospective study 1.2 Study Rationale The World Health Organization (WHO) 2016 Global Hepatitis Strategy targets ≥90% diagnostic coverage and ≥80% treatment coverage for chronic hepatitis B (CHB) by 2030 \[1\]. However, global progress remains sluggish: in 2024, only 26.1% of CHB patients were diagnosed and 14.6% received antiviral treatment, with merely 13% of infected individuals aware of their status and \<3% on therapy by the end of 2022 \[2-3\]. China, the country with the highest HBV burden globally, faces the same gap-antiviral uptake at Union Hospital Wuhan only rose from 55% to 62% after the 2022 Chinese national guidelines, still far below the WHO target \[4\].

Diagnosed-but-untreated (DBU) CHB patients face key barriers including limited disease knowledge, concerns about adverse drug effects, fragmented care pathways, and inadequate physician-patient communication. Traditional hospital-based follow-up relies on integrated hospital information systems, limiting scalability to secondary and community care \[5\]. In contrast, AI-based intelligent-agent models enable web/mobile-delivered education, dynamic follow-up, and risk assessment, with mature applications in CHB management (e.g., 95% diagnostic concordance, real-time guideline-concordant plan generation) \[6\].

This study evaluates Union-Agent-a cloud-based AI assistant developed under China's National Key R\&D Program (2022YFC2305100), integrating speech recognition, natural language processing, knowledge-graph retrieval, and large-language-model reasoning-for DBU CHB patients. The Union-FAST framework (Find DBU patients, Analyze barriers, Strategize interventions, Tool implementation) aims to address care gaps and improve antiviral treatment initiation.

1.3 Study Objectives Primary Objectives To evaluate whether Union-Agent increases the 6-month antiviral treatment initiation rate in DBU CHB patients meeting the 2022 Chinese Guidelines for the Prevention and Treatment of Chronic Hepatitis B by at least 10 percentage points (or a 50% relative increase) from baseline.

Secondary Objectives

1. Identify the multilevel barriers (individual, healthcare system, societal) to antiviral treatment in DBU CHB patients via Union-Agent's data collection and analysis. 2. Improve core HBV knowledge recognition in ≥80% of DBU patients within 6 months through Union-Agent's intelligent education modules. 3. Enhance follow-up adherence (shorten review intervals, increase on-schedule visit rates) in DBU patients with a history of irregular follow-up. 4. Timely detect high-risk events (ALT flares, HBV DNA rebound, incipient cirrhosis, hepatocellular carcinoma risk, and reduce end-stage liver disease incidence via real-time risk alerts and early intervention). 5. Optimize hierarchical care pathways to reduce unnecessary tertiary hospital referrals, waiting times, and economic burden for patients.

1.4 Study Design

A multicenter, prospective, dual-cohort (observational + interventional) study with a self-controlled baseline design. The study includes two core phases:

1. Baseline cross-sectional survey: Characterize DBU patients, identify treatment barriers, and establish stratification variables for intervention. 2. Prospective interventional cohort: Patients receive routine care plus four Union-Agent-driven services, with real-time data collection to evaluate treatment uptake and clinical outcomes.

Study Setting: 8 collaborating centers across China (tertiary, secondary, and community hospitals)

* Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (Leading) * Zhongnan Hospital Affiliated to Wuhan University * First Affiliated Hospital of the School of Medicine, Shihezi University * Taihe Hospital in Shiyan * Yichang Central People's Hospital * The Second Affiliated Hospital of Anhui Medical University * Wuhan Jinyintan Hospital * Zigui County People's Hospital Intervention: Union-Agent (web/mobile-based AI platform) with four core modules for patients and a dedicated physician portal (see Section 3 for details).

Comparator: Self-baseline (no additional digital intervention beyond routine clinical care).

2\. Study Population 2.1 Diagnostic Criteria CHB is defined as persistent HBsAg positivity for ≥6 months, in accordance with the latest Chinese Guidelines for the Prevention and Treatment of Chronic Hepatitis B.

2.2 Inclusion Criteria

1. Aged ≥18 years with persistent HBsAg positivity for ≥6 months. 2. Treatment-naïve to nucleos(t)ide analogues (NUCs), or previous NUC treatment with no medication for the past 6 months (DBU status). 3. Capable of using a smartphone independently or with family member assistance. 4. Voluntarily provides written informed consent (electronic version) and agrees to study procedures and follow-up.

2.3 Exclusion Criteria

1. Severe mental or cognitive impairment that impairs study cooperation. 2. Current participation in other interventional clinical trials that may influence HBV treatment decisions. 3. Inability to use a smartphone even with supervision. 4. Refusal to sign the informed consent form. 2.4 Sample Size Justification The sample size is calculated using the classical single-proportion precision formula for a 95% confidence interval (CI): N = (Z\_{0.975}\^2 \* p\*(1-p)) / d\^2

* Z0.975=1.96 (95% CI standard normal deviate) * p=0.5 (conservative estimate to maximize variance) * d=0.022 (allowable margin of error: ±2.2 percentage points) Calculated sample size: 1,985 patients. Accounting for: DEFF = 1 + (m - 1) \* ρ * Design effect (DEFF): , 1.5-3.0 to adjust for intra-center correlation in the multicenter design. * Loss to follow-up: 10-15%. Final planned enrollment: 2,000 DBU CHB patients (expected screen: 2,200 patients; enrollment period: 6 months).

3\. Intervention: Union-Agent AI Platform Union-Agent is a cloud-based intelligent agent developed for HBV care, with a bidirectional interface (physician + patient) and closed-loop management of education, medication, follow-up, and referral. All functions comply with the 2022 Chinese CHB Guidelines and 2024 WHO HBV Treatment Guidelines.

3.1 Core Services for Patients

1. Guideline-based personalized education: Tailored to baseline HBV knowledge, including concise disease information, antiviral benefits, drug safety, and short knowledge assessments. 2. Standardized treatment recommendations: Clear, interpretable guidance on first-line NUC regimens (entecavir \[ETV\], tenofovir disoproxil fumarate \[TDF\], tenofovir alafenamide fumarate \[TAF\]). 3. Individualized medication reminders: Real-time adherence logging, alerts for missed doses, and coping strategies for common adverse events. 4. Semantic follow-up and risk assessment: Patient self-report of symptoms (voice/image input), conversion to structured data, and automated risk stratification with real-time alerts for high-risk events. 5. Automated scheduling: Reminders for routine reviews, lab tests, and appointment booking; support for uploading lab results for physician review.

3.2 Physician Portal Functions

1. Patient profile creation and verification of antiviral treatment eligibility per national guidelines. 2. Automated alerts for eligible but untreated DBU patients. 3. Review of patient-reported symptoms, structured risk assessments, and high-risk event alerts. 4. Push of tailored education, medication reminders, and follow-up tasks to the patient interface. 5. Direct communication with patients (reminders, referral links, and clinical evaluations) for missed follow-up or incomplete tasks. 6. Generation of individualized long-term risk assessments (cirrhosis, hepatocellular carcinoma, liver failure).

3.3 Data Security and Privacy

* Access limited to authorized personnel via multi-factor authentication. * All data exported in de-identified, encrypted formats (k-anonymization, federated learning; ISO/IEC 27799 compliant). * Unified data governance framework adhering to Chinese national information security regulations, the Declaration of Helsinki, and Good Clinical Practice (GCP) standards. * Privacy breach risk: \<2%. 4. Study Procedures and Timeline

The total study duration is 18 months, divided into 4 phases (Union-FAST implementation aligned with all phases):

Phase Timeframe Key Activities

1. Study Setup \& System Validation Months 0-2 Finalize Union-Agent modules; 2 simulation cycles; obtain Institutional Review Board (IRB)/Ethics Committee (EC) approval; train investigators at all centers. 2. Patient Recruitment \& Baseline Survey Months 3-8 Screen 2,200 patients; enroll 2,000 eligible DBU patients; conduct cross-sectional baseline survey (demographics, medical history, HBV knowledge, treatment barriers) via Union-Agent's mobile interface; establish baseline treatment initiation rate and stratification variables. 3. Intervention \& Real-Time Data Collection Months 9-14 Implement 6-month Union-Agent intervention; collect real-time data (treatment initiation, medication adherence, follow-up visits, lab results, high-risk events); deliver routine clinical care plus Union-Agent services. 4. Data Analysis \& Reporting Months 15-18 Comprehensive data cleaning and statistical analysis; prepare clinical study report; draft 1-2 peer-reviewed manuscripts; present interim results at AASLD 2026/APASL 2027; submit final manuscript in 2027.

4.1 Follow-Up Requirements

* Frequency: Real-time digital follow-up via Union-Agent; in-person/clinical follow-up per routine CHB care guidelines (adjusted per patient risk stratification). * Assessments: HBV knowledge tests (baseline + 6 months); lab tests (ALT, HBV DNA, liver function; per clinical need); treatment initiation status (6 months); high-risk event detection (real-time); patient satisfaction survey (6 months).

4.2 Study Termination for Individual Patients

1. Voluntary withdrawal by the patient. 2. Severe non-study-related illness or death that prevents study participation. 3. Protocol violation by the patient (e.g., non-compliance with Union-Agent use). 4. Investigator decision (e.g., clinical deterioration requiring urgent intervention). 5. Outcome Measures 5.1 Primary Outcome 6-month antiviral treatment initiation rate in DBU patients meeting 2022 Chinese CHB guideline indications:

* Absolute increase: ≥10 percentage points from baseline. * Relative increase: 50% from baseline. 5.2 Secondary Outcomes

<!-- -->

1. Barrier characterization: Identification of individual/healthcare system/societal barriers to treatment via baseline and intervention-phase data (Union-Agent survey and clinical records). 2. HBV knowledge acquisition: Proportion of patients with correct recognition of core HBV knowledge (≥80% at 6 months). 3. Follow-up adherence: Change in average review intervals (30% reduction from baseline); on-schedule visit rate; follow-up completion rate. 4. High-risk event detection: Proportion of high-risk events (ALT flares, HBV DNA rebound, incipient cirrhosis) detected in real time (≥90%); incidence of end-stage liver disease events (compared with historical controls). 5. Care pathway optimization: Reduction in unnecessary tertiary hospital referrals; decrease in average appointment waiting time; patient satisfaction score (≥80/100 at 6 months). 6. Medication adherence: Proportion of patients with ≥80% medication adherence (real-time logging via Union-Agent).

5.3 Safety Outcomes

This study involves no additional drug/surgical intervention (only digital AI-based care support). Safety outcomes include:

• Adverse events (AEs) related to routine HBV antiviral treatment (captured via Union-Agent and clinical records).

* Severe adverse events (SAEs): Any life-threatening, disabling, or fatal event related to HBV or its treatment (reported to the IRB/EC within 24 hours). *

• Multilevel Barriers to Antiviral Treatment in DBU CHB Patients [Time frame: From baseline assessment until 6 months of intervention, assessed at baseline and 6 months of intervention via Union-Agent's built-in structured questionnaire and clinical record review.]
• Proportion of DBU Patients with Correct Recognition of Core HBV Knowledge [Time frame: From baseline assessment until 6 months of intervention, assessed at baseline and 6 months of intervention using the Union-Agent HBV Knowledge Assessment Scale.]
• Change in Average Review Intervals in DBU Patients with Irregular Follow-Up History [Time frame: From the start of Union-Agent intervention until 6 months of intervention, assessed up to 6 months via Union-Agent follow-up logs to determine changes in average review intervals.]
• Proportion of Real-Time High-Risk Event Detection in DBU Patients [Time frame: From the start of Union-Agent intervention until 6 months of intervention, assessed up to 6 months via the Union-Agent risk stratification module to detect high-risk events in real time.]