Gastrointestinal surgeries are crucial for treating digestive tract diseases. However, they often lead to high postoperative infection rates due to long durations, bacterial translocation, weakened immunity, and reduced intestinal function post-surgery. This not only impacts surgical outcomes and patient recovery but also extends hospital stays and increases financial burdens. In severe instances, it can even result in sepsis and death. The mortality risk for emergency gastrointestinal surgeries ranges from 15% to 25%.

Existing scoring systems like APACHE - Ⅱ and SOFA, designed mainly for ICU patients, are insufficient for predicting the death risk of emergency gastrointestinal surgery patients. Some machine learning models for common gastric and colorectal cancer patients lack independent prospective validation.

To overcome these limitations, this study at the First Hospital of Jilin University aims to construct and validate a model predicting all - cause mortality within 30 days post - emergency gastrointestinal surgery.

The study proceeds in two phases. The retrospective development phase examines patients who underwent emergency gastrointestinal surgery from July 2019 to July 2024. Data is collected via the electronic medical record system, and eligible patients form the development cohort for model building. The prospective validation phase, planned to last 5 months for patient enrollment and 30 days for follow - up, is part of a half - year study.

Inclusion criteria involve patients over 18, undergoing emergency gastrointestinal surgery (ICD - 10), and meeting specific critical conditions post - surgery. Exclusion criteria include superficial surgeries, significant data loss, intraoperative death, multiple injuries, and participation in other studies.

Sample size calculation, based on methods by Harrell et al. and Peduzzi et al., requires at least 80 patients with events. With a 15% event incidence, the training set needs about 534 cases, the validation set 229, for a total of 763 cases (7:3 ratio). An additional 100 cases are for external validation.

Investigated factors include demographics, medical history, preoperative, intraoperative, and postoperative indicators, plus pathology. The primary endpoint is 30 - day all - cause mortality, and the secondary is 30 - day postoperative complications, assessed by Clavien - Dindo classification.

Data management involves CRC and double - entry. Analysis uses SPSS 25.0 and R 4.0.2. Bias is controlled through surgeon screening and surgical quality evaluation. The study has ethical approval, and patients provide informed consent.

This research aims to offer clinicians a reliable model for early high - risk patient identification and precise interventions, ultimately enhancing patient outcomes.

1. Research Background Gastrointestinal surgeries play a fundamental role in the treatment of digestive tract disorders. Nevertheless, due to their extended durations, the presence of diverse bacteria in the intestine, which can lead to bacterial translocation during surgery, coupled with decreased immune function and reduced intestinal function post-surgery, the incidence of postoperative infections is relatively high. This not only compromises the effectiveness of the surgery and patient recovery but also lengthens hospital stays and escalates economic burdens. In severe cases, sepsis may occur, potentially resulting in death. Previous studies have indicated that the mortality risk associated with emergency gastrointestinal surgeries can range from 15% to 25%.

   Although widely used scoring systems such as APACHE - Ⅱ and SOFA are employed to assess the risk and predict the prognosis of critically ill patients, they are predominantly designed for patients in the Intensive Care Unit (ICU). As a result, they do not adequately account for the unique characteristics of critically ill patients after emergency gastrointestinal surgeries, such as the absence of surgical - related factors, and thus lack disease - specific predictive power for death risk. Additionally, while some machine learning models have been developed to predict the death risk of patients with common gastric and colorectal cancers, these models lack independent prospective validation, limiting their clinical applicability.

2. Research Objectives 2.1 Primary Objective The main aim of this study is to construct a prediction model for all - cause mortality within 30 days after emergency gastrointestinal surgery in critically ill patients. By leveraging advanced machine learning algorithms, we seek to identify key factors associated with mortality and develop a reliable model that can accurately predict the likelihood of death within this critical time frame.

   2.2 Secondary Objective Based on the mortality predictions generated by the model, we aim to quantify the severity of patients' diseases. This will enable clinicians to identify high - risk patients at an early stage and implement timely and targeted interventions. By doing so, we hope to improve patient outcomes, reduce mortality rates, and enhance the overall quality of care provided to these patients.

3. Research Design 3.1 Research Setting and Participants The study will be conducted at the First Hospital of Jilin University and is divided into two distinct phases.

In the development phase, a retrospective study will be carried out on critically ill patients who underwent emergency gastrointestinal surgery between July 2019 and July 2024. Clinical data relevant to patient deaths will be systematically collected from the hospital's electronic medical record system. Patients who meet the inclusion criteria but do not meet the exclusion criteria will be enrolled in the development cohort. These patients' data will be used to establish the initial prediction model.

The inclusion criteria for patients are as follows: they must be over 18 years old, have undergone emergency gastrointestinal surgery as defined by the tenth revision of the International Classification of Diseases (ICD - 10), and meet any one of the following critical patient diagnostic criteria after surgery: requiring ventilator assistance at the end of the operation, needing a norepinephrine infusion dose greater than 1.0 μg/kg/min or even in combination with other vasopressor drugs to maintain blood pressure, experiencing severe arrhythmia, or having failure of other organ functions.

The exclusion criteria include patients with superficial surgical sites that do not involve the abdominal cavity, those with more than 30% of data missing, patients who died during the operation or had treatment withdrawn, those with multiple injuries, and patients participating in other clinical studies.

In the validation phase, a prospective single - center observational study will be conducted. The patient enrollment period is planned to last for 5 months, and the postoperative follow - up period for each patient will be 30 days. The entire study, from its inception to the completion of data analysis, is expected to span half a year.

3.2 Sample Size Determination The sample size for this study will be calculated using the methods proposed by Harrell et al. and Peduzzi et al. Considering approximately 8 key clinical factors, it is estimated that at least 80 patients with events (i.e., 30 - day all - cause mortality) are required. Based on a projected event incidence rate of 15%, the training set will require approximately 534 cases, and the validation set will need 229 cases. With a 7:3 ratio between the modeling group and the validation group, the total sample size amounts to 763 cases. Additionally, an external validation cohort of approximately 100 cases is planned to further assess the generalizability of the model.

3.3 Data Collection 3.3.1 Exposure Factors Demographic Characteristics: Information on patients' gender and age will be collected. These factors can influence the body's physiological responses and recovery capabilities, potentially affecting the risk of mortality.

Medical History: Details of patients' past medical conditions, including hypertension, diabetes, cardiovascular diseases, cerebrovascular diseases, previous 颅脑 or spinal surgeries, kidney diseases, immune system disorders, atrial fibrillation, chronic obstructive pulmonary disease (COPD), and other relevant conditions, will be recorded. Additionally, lifestyle risk factors such as smoking and alcohol consumption will be documented.

Preoperative Indicators: This includes a comprehensive set of laboratory test results, such as hematocrit, hemoglobin, platelet count, white blood cell count, international normalized ratio, prothrombin time, activated partial thromboplastin time, arterial blood pH, partial pressure of oxygen, partial pressure of carbon dioxide, lactate levels, aspartate aminotransferase, total bilirubin, albumin, serum sodium, serum potassium, high - sensitivity C - reactive protein, procalcitonin, brain natriuretic peptide (BNP), troponin, D - dimer, creatinine, and blood urea nitrogen. Other preoperative data, such as the American Society of Anesthesiologists (ASA) physical status classification, body temperature, heart rate, respiratory rate, mean arterial pressure, and oxygenation index, will also be collected.

Intraoperative Indicators: Data such as the duration of the surgery, the volume of intraoperative fluid infusion, the volume of intraoperative red blood cell transfusion, intraoperative urine output, intraoperative blood loss, intraoperative mean arterial pressure, and the presence or absence of abdominal infection will be recorded. These factors can directly impact the patient's physiological state during and after the surgery.

Postoperative Indicators: Similar to the preoperative laboratory tests, a series of postoperative laboratory results will be collected. Additionally, information on whether blood products were transfused, the ASA classification, and vital signs such as body temperature, heart rate, respiratory rate, mean arterial pressure, and oxygenation index will be documented.

Postoperative Pathology: Details of the postoperative pathology, including tumor length, tumor differentiation, pT, pN, pM, and pTNM staging, as well as the number of cancer nodules, will be obtained. These pathological features can provide crucial information about the severity of the disease and the patient's prognosis.

3.3.2 Outcome Measures Primary Endpoint: The primary endpoint of this study is all - cause death within 30 days after surgery. All - cause death refers to death resulting from any cause, providing a comprehensive measure of the patient's outcome.

Secondary Endpoint: The secondary endpoint is the complication status within 30 days after surgery. Complications will be evaluated using the Clavien - Dindo classification system, and complications of grade Ⅱ and above will be recorded and analyzed. These may include, but are not limited to, intra - abdominal hemorrhage, gastrointestinal bleeding, anastomotic leakage, chylous fistula, surgical site infection (including intra - abdominal and incision infections), intestinal obstruction, postoperative diarrhea, pulmonary infection, urinary tract infection, cardiovascular accidents, cerebrovascular accidents, and thrombotic diseases.

3.4 Data Management and Statistical Analysis 3.4.1 Data Management Clinical research coordinators (CRC) will be responsible for screening patients who underwent emergency gastrointestinal surgery between July 2019 and May 2024. Data will be collected from the electronic medical record system, and patients who meet the inclusion criteria but not the exclusion criteria will be recorded. For patients who die during hospitalization, the data will be directly recorded. Data on complications within 30 days after surgery and deaths after discharge will be obtained through follow - up.

During each surgical procedure, the name of the operating surgeon will be recorded in the medical record. All patients who meet the inclusion criteria, including those who refused to participate, were excluded due to exclusion criteria, or were later removed due to exclusion criteria, will be documented, along with the reasons for their exclusion.

Data collection will be carried out daily by CRC, who will verify and record patients' enrollment, withdrawal, achievement of the primary endpoint, and measurement of the secondary endpoint. Two postgraduate students will serve as data entry operators, and they, together with CRC, will collect baseline data. A double - entry data entry system will be employed, where one person enters the data, and the other verifies its accuracy. Both paper - based and electronic case report forms (CRFs) will be used. Baseline data will be entered on the day of the surgery. For patients who die during hospitalization, the primary endpoint data will be entered before their death. Pathological and other hospital information will be entered within one week after the patient's discharge. Data on complications within 30 days after surgery and deaths after discharge will be entered within 30 days after the surgery.

3.4.2 Statistical Analysis A database will be established, and statistical analysis will be performed using SPSS 25.0 and R 4.0.2 software. For continuous data that follow a normal distribution, the mean ± standard deviation (x ± s) will be used for description, and the t - test will be employed for group comparisons. For data that do not follow a normal distribution, the median and interquartile range [M(P25 - P75)] will be used, and the Mann - Whitney U test will be used for group comparisons. Categorical data will be presented as percentages, and the chi - square test will be used for group comparisons. Missing data will be imputed using the Monte Carlo multiple imputation method.

First, univariate analysis will be conducted to identify factors that are significantly associated with the outcome. Then, variables that show statistical significance in the univariate analysis will be included in the Least Absolute Shrinkage and Selection Operator (LASSO) regression to select the final variables for model building. The performance of the model will be evaluated using several metrics, including the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, Youden index, and F1 score. All statistical tests will be two - sided, and statistical significance will be set at P < 0.05.

3.5 Bias Control and Quality Management To minimize bias, strict criteria will be set for the selection of surgeons. Emergency physicians participating in the study must be proficient in emergency gastrointestinal surgery techniques, having completed at least 50 such surgeries. A trial steering committee, composed of surgeons with over 10 years of experience in gastrointestinal surgeries, will be established. Before a surgeon is allowed to enroll patients, they will be required to provide videos of their recent three consecutive emergency gastrointestinal surgeries. The committee will use the Delphi method to evaluate the surgical quality and determine whether the surgeon meets the enrollment criteria.

During the clinical trial, the surgical quality will be evaluated through videos and photos. Surgeons will be required to obtain images of the pelvic exploration during the surgery. If it is not possible to obtain such imaging data, the reasons for the inability to do so must be documented.

Surgeons will be required to record the surgical process in a surgical notebook provided by the research center and explain any deviations from the standardized surgical procedures. The trial steering committee will review the enrollment status and the quality of the enrolled cases monthly to ensure the quantity and quality of cases enrolled by each surgeon.

To ensure the accurate collection of the primary endpoint and predictive factors, a head nurse will be appointed as the clinical research coordinator (CRC). The CRC will daily verify patients' enrollment, withdrawal, and achievement of the primary endpoint. Two postgraduate students, together with the CRC, will promptly enter the data into the electronic case report form (ECRF).

3.6 Safety Evaluation Before the start of the study, the baseline physical signs and symptoms of the participants, including body temperature, pulse rate, respiratory rate, and blood pressure, will be measured. Preoperative hematological and imaging examinations will be completed to ensure that the participants meet the inclusion and exclusion criteria.

After the surgery, continuous variables will be used to describe the postoperative measurements of each participant and their changes relative to the preoperative baseline. Laboratory indicators will be classified as low, normal, or high based on the normal reference range. The changes from normal or high baseline values to low values after the surgery, and from normal or low baseline values to high values after the surgery, will be recorded. The lowest or highest observed values after the surgery will be used to calculate the post - baseline low or high values. Physicians will evaluate these abnormal indicators to determine whether they have clinical significance, and the proportion of participants with "abnormal and clinically significant" changes will be reported.

Physical examinations, including general examinations, head, neck, chest, lung, heart, blood vessel, abdomen, spine, limb, and nervous system examinations, as well as vital signs such as blood pressure, heart rate, and body temperature, will be performed. The measurements of physical examinations, vital signs, and body weight before and after treatment, as well as their changes relative to the baseline, will be statistically described. Any vital signs that exceed the normal range or show abnormal changes will be documented.

3.7 Ethical Considerations Before the commencement of the study, ethical approval will be obtained from the Ethics Committee of the First Hospital of Jilin University. Eligible patients will be informed face - to - face by their treating surgeons and will be required to sign an informed consent form. The informed consent process will be conducted by attending physicians or above.

The information provided to patients in the informed consent includes: a statement about the research nature of the trial; a comprehensive and fair explanation of the procedures to be followed; a detailed description of the nature, expected duration, and purpose of the study; a description of any foreseeable risks or discomforts; a description of any potential benefits; a statement regarding the careful handling and confidentiality of patient data, as well as the data retention period of 15 years; a statement that patient biological materials will be stored for 15 years; and a statement about voluntary participation and withdrawal, clarifying that refusal to participate will not result in any punishment or loss of benefits to which the patient is entitled, and that the patient can withdraw from the study at any time without penalty or loss of benefits. In the event of withdrawal, the patient will receive standard treatment of the same level of care.

Inclusion Criteria:

• Age > 18 years old; Underwent emergency gastrointestinal surgery [referring to the tenth edition of the International Classification of Diseases (ICD - 10)]; Met any of the following criteria for critically ill patients after surgery: those who still required ventilator - assisted ventilation after the surgery; those who needed norepinephrine infusion at a dose > 1.0 μg/kg/min and even required combination with other vasopressor drugs to maintain blood pressure; those with severe arrhythmia; those with combined failure of other organ functions.

Exclusion Criteria:

• The surgical site is superficial and does not involve the abdominal cavity. Cases with data missing by more than 30%. Patients who died during the operation, or those for whom treatment was terminated or abandoned.

Cases complicated with multiple injuries. Patients participating in other clinical studies.

emergency gastrointestinal surgery patients

The IPD in this study will not be shared mainly due to the following reasons. First, there are significant data privacy risks. The IPD contains highly sensitive personal health information, such as genetic data, detailed mental health records, and comprehensive medical histories of patients. Despite anonymization efforts, there remains an inherent risk of re - identification. A data breach could lead to severe harm to patients' privacy and a crisis of trust.