Abstract:Acute kidney injury (AKI) is a serious clinical condition that affects up to 20% of hospitalised patients. AKI is associated with short term unplanned hospital readmission and post-discharge mortality risk. Patient risk and healthcare expenditures can be minimised by followup planning grounded on predictive models and machine learning. Since AKI is multi-factorial, predictive models specialised in different categories of patients can increase accuracy of predictions. In the present article we present some results following this approach.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **How to improve the accuracy of predicting the mortality rate within 90 days after discharge for patients with acute kidney injury (AKI) through patient stratification, while ensuring the fairness, transparency, and patient - centered performance optimization of the prediction model**. ### Problem Background Acute kidney injury (AKI) is a serious clinical condition, affecting 10% - 20% of hospitalized patients. AKI is associated with the following problems: - An increased short - and medium - term risk of death after discharge. - Prolonged hospital stays. - Increased medical expenditures. - Unplanned readmissions in the short term. - An increased probability of progressing to a chronic disease in the medium term. ### Research Motivation In order to reduce patients' readmission risks and medical expenditures, researchers hope to optimize the follow - up plans after discharge through prediction models. Since AKI is caused by multiple factors, developing specialized prediction models for different categories of patients can improve the accuracy of prediction. ### Main Research Contents This study proposes an automated method based on machine learning for stratifying AKI patients and constructing prediction models for the mortality rate within 90 days for each stratified group. The specific steps include: 1. **Initial Stratification**: Use the training data set to initially group patients. 2. **Prediction Model Construction**: Construct generalized additive model (GAM) predictors for each group. 3. **Patient Allocation**: Allocate new patients to existing groups according to heuristic methods and evaluate the accuracy of the prediction model. 4. **Optimized Grouping**: Gradually optimize the grouping through random small perturbations to maximize the sum of the AUROC (area under the receiver operating characteristic curve) of the prediction model. 5. **Final Evaluation**: Use the test data set to evaluate the accuracy and robustness of the final prediction model. ### Key Contributions - **Ethical Compliance**: Ensure the fairness, transparency, and patient - centered performance optimization of the prediction model. - **Stratification Method**: Achieve patient stratification through the Clustering Aware Classification heuristic method. - **Evaluation Indicator**: Use AUROC as the main indicator for evaluating the accuracy of the prediction model. Through this method, researchers hope to be able to provide more accurate and reliable prediction tools in practical applications, helping clinicians develop better follow - up plans, thereby improving the prognosis of AKI patients.

Population stratification for prediction of mortality in post-AKI patients

Using artificial intelligence to predict mortality in AKI patients: a systematic review/meta-analysis

Prediction of Mortality Risk after Ischemic Acute Kidney Injury with a Novel Prognostic Model: A Multivariable Prediction Model Development and Validation Study

Identification of risk factors and establishment of prediction models for mortality risk in patients with acute kidney injury: A retrospective cohort study

Predicting acute kidney injury at hospital re-entry using high-dimensional electronic health record data

Predictive value of machine learning for the risk of acute kidney injury (AKI) in hospital intensive care units (ICU) patients: a systematic review and meta-analysis

Machine learning-based risk prediction of acute kidney disease and hospital mortality in older patients

The prediction of in-hospital mortality in elderly patients with sepsis-associated acute kidney injury utilizing machine learning models

Predicting long-term mortality of patients with postoperative acute kidney injury following noncardiac general anesthesia surgery using machine learning

Development of mortality prediction model in the elderly hospitalized AKI patients

Development of interpretable machine learning models for prediction of acute kidney injury after noncardiac surgery: a retrospective cohort study

Application of Machine Learning Algorithms to Predict Acute Kidney Injury in Elderly Orthopedic Postoperative Patients

Personalized prediction of mortality in patients with acute ischemic stroke using explainable artificial intelligence

Prediction of persistent acute kidney injury in postoperative intensive care unit patients using integrated machine learning: a retrospective cohort study

Forecasting acute kidney injury and resource utilization in ICU patients using longitudinal, multimodal models

Predicting in-hospital outcomes of patients with acute kidney injury

Predicting renal function recovery and short-term reversibility among acute kidney injury patients in the ICU: comparison of machine learning methods and conventional regression

Acute kidney injury prediction for non-critical care patients: a retrospective external and internal validation study

Predicting In-Hospital Mortality of Patients with Acute Kidney Injury in the ICU Using Random Forest Model.

The Development of a Machine Learning Inpatient Acute Kidney Injury Prediction Model*