Table 1 Comparison of calculated features between ESRD positive and ESRD negative patients (performed on all the data)
From: Machine learning algorithm for early detection of end-stage renal disease
Feature | Control group ESRD negative Mean (SD) n = 25,406 | Case group ESRD positive Mean (SD) n = 1585 | p-value |
|---|---|---|---|
Acute kidney injury (AKI) episodes (per year) | 0.43[±2.01] | 0.88[±2.83] | < 0.001 |
Electrolyte imbalance events (per year) | 0.07[±0.41] | 0.14[±0.72] | < 0.001 |
Fluid retention events (per year) | 0.01[±0.30] | 0.04[±0.56] | < 0.001 |
Urinalysis exams (per year) | 1.39[±1.99] | 1.72[±2.50] | < 0.001 |
Kidney biopsies (per year) | 0.001[±0.029] | 0.004[±0.050] | < 0.001 |
Days under ACEi treatment | 66.83 ± [107.48] | 55.97 ± [93.83] | < 0.001 |
Hospitalizations | 1.10[±2.40] | 1.45[±2.83] | < 0.001 |
Hypertensive crisis episodes (per year) | 0.44[±0.99] | 1.39[±2.37] | < 0.001 |
Loop diuretics prescriptions (per year) | 0.70[±1.86] | 0.97[±1.97] | < 0.001 |
Lab proteinuria | 0.11[±0.51] | 0.20[±0.69] | < 0.001 |
Hyperparathyroidism | 0.05[±0.33] | 0.13[±0.56] | < 0.001 |
Phosphorus abnormalities | 0.0008[±0.020] | 0.0127[±0.215] | < 0.001 |
Chronic nephritic syndrome | 0.001[±0.031] | 0.012[±0.19] | < 0.001 |
Non-nephrogenic complications of diabetes | 0.58[±2.16] | 0.81[±2.75] | < 0.001 |
CHF (percent positives) | 19.8% | 27.3% | < 0.001 |
Stroke (percent positives) | 8.3% | 10.5% | 0.003 |
Ischemic heart disease (percent positive) | 31.9% | 39.3% | < 0.001 |
Myocardial infarction | 36.4% | 56.1% | < 0.001 |
Anemia | 35.0% | 45.8% | < 0.001 |
Obesity | 18.5% | 12.3% | < 0.001 |
Sex–Female | 50% | 47.1% | 0.08 |
Sex–Male | 50% | 52.9% | 0.08 |
Age (years) | 70.72 [±13.12] | 70.00 [±11.95] | 0.83 |
CKD Stage 1, 6 months before index data | 3.67% | 0.82% | < 0.001 |
CKD Stage 2, 6 months before index data | 10.00% | 3.09% | < 0.001 |
CKD Stage 3, 6 months before index data | 47.04% | 35.01% | < 0.001 |
CKD Stage 4, 6 months before index data | 8.14% | 52.55% | < 0.001 |