Research on Risk Assessment and Early Warning Models for Adverse Clinical Outcomes in Critically Ill Patients

Chongqing Medical University55,940 enrolled

Overview

This is a medical research study that uses information from past patient hospital records. It focuses on three serious conditions that often affect critically ill patients: sepsis (a life-threatening body-wide infection), ARDS (a severe lung injury that makes breathing very difficult), and acute kidney injury (sudden loss of kidney function). The goal is to better understand which patients in the ICU are at highest risk of developing these conditions or getting worse. Researchers will look at de-identified information from medical records of patients treated in the ICU . The study will use computer analysis to find patterns in the data that may help doctors predict these risks earlier. No new treatments are being tested, and no patients will be contacted or recruited for this study. All data used is anonymous to protect patient privacy.

Study Type

OBSERVATIONAL

Enrollment

55,940

Conditions

Sepsis ARDS (Acute Respiratory Distress Syndrome)AKI - Acute Kidney Injury

Interventions

This is a non-interventional, observational study. The aim is to develop and validate a predictive model using existing clinical data. No medical interventions (such as drugs, devices, or procedures) are being administered, assigned, or compared as part of this research protocol. The "intervention" of interest is the application of the predictive model for risk assessment, which is an analytical procedure, not a patient-directed intervention.

Eligibility

Sex: ALLMin age: 18 Years

Medical Language ↔ Plain English

Inclusion Criteria: * Adult patients (age ≥ 18 years). * Admitted to the ICU with a length of stay ≥ 24 hours. * Availability of key clinical variables within the first 24 hours of ICU admission (e.g., vital signs, laboratory results, admission diagnosis). Exclusion Criteria: * Patients with incomplete or missing key data for model variables (e.g., missing baseline creatinine, or missing Sequential Organ Failure Assessment (SOFA) score components). * Patients admitted for palliative care or comfort measures only upon ICU admission. * Readmissions during the same hospitalization (only the first ICU admission will be included).

Outcomes

Primary Outcomes

Area Under the Receiver Operating Characteristic Curve (AUROC) for predicting the composite outcome of Sepsis, ARDS, or Acute Kidney Injury

The discriminatory power of the machine learning model will be assessed by the AUROC. The value ranges from 0 to 1, with a higher value indicating better ability to distinguish between patients who will and will not experience the composite outcome.

Time frame: From ICU admission to 7 days after admission (for outcome prediction)

Calibration of predicted risk, measured by the Brier Score

The accuracy of the model's predicted probabilities will be assessed using the Brier Score (range 0 to 1, lower scores indicate better calibration). A calibration plot will be presented to visualize the agreement between predicted and observed event rates.

Time frame: From ICU admission to 7 days after admission (for outcome assessment).

Sensitivity (Recall) for the composite outcome at a pre-defined risk threshold

Performance metric calculated after applying a pre-defined probability cut-off to classify patients as high-risk or low-risk.