AI-Driven Prediction of Dialysis Outcome With EHR

Predicting Clinical Outcomes in Dialysis Patients Using Electronic Health Records: An AI-Based Approach

This is a multi-center, clinical study designed to evaluate the application and effectiveness of an AI-assisted predictive model for outcome of dialysis patients, leveraging multimodal health data.

Study Overview

• Status: Recruiting
• Conditions: Dialysis Patients
• Intervention / Treatment: Other: AI-assisted Predictive Model for Dialysis Outcomes

Detailed Description

This study aims to develop an AI-assisted model to predict clinical outcomes in dialysis patients, focusing on both primary outcomes (e.g., mortality) and intermediate outcomes (e.g., anemia, blood pressure, nutritional status, and calcium-phosphate metabolism). The study will utilize patients' EHR data, including laboratory test results, medical history, dialysis treatment information, and clinical observations, to predict these health outcomes. The goal is to improve early identification of at-risk patients, enabling better clinical decision-making and personalized care strategies.

• Study Type: Observational
• Enrollment (Estimated): 1000000

Contacts and Locations

• Study Contact: Name: Fei Liu, MD; Phone Number: +86 13810512704; Email: liufei_2359@163.com

Study Locations

• China
  • Beijing
    • Beijing, Beijing, China
      • Recruiting
      • General Hospital of PLA
      • Contact: Delong Zhao; Phone Number: +86 13810512704; Email: feiliu0108@gmail.com

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

The study population consists of dialysis patients from the China Hemodialysis National Network Center, which includes a wide range of patients undergoing hemodialysis treatment at participating hospitals across China. Participants will be selected based on the availability of comprehensive electronic health records (EHR), including medical history, laboratory test results, dialysis treatment details, and clinical observations. The cohort will include both male and female patients, with varying degrees of health status, including those with comorbidities commonly associated with dialysis. The study aims to utilize this diverse group to assess and predict outcomes related to mortality and complications in dialysis patients.

Description

Inclusion Criteria:

1. Patients who have been undergoing dialysis (either hemodialysis or peritoneal dialysis) for at least 3 months.
2. Complete and accessible EHR data, including medical history, laboratory test results, dialysis treatment details, and clinical observations.
3. Participants must provide informed consent for the use of their health data for research purposes.

Exclusion Criteria:

1. Patients with incomplete or missing critical EHR data, including medical history, laboratory results, dialysis data, or treatment details necessary for the study.
2. Patients who have been on dialysis for less than 3 months, to ensure stable data for outcome prediction.

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
High Risk Group; Participants predicted to have a high risk of mortality based on AI-assisted prediction models using their EHR data, including medical history, lab results, dialysis treatment details, and clinical observations.	Other: AI-assisted Predictive Model for Dialysis Outcomes; This study utilizes an AI-assisted predictive model that analyzes multimodal data from electronic health records, including medical history, laboratory results, dialysis treatment details, and clinical observations, to predict outcomes for dialysis patients. The model employs deep learning algorithms to predict mortality risk, intermediate outcomes such as anemia, blood pressure control, nutrition, and calcium-phosphate metabolism, and helps identify early signs of deterioration. The intervention is not a direct treatment or procedure but aims to develop a tool for predicting patient outcomes and optimizing treatment strategies to improve overall health and survival rates for dialysis patients.
Low Risk Group; Participants predicted to have a low risk of mortality based on the AI-assisted prediction model, who will be compared with the high-risk group for evaluating the effectiveness of early intervention strategies.	Other: AI-assisted Predictive Model for Dialysis Outcomes; This study utilizes an AI-assisted predictive model that analyzes multimodal data from electronic health records, including medical history, laboratory results, dialysis treatment details, and clinical observations, to predict outcomes for dialysis patients. The model employs deep learning algorithms to predict mortality risk, intermediate outcomes such as anemia, blood pressure control, nutrition, and calcium-phosphate metabolism, and helps identify early signs of deterioration. The intervention is not a direct treatment or procedure but aims to develop a tool for predicting patient outcomes and optimizing treatment strategies to improve overall health and survival rates for dialysis patients.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Mortality Prediction Accuracy	The ability of the AI-assisted predictive model to accurately predict the risk of mortality in dialysis patients. Prediction accuracy will be assessed using the Area Under the Curve (AUC), F1 score, and sensitivity/specificity. The model will be evaluated by comparing the predicted mortality risk with actual outcomes (i.e., whether patients survived or passed away during the study period).	1 year

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Complications Prediction Accuracy	The accuracy of the AI-assisted predictive model in forecasting complications commonly experienced by dialysis patients, including anemia, uncontrolled blood pressure, poor nutritional status, and abnormalities in calcium-phosphate metabolism. The model's performance will be assessed using metrics such as AUC, F1 score, and accuracy by comparing predicted values to actual clinical outcomes, such as lab results, clinical diagnoses, and patient health status.	1 year

Collaborators and Investigators

• Sponsor: The Eye Hospital of Wenzhou Medical University

Study record dates

Study Major Dates

• Study Start (Actual): January 1, 2023
• Primary Completion (Estimated): May 1, 2025
• Study Completion (Estimated): May 1, 2025

Study Registration Dates

• First Submitted: January 19, 2025
• First Submitted That Met QC Criteria: January 19, 2025
• First Posted (Actual): January 24, 2025

Study Record Updates

• Last Update Posted (Actual): April 17, 2025
• Last Update Submitted That Met QC Criteria: April 16, 2025
• Last Verified: April 1, 2025

More Information

Terms related to this study

• Keywords: Dialysis Patients; AI-Assisted Prediction
• Other Study ID Numbers: Dialysis

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No