Predicting dry weight change in Hemodialysis patients using machine learning

BACKGROUND

Machine Learning has been increasingly used in the medical field, including managing patients undergoing hemodialysis. The random forest classifier is a Machine Learning method that can generate high accuracy and interpretability in the data analysis of various diseases. We attempted to apply Machine Learning to adjust dry weight, the appropriate volume status of patients undergoing hemodialysis, which requires a complex decision-making process considering multiple indicators and the patient's physical conditions.

METHODS

All medical data and 69,375 dialysis records of 314 Asian patients undergoing hemodialysis at a single dialysis center in Japan between July 2018 and April 2020 were collected from the electronic medical record system. Using the random forest classifier, we developed models to predict the probabilities of adjusting the dry weight at each dialysis session.

RESULTS

The areas under the receiver-operating-characteristic curves of the models for adjusting the dry weight upward and downward were 0.70 and 0.74, respectively. The average probability of upward adjustment of the dry weight had sharp a peak around the actual change over time, while the average probability of downward adjustment of the dry weight formed a gradual peak. Feature importance analysis revealed that median blood pressure decline was a strong predictor for adjusting the dry weight upward. In contrast, elevated serum levels of C-reactive protein and hypoalbuminemia were important indicators for adjusting the dry weight downward.

CONCLUSIONS

The random forest classifier should provide a helpful guide to predict the optimal changes to the dry weight with relative accuracy and may be useful in clinical practice.

Hepatic phosphate uptake and subsequent nerve-mediated phosphaturia are crucial for phosphate homeostasis following portal vein passage of phosphate in rats

Fibroblast growth factor 23, parathyroid hormone, and 1,25-dihydroxyvitamin D are critical in phosphate homeostasis. Despite these factors' importance, regulators of phosphaturia in the acute postprandial phase remain largely unknown. This study investigated the mechanism of acute phosphate regulation in the postprandial phase in rats. Duodenal administration of radiolabeled phosphate (³²P) showed that ³²P levels in the inferior vena cava (IVC) blood were lower than those in the portal vein (PV) blood. Serum phosphate concentration transiently increased 5 min after phosphate solution administration through IVC, while it was maintained after the administration through PV. Phosphate administration through both IVC and PV resulted in increased fractional excretion of phosphate (FEPi) at 10 min without elevation of the known circulating factors, but urinary phosphate excretion during the period was 8% of the dose. Experiments using ³²P or partial hepatectomy showed that the liver was one of the phosphate reservoirs. The elevation of FEPi and suppression of sodium-phosphate cotransporter 2a in the kidney at 10 min was attenuated in rats with SCH23390, hepatic denervation, or renal denervation, thus indicating that the liver communicated with the kidney via the nervous system to promote phosphaturia. These results revealed previously unknown mechanisms for serum phosphate maintenance.

Paratubular basement membrane insudative lesions predict renal prognosis in patients with type 2 diabetes and biopsy-proven diabetic nephropathy

Aims

Glomerular insudative lesions are a pathological hallmark of diabetic nephropathy (DN). However, paratubular basement membrane insudative lesions (PTBMIL) have not attracted much attention, and the association between such lesions and the renal prognosis remains unclear.

Methods

Among 142 patients with biopsy-proven DN and type 2 diabetes encountered from 1998 to 2011, 136 patients were enrolled in this study. Patients were classified into 3 groups (Group 1: mild, Group 2: moderate, Group 3: severe) according to the extent of cortical and medullary PTBMIL. The endpoint was a decline of the estimated glomerular filtration rate (eGFR) by ≥ 40% from baseline or commencement of dialysis for end-stage renal disease. The Cox proportional hazard model was employed to calculate hazard ratios (HRs) and 95% confidence interval (CIs) for the death-censored endpoint.

Results

During a median follow-up period of 1.8 years (IQR: 0.9–3.5), the endpoint occurred in 104 patients. Baseline mean eGFR was 43.9 ± 22.8 ml/min/1.73 m², and 125 patients (92%) had overt proteinuria. After adjusting for known indicators of DN progression, the HR for the endpoint was 2.32 (95% CI: 1.20–4.51) in PTBMIL Group 2 and 3.12 (1.48–6.58) in PTBMIL Group 3 versus PTBMIL Group 1. Furthermore, adding the PTBMIL Group to a multivariate model including known promoters of DN progression improved prediction of the endpoint (c-index increased by 0.02 [95% CI: 0.00–0.04]).

Conclusions

PTBMIL may be useful for predicting the renal prognosis of patients with biopsy-proven DN, but further investigation of these lesions in various stages of DN is needed.