Prediction of Mortality and Hospitalization Risk Using Nutritional Indicators and Their Changes Over Time in a Large Prevalent Hemodialysis Cohort

Prediction of mortality in hemodialysis patients based on autoencoders

BACKGROUND

Patients with end-stage renal disease (ESRD) undergoing hemodialysis (HD) exhibit a high mortality risk, particularly at the onset of treatment. Conventional risk assessment models, dependent on extensive temporal data accumulation, frequently encounter issues of data incompleteness and lengthy collection periods.

OBJECTIVE

This study addresses the imbalance in short-term HD data and the issue of missing data features, achieving a robust assessment of mortality risk for HD patients over the subsequent 30 to 450 days.

METHODS

An autoencoder-based mortality prediction model for HD patients is proposed. Leveraging the manifold structure of the non-missing features and the intrinsic relationship between the features in the high-dimensional data space, the model infers the values of the missing features. Noise and redundant information typically distort the manifold structure, impacting the accuracy of inferences about missing features. Consequently, we generate feature dropping masks to simulate the missing data distribution in the deep learning framework and design an autoencoder, forming an adaptive feature extraction module. The module utilizes readily available short-term data for unsupervised learning, enabling the encoder to reconstruct missing features and derive latent representations. Finally, a classifier based on the latent representations achieves the mortality prediction.

RESULTS

Over a 30-day observation window, the model demonstrated superior mortality prediction performance compared to other models across all prediction windows. Feature importance analysis showed that creatinine and age are consistently the most critical features across all prediction windows. Glucose (fasting) and platelet count also remain significant, with their correlation with mortality risk increasing over time. Serum albumin, international standard ratio, and phosphate are particularly important for short-term predictions, while conjugated bilirubin and prothrombin time gain prominence in mid- and long-term predictions.

CONCLUSION

The proposed model proficiently leverages short-term HD data to provide precise mortality risk evaluations in HD patients, with particular efficacy in the short-term. Its application holds considerable value for clinical decision-making and risk management in this patient population.

Analyzing Monthly Blood Test Data to Forecast 30-Day Hospital Readmissions among Maintenance Hemodialysis Patients

Background: The increase in the global population of hemodialysis patients is linked to aging demographics and the prevalence of conditions such as arterial hypertension and diabetes mellitus. While previous research in hemodialysis has mainly focused on mortality predictions, there is a gap in studies targeting short-term hospitalization predictions using detailed, monthly blood test data. Methods: This study employs advanced data preprocessing and machine learning techniques to predict hospitalizations within a 30-day period among hemodialysis patients. Initial steps include employing K-Nearest Neighbor (KNN) imputation to address missing data and using the Synthesized Minority Oversampling Technique (SMOTE) to ensure data balance. The study then applies a Support Vector Machine (SVM) algorithm for the predictive analysis, with an additional enhancement through ensemble learning techniques, in order to improve prediction accuracy. Results: The application of SVM in predicting hospitalizations within a 30-day period among hemodialysis patients resulted in an impressive accuracy rate of 93%. This accuracy rate further improved to 96% upon incorporating ensemble learning methods, demonstrating the efficacy of the chosen machine learning approach in this context. Conclusions: This study highlights the potential of utilizing machine learning to predict hospital readmissions within a 30-day period among hemodialysis patients based on monthly blood test data. It represents a significant leap towards precision medicine and personalized healthcare for this patient group, suggesting a paradigm shift in patient care through the proactive identification of hospitalization risks.

Creatinine index: a retrospective cohort study in an urban Australian dialysis context

AIM

This study aimed to described the relationship between the CI and mortality in an Australian context.

INTRODUCTION

Maintenance haemodialysis is a catabolic state associated with a significant decrease in lean body mass (LBM) and protein energy wasting. LBM can be derived or estimated from creatinine kinetic modelling, specifically the creatinine index (CI). This has been demonstrated in cohort studies to predict mortality.

METHODS

One hundred seventy-nine patients undergoing haemodialysis in 2015 were included in this cohort. They were followed for 5 years with pertinent clinical data collected to calculate the CI as of December 2015. For analysis, patients were split into a high and low CI group based on the median (18.32 mg/kg/day). The primary outcome of interest was all-cause mortality, and secondary outcomes included myocardial infarction, stroke and transplantation.

RESULTS

During follow-up, 69 (76.7%) patients in the low CI group and 28 (31.5%) patients in the high CI group died (P < 0.001). The relative risk (RR) of mortality within the low compared with the high CI group was 2.43 (95% confidence interval, 1.75-3.38). Fully adjusted Cox proportional hazards modelling demonstrated a hazard ratio (HR) of 0.498 (95% CI, 0.292-0.848) for survival in the high CI group. Lower CI was associated with increased risk of stroke (RR, 5.43 [95% CI, 1.24-23.84]), whereas transplant was more likely in the high CI group (RR, 6.4 [95% confidence interval, 1.96-20.88]).

CONCLUSIONS

In a single-centre Australian haemodialysis cohort, the CI was strongly associated with mortality and stroke risk. The CI is an accurate and simple method to identify patients with low LBM at risk for significant morbidity and mortality.

Trajectories of Nutritional Parameters Before and After Prescribed Oral Nutritional Supplements: A Longitudinal Cohort Study of Patients With Chronic Kidney Disease Not Requiring Dialysis

Background:

The association between oral nutritional supplement use and nutritional parameters among patients with nondialysis chronic kidney disease (CKD-ND) with or at high risk of undernutrition/protein-energy wasting has not been previously studied. The definition of patient subgroups most likely to benefit from oral nutritional supplementation (ONS) is also an area where more research is needed.

Objective:

To assess nutritional parameter trajectories among patients with CKD-ND prescribed oral nutritional supplements in British Columbia, and to compare trajectories by nutritional phenotype.

Design:

Longitudinal cohort study, pre-post design.

Setting:

Multidisciplinary CKD clinics across British Columbia.

Patients:

A total of 3957 adult patients with CKD-ND, who entered multidisciplinary CKD clinics during 2010 to 2019, met criteria for oral nutritional supplement prescription based on dietitian assessment, and received ≥1 oral nutritional supplement prescription.

Measurements:

Longitudinal nutritional parameters, including body mass index (BMI), serum albumin, serum bicarbonate, serum phosphate, and neutrophil-to-lymphocyte ratio (NLR).

Methods:

Using linear mixed models, slopes for nutritional and inflammation parameters were assessed in the 2-year periods before and after the first oral nutritional supplement prescription. Hierarchical cluster analysis was applied to identify nutritional phenotypes using baseline data, and slope analysis was repeated by cluster.

Results:

In the pre-oral-nutritional-supplement period, declines in BMI (−0.87 kg/m2/year, 95% confidence interval [CI]: −0.99 to −0.75), albumin (−1.11 g/L/year, 95% CI: −1.27 to −0.95), and bicarbonate (−0.49 mmol/L/year; 95% CI: −0.59 to −0.39), and increases in NLR (+0.79/year; 95% CI: 0.60 to 0.98) and phosphate (+0.05 mmol/L/year; 95% CI: 0.04 to 0.06) were observed. Following oral nutritional supplement prescription, there were statistically significant increases in BMI slope (+0.91 kg/m2/year, P < .0001), albumin slope (+0.82 g/L/year, P < .0001), and phosphate slope (+0.02 mmol/L/year, P = .005), as well as a decline in NLR slope of −0.55/year ( P < .0001). There was no significant change in bicarbonate slope. Cluster analysis identified 5 distinct phenotypes. The cluster with the highest mean baseline NLR and lowest mean BMI demonstrated the greatest number of improvements in nutritional parameter slopes in the post-oral-nutritional-supplement period.

Limitations:

Possibility of residual confounding. Data on dietary intake, muscle mass, and nutritional scoring systems were not available in the registry.

Conclusions:

Among patients with CKD-ND prescribed oral nutritional supplements, there were improvements in nutrition/inflammation parameters over time following the first ONS prescription. The heterogeneity in response to ONS by cluster subgroup suggests an individualized approach to nutritional management may be beneficial.

Clinical and predictive value of simplified creatinine index used as muscle mass surrogate in end-stage kidney disease haemodialysis patients-results from the international MONitoring Dialysis Outcome initiative

Background

Protein-energy wasting, muscle mass (MM) loss and sarcopenia are highly prevalent and associated with poor outcome in haemodialysis (HD) patients. Monitoring of MM and/or muscle metabolism in HD patients is of paramount importance for timely detection of muscle loss and to intervene adequately. In this study we assessed the reliability and reproducibility of a simplified creatinine index (SCI) as a surrogate marker of MM and explored its predictive value on outcome.

Method

We included all in-centre HD patients from 16 European countries with at least one SCI. The baseline period was defined as 30 days before and after the first multifrequency bioimpedance spectroscopy measurement; the subsequent 7 years constituted the follow-up. SCI was calculated by the Canaud equation. Multivariate Cox proportional hazards models were applied to assess the association of SCI with all-cause mortality. Using backward analysis, we explored the trends of SCI before death. Bland–Altman analysis was performed to analyse the agreement between estimated and measured MM.

Results

We included 23 495 HD patients; 3662 were incident. Females and older patients have lower baseline SCI. Higher SCI was associated with a lower risk of mortality [hazard ratio 0.81 (95% confidence interval 0.79–0.82)]. SCI decline accelerated ∼5–7 months before death. Lean tissue index (LTI) estimated by SCI was correlated with measured LTI in both sexes (males: R² = 0.94; females: R² = 0.92; both P < 0.001). Bland–Altman analysis showed that measured LTI was 4.71 kg/m² (±2 SD: −12.54–3.12) lower than estimated LTI.

Conclusion

SCI is a simple, easily obtainable and clinically relevant surrogate marker of MM in HD patients.