A Clinical Prediction Model for Surgical Site Infection After Central Nervous System Tumor Surgery

NCT07378683 | Recruiting DMC

• 2 other identifiers: NCC2025-CNS-SSI2024-I2M-3-014
• Study Type: interventional
• Target: 500
• Locations: 1 country
• Sites: 1

Brief Summary

1. 1.Study Background For patients undergoing brain or spinal tumor surgery, postoperative surgical wound infection (known as "surgical site infection," SSI) is a recognized risk. Once an infection occurs, it may complicate and prolong the recovery process. Currently, doctors primarily rely on clinical experience to judge which patients are at higher risk of infection. Our research team has previously developed an intelligent prediction model (a machine learning tool) that can very accurately identify which patients are at higher risk of postoperative infection. This study aims to validate whether implementing enhanced preventive measures for high-risk patients in advance, based on the model's predictions, can effectively reduce the occurrence of infection.
2. 2.Study Purpose The primary purpose of this study is to validate whether an individualized intervention strategy based on an intelligent prediction model can effectively reduce the incidence of surgical site infection in patients after central nervous system tumor surgery. Simultaneously, we will also evaluate the safety of this strategy, its impact on patient hospital stay length and medical costs, as well as its feasibility in practical clinical application.
3. 3.Study Design Type: This is a prospective, single-center clinical validation study. "Prospective" means the research plan is established first, followed by patient recruitment and data collection according to the plan; "single-center" indicates the study is conducted solely at the Chinese Academy of Medical Sciences Cancer Hospital.
4. 4.Primary Evaluation Indicators Primary Indicator: The incidence of surgical site infection within 90 days after surgery. The diagnosis of infection will be made by experts who are unaware of the patient's group assignment, strictly following international standards.
5. 5.Eligibility Criteria (Inclusion Criteria Summary)
6. 6.Patients' Rights and Safety Voluntary Participation: Participation in this study is entirely voluntary. You have the right to withdraw from the study at any time, which will not affect the quality of any normal medical services you are entitled to at our hospital or your relationship with it.
7. 7.Contact Information

Conditions

Surgical Site Infection (SSI); Central Nervous System Tumor

Keywords

surgical site infection; central nervous system tumor; machine learning; explainable artificial intelligence; SHAP; clinical decision support system

Outcome Measures

Primary Outcomes (1)

• Incidence of surgical site infection within 90 days postoperatively

  The occurrence of a surgical site infection (SSI) following central nervous system tumor surgery, diagnosed according to the U.S. National Healthcare Safety Network (NHSN) criteria. Diagnosis will be adjudicated by a blinded independent endpoint adjudication committee.

  90 days post-surgery

Study Arms (2)

High-Risk: ML-Based Intervention

EXPERIMENTAL

This is the High-Risk Intervention Group. Patients identified as high-risk for SSI (predicted probability ≥ 50%) by the LightGBM model within 72 hours post-surgery receive an enhanced, individualized intervention package, which may include an escalated antibiotic regimen, intensified nutritional support, and closer monitoring.

Other: ML-Based Enhanced Intervention Package for SSI Prevention

Low-Risk: Standard Care

ACTIVE COMPARATOR

This is the Low-Risk Routine Care Group. Patients identified as low-risk for SSI (predicted probability \< 50%) by the same model receive the current standard postoperative care of the hospital, following established clinical pathways without the enhanced intervention package. This arm serves as the control.

Other: Standard Postoperative Care

Interventions

ML-Based Enhanced Intervention Package for SSI Prevention OTHER

This is a comprehensive, individualized intervention package administered to patients identified as high-risk for surgical site infection (SSI) by a machine learning (LightGBM) model. The package is activated postoperatively and may include, but is not limited to: 1) \*\*Escalated/optimized antibiotic prophylaxis regimen\*\*; 2) \*\*Intensified nutritional support protocols\*\*; 3) \*\*Enhanced monitoring\*\* of vital signs and wound status. The specific components are tailored based on the patient's risk profile and clinical context. The intervention aims to validate the effectiveness of a model-driven strategy in reducing SSI incidence.

High-Risk: ML-Based Intervention

Standard Postoperative Care OTHER

This intervention represents the current standard of care provided to patients after elective central nervous system tumor surgery at this institution. It includes routine antibiotic prophylaxis, standard wound care, basic vital signs monitoring, and follow-up according to established clinical pathways. Patients in this group do not receive the enhanced, study-specific intervention package and serve as the control population for comparison.

Low-Risk: Standard Care

Eligibility Criteria

• Sex: all
• Healthy Volunteers: No
• Age Groups: Child (0-17), Adult (18-64), Older Adult (65+)

You may qualify if:

• Patients with a pathologically confirmed primary or metastatic central nervous system (brain or spinal) tumor.
• Scheduled for elective craniotomy or spinal tumor resection surgery.
• Age ≥ 18 years.
• Expected survival \> 3 months, and able/willing to comply with postoperative follow-up.
• Voluntary participation and provision of written informed consent.

You may not qualify if:

• Presence of active systemic or local surgical site infection before surgery.
• Use of therapeutic antibiotics for any reason within 72 hours prior to surgery.
• Concurrent severe immunosuppressive conditions, or chronic use of high-dose immunosuppressants.
• Pregnancy or lactation.
• Known contraindications or severe allergy to the antibiotics planned for use in the study.
• Any other condition that, in the investigator's judgment, may affect study participation or outcome assessment.

Sponsors & Collaborators

• Ming Yang: lead

Study Sites (1)

National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College

Beijing, PUMC, 10010, China

RECRUITING

MeSH Terms

Conditions

Surgical Wound Infection; Central Nervous System Neoplasms

Condition Hierarchy (Ancestors)

Wound Infection; Infections; Postoperative Complications; Pathologic Processes; Pathological Conditions, Signs and Symptoms; Nervous System Neoplasms; Neoplasms by Site; Neoplasms; Nervous System Diseases

Central Study Contacts

Ming Yang, MD

CONTACT

+8613810655237; yangming@cicams.ac.cn

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: NON RANDOMIZED
• Masking: NONE
• Purpose: DIAGNOSTIC
• Intervention Model: PARALLEL
• Sponsor Type: OTHER
• Responsible Party: SPONSOR INVESTIGATOR
• PI Title: Clinical Professor

Model Details: A prospective, single-center clinical validation study aimed at evaluating the effectiveness of a machine learning (LightGBM) model-based risk stratification intervention strategy in reducing the incidence of surgical site infection (SSI) in patients after central nervous system tumor surgery. The study plans to enroll 500 patients undergoing elective surgery. SSI risk will be calculated within 72 hours postoperatively using the model, which is based on 8 clinical features. The high-risk group (predicted probability ≥50%) will receive individualized enhanced interventions (e.g., antibiotic escalation, nutritional support), while the low-risk group (predicted probability \<50%) will receive standard care. The primary endpoint is the incidence of SSI within 90 days postoperatively (NHSN criteria), and secondary endpoints include model performance, hospital stay length, medical costs, etc.

Study Record Dates

• First Submitted: January 22, 2026
• First Posted: January 30, 2026
• Study Start: June 1, 2025
• Primary Completion (Estimated): October 30, 2029
• Study Completion (Estimated): October 30, 2029
• Last Updated: January 30, 2026

Record last verified: 2026-01

Data Sharing

• IPD Sharing: Will share
• Shared Documents: ANALYTIC CODE
• Time Frame: The de-identified IPD and supporting documents will become available no later than 6 months after the publication of the primary study findings in a peer-reviewed journal. The data will be accessible for a minimum of 5 years thereafter.
• Access Criteria: Access to IPD will be granted to qualified researchers for legitimate scientific research purposes. Requestors must submit a methodologically sound research proposal to the study Principal Investigator (PI) for review by a designated scientific committee. Proposals will be evaluated for scientific merit, feasibility, and ethical compliance. Data requestors will be required to sign a data use agreement (DUA) prohibiting re-identification attempts, unauthorized redistribution, and commercial use.

Locations

CN (1)