AI-Guided Intensified Follow-up After Surgery for Acute Type A Aortic Dissection

Official title

Artificial Intelligence Multimodal Prediction Model-Guided Intensified Follow-up Strategy for 1-Year Outcomes After Surgery for Acute Type A Aortic Dissection: A Multicenter Randomized Controlled Trial

Overview

This multicenter, prospective, randomized controlled trial will evaluate whether an artificial intelligence (AI) multimodal prediction model-guided intensified follow-up strategy improves 1-year outcomes after surgery for acute Stanford type A aortic dissection. Eligible adult patients who have undergone open surgical repair or open plus endovascular/hybrid repair and are clinically stable to enter the postoperative follow-up phase will be randomized 1:1 to usual postoperative follow-up or AI-guided intensified follow-up. The AI-guided arm will receive usual follow-up plus an AI-generated risk stratification report for 1-year mortality and adverse aortic remodeling. Higher-risk patients may receive more frequent follow-up, prioritized CTA review, multidisciplinary assessment, and targeted management reminders. The primary outcome is all-cause mortality through postoperative day 365. Key secondary outcomes include aortic reintervention, adverse aortic remodeling, and ICU readmission within 1 year.

Detailed description

Acute Stanford type A aortic dissection is a life-threatening aortic disease. Even after successful surgical repair, patients may experience death, residual dissection progression, false lumen patency, distal aortic dilation, reintervention, infection, renal dysfunction, readmission, or other adverse outcomes during the first postoperative year. Current follow-up pathways may not fully integrate longitudinal clinical, biochemical, CTA imaging, and hemodynamic/biomechanical information for individualized risk management.

This study is a multicenter, prospective, randomized, controlled, open-label trial with blinded endpoint adjudication. It tests whether embedding a previously developed and validated AI multimodal prediction model into the postoperative follow-up pathway can identify high-risk patients earlier, improve completion of imaging follow-up, trigger multidisciplinary review, and support intensified management, thereby improving 1-year clinical outcomes.

Eligible participants will be adults with acute Stanford type A aortic dissection who have undergone open surgery or open combined with endovascular/hybrid repair, have stable postoperative status, and are ready for discharge or early postoperative follow-up. After written informed consent and confirmation of eligibility, participants will be randomized in a 1:1 ratio to usual postoperative follow-up or AI prediction model-guided intensified follow-up. Randomization is planned to be center-stratified with block randomization, with optional pre-specified stratification by early postoperative risk, DeBakey type, or Penn class according to the final randomization plan.

The usual follow-up group will receive each center's standard ATAAD postoperative follow-up pathway, including discharge education, blood pressure and medication management, outpatient and telephone follow-up, and CTA or ultrasound review according to local practice. The AI-guided intensified follow-up group will receive usual follow-up plus an AI-generated risk report classifying participants as low, moderate, high, or very high risk for 1-year mortality and adverse aortic remodeling. Depending on the risk level, clinicians may arrange intensified telephone or outpatient follow-up, earlier or prioritized CTA review, multidisciplinary team discussion, reminders for blood pressure, renal function, infection or nutritional management, rapid review of abnormal imaging findings, and reintervention assessment pathways. The AI report is an auxiliary decision-support tool and does not replace guideline-based care, imaging review, or the treating surgical team's judgment.

The planned total sample size is 1,314 participants, approximately 657 per group, allowing for 10% loss to follow-up or major protocol deviation. The primary endpoint is all-cause mortality from randomization through postoperative day 365. Key secondary endpoints include 1-year aortic reintervention, adverse aortic remodeling, and ICU readmission. Other secondary outcomes include aorta-related death, unplanned readmission, major adverse cardiovascular events, stroke, renal failure or continuous renal replacement therapy, infection, imaging follow-up completion, follow-up adherence, and execution of AI-triggered follow-up actions. Outcome events will be supported by hospital records, follow-up data, death registry information where available, imaging core laboratory review, AI system logs, and blinded Clinical Event Committee adjudication.

Interventions

• Other Usual Postoperative Follow-up Strategy
  Routine postoperative surveillance and management after acute type A aortic dissection repair according to each center's standard clinical practice.
• Other AI Prediction Model-Guided Intensified Follow-up Strategy
  Use of a previously developed AI multimodal prediction model to generate risk stratification for postoperative 1-year mortality and adverse aortic remodeling, triggering protocolized intensified follow-up actions when indicated.

Primary outcome measures

• All-Cause Mortality Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]

Secondary outcome measures (6)

• Aortic Reintervention Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]
• Adverse Aortic Remodeling Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]
• ICU Readmission Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]
• Aorta-Related Mortality Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]
• Unplanned Hospital Readmission Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]
• Completion Rate of Follow-up CTA Imaging Within 365 Days After Surgery [Time frame: From randomization to postoperative day 365]

Eligibility criteria

Inclusion criteria

• 1\. Age 18 years or older. 2. Acute Stanford type A aortic dissection confirmed by CTA, intraoperative findings, or medical records.

3\. Underwent open surgical repair or open repair combined with endovascular/hybrid repair.

4\. Postoperative condition is stable and the patient is planned for discharge or has entered early post-discharge follow-up.

5\. Core baseline data required for AI model operation are available, including at least one eligible preoperative or postoperative CTA imaging dataset.

6\. Able to complete telephone, outpatient, or inpatient follow-up and willing to provide written informed consent.

Exclusion criteria

• 1\. Predominantly chronic type A dissection, traumatic dissection, or iatrogenic dissection, if the investigator judges the patient unsuitable for this follow-up strategy study.

2\. Did not undergo surgical repair, received only palliative treatment, or did not meet postoperative randomization conditions.

3\. Expected inability to complete 12-month follow-up, no stable contact information, or inability to obtain outpatient, telephone, or inpatient follow-up data.

4\. Severe missing baseline data preventing generation of the AI risk report. 5. Concurrent participation in another interventional study that may substantially affect postoperative follow-up intensity or the primary outcome.

6\. Any other condition that, in the investigator's judgment, makes the patient unsuitable for participation.

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: No

Study design

Allocation

Randomized

Model

Parallel assignment

Masking

Single blind

Primary purpose

Health services research

Study locations

Center list to be confirmed — check the primary protocol.

Publications

• Zhu Y, Xu XY, Rosendahl U, Pepper J, Mirsadraee S. Prediction of aortic dilatation in surgically repaired type A dissection: A longitudinal study using computational fluid dynamics. JTCVS Open. 2022 Feb 9;9:11-27. doi: 10.1016/j.xjon.2022.01.019. eCollection 2022 Mar. PMID 36003481
• Shad R, Kong S, Fong R, Quach N, Kasinpila P, Bowles C, Lee A, Hiesinger W. Computational Fluid Dynamics Simulations to Predict False Lumen Enlargement After Surgical Repair of Type-A Aortic Dissection. Semin Thorac Cardiovasc Surg. 2022 Summer;34(2):443-448. doi: 10.1053/j.semtcvs.2021.05.012. Epub 2021 Jun 3. PMID 34091015
• Nappi F, Gambardella I, Singh SSA, Salsano A, Santini F, Spadaccio C, Biancari F, Dominguez J, Fiore A. Survival following acute type A aortic dissection: a multicenter study. J Thorac Dis. 2023 Dec 30;15(12):6604-6622. doi: 10.21037/jtd-23-1137. Epub 2023 Dec 21. PMID 38249919
• Gemelli M, Di Tommaso E, Chivasso P, Sinha S, Ahmed EM, Rajakaruna C, Bruno VD. Blood lactate predicts mortality after surgical repair of type A acute aortic dissection. J Card Surg. 2022 May;37(5):1206-1211. doi: 10.1111/jocs.16324. Epub 2022 Feb 13. PMID 35152486
• Zhou T, Li JX, Zhang CY, Li YG, Peng J, Wei CL, Chen MH, Zhou HF. Risk factors for one-year mortality following discharge in patients with acute aortic dissection: development and validation of a predictive model in a cross-sectional study. BMC Cardiovasc Disord. 2024 Feb 29;24(1):129. doi: 10.1186/s12872-024-03766-6. PMID 38424525
• Yang Y, Xue J, Li H, Tong J, Jin M. Perioperative risk factors predict one-year mortality in patients with acute type-A aortic dissection. J Cardiothorac Surg. 2020 Sep 11;15(1):249. doi: 10.1186/s13019-020-01296-8. PMID 32917250

Identifiers

NCT: NCT07623213 · ATAAD-AI-RCT-001

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗