DeepDyve requires Javascript to function. Please enable Javascript on your browser to continue.
Construction of a Prediction Model for the Mortality of Elderly Patients with Diabetic Nephropathy
Construction of a Prediction Model for the Mortality of Elderly Patients with Diabetic Nephropathy
Wang, Li;Lv, Yan
2022-09-12 00:00:00
Hindawi Journal of Healthcare Engineering Volume 2022, Article ID 5724050, 14 pages https://doi.org/10.1155/2022/5724050 Research Article Construction of a Prediction Model for the Mortality of Elderly Patients with Diabetic Nephropathy 1 2 Li Wang and Yan Lv Geriatrics Department of Shenzhen Luohu People’s Hospital, Shenzhen 518000, Guangdong, China Department of Nephrology, Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Tongji Shanxi Hospital, ƒird Hospital of Shanxi Medical University, Taiyuan 030032, Shanxi, China Correspondence should be addressed to Yan Lv; lvyanshenzhen@outlook.com Received 13 April 2022; Revised 9 June 2022; Accepted 26 July 2022; Published 12 September 2022 Academic Editor: Emanuele Rizzuto Copyright © 2022 Li Wang and Yan Lv. �is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. To construct a prediction model for all-cause mortality in elderly diabetic nephropathy (DN) patients, in this cohort study, the data of 511 DN patients aged ≥65 years were collected and the participants were divided into the training set (n 358) and the testing set (n 153). �e median survival time of all participants was 2 years. �e data in the training set were grouped into the survival group (n 203) or the death group (n 155). Variables with P ≤ 0.1 between the two groups were selected as preliminary predictors and involved into the multivariable logistic regression model and the covariables were gradually adjusted. �e receiver operator characteristic (ROC), Kolmogorov-Smirnov (KS), and calibration curves were plotted for evaluating the predictive performance of the model. Internal validation of the performance of the model was veri•ed in the testing set. �e predictive values of the model were also conducted in terms of people with di–erent genders and ages or accompanied with chronic kidney disease (CKD) or cardiovascular diseases (CVD), respectively. In total, 216 (42.27%) elderly DN patients were dead within 2 years. �e prediction model for the 2-year mortality of elderly patients with DN was established based on length of stay (LOS), temperature, heart rate, peripheral oxygen saturation (SpO ), serum creatinine (Scr), red cell distribution width (RDW), the simpli•ed acute physiology score-II (SAPS-II), hyperlipidemia, and the Chronic Kidney Disease Epidemiology Collaboration equation for es- timated glomerular •ltration rate (eGFR-CKD-EPI). �e AUC of the model was 0.78 (95% CI: 0.73–0.83) in the training set and 0.72 (95% CI: 0.63–0.80) in the testing set. �e AUC of the model was 0.78 (95% CI: 0.65–0.91) in females and 0.78 (95%CI: 0.68–0.88) in patients ≤75 years. �e AUC of the model was 0.74 (95% CI: 0.64–0.84) in patients accompanied with CKD. �e model had good predictive value for the mortality of elderly patients with DN within 2 years. In addition, the model showed good predictive values for female DN patients, DN patients ≤75 years, and DN patients accompanied with CKD. clinical outcomes, which were largely due to the serious 1. Introduction complications [6]. �erefore, predicting the all-cause mortality in DN patients was of great value for providing Diabetic nephropathy (DN) is a common microvascular complication of diabetes mellitus (DM) [1]. Approximately timely interventions in these patients and improving the outcomes of these patients. 30% of DM patients are diagnosed with renal complications including DN [2]. DN in patients can lead to end-stage renal Previously, various studies have explored the risk factors failure and disability, which is associated with high mortality for the mortality in DN patients [7–9], but the risk of all over the world [3]. DN patients tend to be elderly and may mortality could not be estimated based on the •ndings of be associated with various complications, such as cerebro- these studies, as they did not form a prediction model. vascular, cardiovascular, peripheral vascular, connective Currently, the model for predicting the mortality of DN tissue, liver, and chronic pulmonary diseases and tumors patients was rare. In 2017, Sato et al. [10] established a [4, 5]. DN is associated with higher mortality rates and worse prediction model for all-cause mortality in DN patients [10]. 2 Journal of Healthcare Engineering )e model had an area under the curve (AUC) of 0.791, bicarbonate, hematocrit, hemoglobin, mean corpuscular which had good predictive ability for the mortality of DN hemoglobin concentration (MCHC, 10 g/L), red cell dis- patients. Previously, multiple studies have indicated that tribution width (RDW, %), chronic obstructive pulmonary prediction model based on combined variables might be disease (COPD, no or yes), atrial fibrillation (AF, no or yes), better than those including only one variable [11]. )e liver cirrhosis (no or yes), respiratory failure (no or yes), prediction model by Sato et al. [10] was focused on pre- hyperlipidemia (no or yes), malignant cancer (no or yes), dialysis neutrophil-lymphocyte ratio, and validation was not SAPS-II, SOFA score, insulin (no or yes), metformin (no or performed to verify the performance of the model. Due to yes), survival time, the Chronic Kidney Disease Epidemi- the poor prognosis of DN patients at old age [12], a suitable ology Collaboration equation for estimated glomerular fil- prediction model was required for the all-cause mortality in tration rate (eGFR-CKD-EPI, mL/min/m ), the elderly DN patients to quickly identify those at high risk of Modification of Diet in Renal Disease equation for estimated mortality and provide timely treatments for these patients. glomerular filtration rate (eGFR-MDRD, mL/min/m ), In this study, the purpose was to construct a prediction CVD (no or yes), CKD (no or yes), myocardial infarction model for all-cause mortality in elderly DN patients. )e (no or yes), hypertension (no or yes), and peripheral vascular predictors were screened out and included in the model. )e disease (no or yes). internal validation was performed to evaluate the predictive value of the model. Subgroup analysis was also conducted in 2.3. Outcome Variables. )e outcome variable was the death terms of gender and being complicated with chronic kidney of elderly DN patients within 2 years. )e follow-up time disease (CKD) or cardiovascular diseases (CVD). was 10 years and the median survival time was 2 years. 2. Methods 2.4. Definitions of Variables. eGFR-MDRD � 175.0 × Scr −1.154 × age−0.203 × 0.742 (if female) × 1.212 (if black); 2.1. Study Population. In this cohort study, the data of 522 eGFR-CKD-EPI � 141 × min (Scr/κ, 1) α × max (Scr/κ, 1) DN patients aged ≥65 years were derived from Medical − 1.029 × 0.993 age × 1.108 (if female) × 1.159 (if black). κ is Information Mart for Intensive Care (MIMIC-III) database. 0.7 for females and 0.9 for males, α is −0.329 for females and MIMIC-III database is an extensive and single-center da- −0.411 for males, min indicates the minimum of Scr/κ or 1, tabase, constructed by Institutional Review Boards (IRB) of and max indicates the maximum of Scr/κ or 1. LOS is the the Massachusetts Institute of Technology (Cambridge, MA, length of stay in the ICUs. USA) and Beth Israel Deaconess Medical Center. It con- tained the data of over 50000 hospital patients admitted to intensive care units (ICUs) between 2001 and 2012 including 2.5. Logistic Regression Model. Logistic regression is a the demographic details, admission and discharge times, classification method applied for binary or classification dates of death, procedures such as dialysis, imaging studies, method generalizing logistic regression to multiclass blood chemistry, hematology, urine analysis, microbiology problems multinomial outcome variables. It evaluates the test results, administration records of intravenous medica- associations between a dependent categorical outcome and tions, medication orders, free text notes such as provider one or more independent predictor variables, which pro- progress notes and hospital discharge summaries, and vides predicted probabilities for each category [14] (1). )e nurse-verified vital signs [13]. After excluding participants detailed formula of the logistic regression model is as without the data on Sequential Organ Failure Assessment follows: (SOFA) score, the simplified acute physiology score-II (SAPS-II), and temperature, 511 patients were finally in- log it P � ln � a + b x + b x + · · · + b x , 1 1 2 2 m m 1 − P volved in our study. (1) a+b x +b x +···+b x 1 1 2 2 m m P � . a+b x +b x +···+b x 1 1 2 2 m m 2.2. Potential Predictors. Potential predictors were analyzed 1 + e in this study including gender, marital status (divorced, married, separated, single, widowed, or unknown), ethnicity (Asian, Black, Hispanic or Latino, White, others, or un- 2.6. Statistical Analysis. )e normal distributed measure- known), length of stay (LOS, day), age (years), respiratory ment data were expressed as mean± standard deviation rate (times/min), temperature ( C), heart rate (times/min), (mean± SD), and comparisons between groups were sub- systolic blood pressure (SBP, mmHg), diastolic blood jected to independent-sample t-test. Nonnormal distributed pressure (DBP, mmHg), mean arterial pressure (MAP, data were described as M (Q , Q ), and the Mann-Whitney U 1 3 mmHg), peripheral oxygen saturation (SpO , %), white rank-sum test was used for comparing differences between 3 3 blood cells (WBC, 10 /μL), red blood cells (RBC, 10 /μL), groups. )e enumeration data were displayed as n (%), and sodium (mEq/L), potassium (mEq/L), phosphate (mEq/L), comparisons between groups were performed by χ test or calcium (mEq/L), magnesium (mEq/L), platelets (PLT, k/ Fisher’s exact probability method [15]. All the data were μL), lactate, international normalized ratio (INR), mean divided into the training set (n � 358) and the testing set corpuscular volume (MCV, fl), glucose (mg/dL), serum (n � 153) at a ratio of 7 : 3 . )e prediction model was creatinine (Scr, mg/dL), blood urea nitrogen (BUN, mg/dL), constructed in the training set and verified in the testing set. Journal of Healthcare Engineering 3 DN patients aged ≥65 years in MIMIC-III (n=522) Excluded Missing the data on SOFA score and SAPS-II (n=9) Missing the data on temperature (n=2) Patients finally included (n=511) Figure 1: )e screen process of the participants. )e data in the training set were grouped into the survival dL. )e median survival time of all patients was 652.00 days. group (n � 203) or the death group (n � 155), and com- )e median eGFR-CKD-EPI was 21.44 mL/min/m and the parisons between the two groups were performed. Variables median eGFR-MDRD was 21.8 mL/min/m . )ere were 389 with P≤ 0.1 were selected as preliminary predictors. )e patients accompanied with CVD, accounting for 76.13%, and preliminarily screened predictors were then involved in the 333 patients accompanied with CKD, accounting for 65.17%. multivariable logistic regression model and the covariables )e median survival time of all participants was 730 days and were gradually adjusted. Subgroup analysis was conducted 216 people died within 2 years, accounting for 42.27%. )e in male group and female group, CKD group and non-CKD LOS in the survival group was shorter than that in the death group, CVD group and non-CVD group, age ≤75 years group (2.15 days versus 3.24 days). )e median survival time group, and age>75 years group, respectively. )e area under of the participants in the survival group was longer than that the curve (AUC), Kolmogorov-Smirnov (KS), calibration in the death group (730.00 days versus 61.50 days) (Table 1). curve, sensitivity, specificity, negative predictive value )e equilibrium test revealed that there was no significant (NPV), positive predictive value (PPV), and accuracy were difference between the data of participants in the training set employed for evaluating the predictive performance of the and the testing set (Table 2). model. A nomogram was also plotted to evaluate the pos- LOS: length of stay, SBP: systolic blood pressure, DBP: sibility of mortality of elderly patients with DN. )e con- diastolic blood pressure, MAP: mean arterial pressure, SpO : fidence level was 0.05 and Python 3 was used for statistical peripheral oxygen saturation, WBC: white blood cells, RBC: analysis. red blood cells, INR: international normalized ratio, MCV: mean corpuscular volume, MCHC: mean corpuscular he- moglobin concentration, RDW: red cell distribution width, 3. Results COPD: chronic obstructive pulmonary disease, AF: atrial 3.1. Missing Value Manipulation and Sensitivity Analysis. fibrillation, eGFR-CKD-EPI: the Chronic Kidney Disease )e missing values of variables are shown in Supplementary Epidemiology Collaboration equation for estimated glo- Table 1. )e missing data were manipulated via multiple merular filtration rate, eGFR-MDRD: the Modification of interpolation using R mice. Sensitivity analysis was per- Diet in Renal Disease equation for estimated glomerular formed in the data before and after the manipulation. )e filtration rate, CKD: chronic kidney disease, CVD: cardio- results delineated that there was no statistical difference vascular diseases, SOFA: Sequential Organ Failure Assess- between the data before and after the manipulation, indi- ment, SAPS-II: the simplified acute physiology score-II. cating that the data after manipulation could be used for further analysis. 3.3. Comparisons between the Characteristics of Patients in the Survival Group and Death Group in the Training Set. )e 3.2. Baseline Characteristics of Participants. In total, 522 DN median LOS (2.15 days versus 3.01 days, Z � 3.734), age (73.59 years versus 76.03 years, Z � 1.770), INR (1.20 versus patients aged≥65 years from MIMIC-III were involved in our study. Participants without the data on SOFA score and 1.30, Z � 2.767), Scr (2.30 mg/dL versus 2.90 mg/dL, SAPS-II (n � 9) and those without the data on temperature Z � 2.100), BUN (43.00 mg/dL versus 50.00 mg/dL, (n � 2) were excluded, and 511 patients were finally included. Z � 2.447), SOFA score (5.00 versus 6.00, Z � 4.397), the )e detailed screen process is shown in Figure 1. Among average heart rate (80.12 times/min versus 85.81 times/min, them, 292 people were males, accounting for 57.14%. )e t � −2.95), SpO (96.52 versus 97.42, t � −1.77), RBC (3.72 3 3 median LOS was 2.6 days. )e median age of all participants 10 /μL versus 3.58 10 /μL, t � 1.85), SAPS-II (40.28 versus was 74.39 years. )e median glucose level was 166 mg/dL. )e 45.77, t � −4.62), and the proportion of patients with re- median Scr level was 2.7 mg/dL. )e median BUN was 45 mg/ spiratory failure (23.65% versus 33.55%, χ � 4.282) were 4 Journal of Healthcare Engineering Table 1: Comparisons of the characteristics of surviving and dead patients. Group Variables Total (n � 511) Survival group (n � 295) Death group (n � 216) Statistics P Gender, n (%) χ � 3.000 0.083 Male 292 (57.14) 159 (53.90) 133 (61.57) Female 219 (42.86) 136 (46.10) 83 (38.43) Marital status, n (%) χ � 9.472 0.092 Divorced 34 (6.65) 22 (7.46) 12 (5.56) Married 247 (48.34) 143 (48.47) 104 (48.15) Separated 3 (0.59) 0 (0.00) 3 (1.39) Single 85 (16.63) 53 (17.97) 32 (14.81) Unknown 14 (2.74) 5 (1.69) 9 (4.17) Widowed 128 (25.05) 72 (24.41) 56 (25.93) Ethnicity, n (%) χ � 9.861 0.079 Asian 18 (3.52) 7 (2.37) 11 (5.09) Black 93 (18.20) 63 (21.36) 30 (13.89) Hispanic or Latino 12 (2.35) 7 (2.37) 5 (2.31) Others 11 (2.15) 8 (2.71) 3 (1.39) Unknown 43 (8.41) 20 (6.78) 23 (10.65) White 334 (65.36) 190 (64.41) 144 (66.67) LOS, M (Q , Q ) 2.60 (1.37, 4.79) 2.15 (1.24, 3.84) 3.24 (1.64, 6.93) Z � 4.748 <0.001 1 3 Age, M (Q , Q ) 74.39 (69.69, 80.12) 73.90 (69.53, 80.02) 75.14 (70.13, 80.31) Z � 1.343 0.179 1 3 Respiratory rate, mean± SD 19.08± 5.76 18.66± 5.36 19.66± 6.23 t � −1.90 0.058 Temperature, mean± SD 36.46± 0.95 36.54± 0.91 36.36± 1.00 t � 2.10 0.036 Heart rate, mean± SD 82.59± 17.99 80.94± 17.64 84.84± 18.25 t � −2.43 0.015 SBP, mean± SD 126.43± 28.02 127.96± 28.29 124.35± 27.58 t � 1.44 0.150 DBP, mean± SD 58.35± 16.25 58.07± 15.74 58.73± 16.96 t � −0.45 0.650 MAP, mean± SD 78.60± 18.90 78.16± 17.55 79.20± 20.63 t � −0.60 0.551 SpO , mean± SD 96.98± 4.73 96.90± 5.14 97.10± 4.10 t � −0.50 0.620 WBC, M (Q , Q ) 9.70 (7.30, 12.70) 9.60 (7.20, 12.70) 9.70 (7.40, 12.65) Z � 0.488 0.626 1 3 RBC, mean± SD 3.65± 0.71 3.69± 0.75 3.60± 0.65 t � 1.37 0.171 Sodium, mean± SD 137.70± 4.71 137.60± 4.75 137.83± 4.66 t � −0.54 0.586 Potassium, mean± SD 4.64± 0.97 4.67± 1.00 4.60± 0.93 t � 0.80 0.426 Phosphate, M (Q ,Q ) 4.00 (3.30, 4.90) 3.90 (3.20, 4.70) 4.10 (3.30, 5.10) Z � 1.918 0.055 1 3 Calcium, mean± SD 8.69± 0.95 8.78± 0.95 8.57± 0.93 t � 2.45 0.014 208.50 (166.50, PLT, M (Q , Q ) 216.00 (169.00, 288.00) 218.00 (173.00, 277.00) Z � −0.609 0.542 1 3 295.00) Lactate, M (Q , Q ) 1.60 (1.22, 2.20) 1.60 (1.20, 2.20) 1.70 (1.30, 2.38) Z � 1.750 0.080 1 3 INR, M (Q ,Q ) 1.20 (1.10,1.50) 1.20 (1.10,1.40) 1.30 (1.10,1.50) Z � 2.904 0.004 1 3 MCV, mean± SD 90.89± 7.56 90.67± 7.72 91.20± 7.34 t � −0.78 0.436 Magnesium, mean± SD 2.05± 0.45 2.05± 0.44 2.05± 0.46 t � 0.01 0.989 158.50 (119.50, Glucose, M (Q , Q ) 166.00 (125.00, 242.00) 176.00 (130.00, 249.00) Z � −1.983 0.047 1 3 229.00) Creatinine, M (Q , Q ) 2.70 (1.70, 4.30) 2.40 (1.60, 4.10) 2.90 (1.90, 4.45) Z � 2.571 0.010 1 3 BUN, M (Q , Q ) 45.00 (31.00, 68.00) 44.00 (30.00, 65.00) 48.00 (32.00, 71.00) Z � 2.022 0.043 1 3 Bicarbonate, mean± SD 24.37± 5.39 24.06± 4.99 24.79± 5.88 t � −1.47 0.142 Hematocrit, mean± SD 32.91± 6.02 33.11± 6.42 32.64± 5.43 t � 0.89 0.374 Hemoglobin, mean± SD 10.81± 1.95 10.95± 2.07 10.62± 1.77 t � 1.97 0.049 MCHC, mean± SD 32.85± 1.60 33.06± 1.56 32.56± 1.61 t � 3.55 <0.001 RDW, mean± SD 15.81± 1.89 15.38± 1.72 16.39± 1.95 t � −6.22 <0.001 COPD, n (%) χ � 0.526 0.468 No 419 (82.00) 245 (83.05) 174 (80.56) Yes 92 (18.00) 50 (16.95) 42 (19.44) AF, n (%) χ �1.546 0.214 No 286 (55.97) 172 (58.31) 114 (52.78) Yes 225 (44.03) 123 (41.69) 102 (47.22) Liver cirrhosis, n (%) χ � 0.097 0.755 No 488 (95.50) 281 (95.25) 207 (95.83) Yes 23 (4.50) 14 (4.75) 9 (4.17) Respiratory failure, n (%) χ �13.735 <0.001 No 355 (69.47) 224 (75.93) 131 (60.65) Journal of Healthcare Engineering 5 Table 1: Continued. Group Variables Total (n � 511) Survival group (n � 295) Death group (n � 216) Statistics P Yes 156 (30.53) 71 (24.07) 85 (39.35) Hyperlipidemia, n (%) χ � 27.292 <0.001 No 267 (52.25) 125 (42.37) 142 (65.74) Yes 244 (47.75) 170 (57.63) 74 (34.26) Malignant cancer, n (%) χ � 0.070 0.792 No 405 (79.26) 235 (79.66) 170 (78.70) Yes 106 (20.74) 60 (20.34) 46 (21.30) SAPS-II score, mean± SD 42.79± 11.78 40.74± 11.18 45.60± 12.02 t � −4.70 <0.001 SOFA score, M (Q , Q ) 6.00 (4.00, 8.00) 5.00 (3.00, 7.00) 6.00 (4.00, 8.00) Z � 4.448 <0.001 1 3 Insulin, n (%) χ � 4.861 0.027 No 33 (6.46) 13 (4.41) 20 (9.26) Yes 478 (93.54) 282 (95.59) 196 (90.74) Metformin, n (%) χ � 0.254 0.615 No 497 (97.26) 286 (96.95) 211 (97.69) Yes 14 (2.74) 9 (3.05) 5 (2.31) 652.00 (87.00, 3650.00 (1088.00, Survival time, M (Q , Q ) 61.50 (17.00, 165.50) Z � −19.702 <0.001 1 3 3650.00) 3650.00) eGFR-MDRD, M (Q , Q ) 21.80 (11.97, 34.05) 23.02 (12.55, 37.12) 19.63 (11.32, 30.85) Z � −2.626 0.009 1 3 eGFR-CKD-EPI, M (Q , Q ) 21.44 (12.93, 33.35) 23.94 (13.67, 36.22) 19.51 (11.86, 29.22) Z � −3.189 0.001 1 3 CVD, n (%) χ � 2.528 0.112 No 122 (23.87) 78 (26.44) 44 (20.37) Yes 389 (76.13) 217 (73.56) 172 (79.63) CKD, n (%) χ � 8.774 0.003 No 178 (34.83) 87 (29.49) 91 (42.13) Yes 333 (65.17) 208 (70.51) 125 (57.87) Myocardial infarction, n (%) χ � 0.027 0.870 No 341 (66.73) 196 (66.44) 145 (67.13) Yes 170 (33.27) 99 (33.56) 71 (32.87) Hypertension, n (%) χ �1.142 0.285 No 400 (78.28) 226 (76.61) 174 (80.56) Yes 111 (21.72) 69 (23.39) 42 (19.44) Peripheral vascular disease, n (%) χ � 4.106 0.043 No 481 (94.13) 283 (95.93) 198 (91.67) Yes 30 (5.87) 12 (4.07) 18 (8.33) Survival time within 2 years, M (Q ,Q ) 730.00 (87.00, 730.00) 730.00 (730.00, 730.00) 61.50 (17.00, 165.50) Z � −21.501 <0.001 1 3 Death within 10 years, n (%) χ �189.837 <0.001 No 172 (33.66) 172 (58.31) 0 (0.00) Yes 339 (66.34) 123 (41.69) 216 (100.00) lower in the survival group than in the death group. )e 3.4. Predictors for Mortality of Elderly Patients with DN. median eGFR-MDRD (25.42 mL/min/m versus 20.41 mL/ Variables with P≤ 0.1 in the survival group and the death 2 2 min/m , Z � −2.266), eGFR-CKD-EPI (25.60 mL/min/m group were included in the multivariable logistical analysis. versus 19.68 mL/min/m , Z � −2.705), the average tem- Stepwise regression was applied to identify the predictors for ° ° perature (36.59 C versus 36.41 C, t � 1.75), calcium mortality of elderly patients with DN within 2 years. As (8.82 mEq/L versus 8.57 mEq/L, t � 2.48), hemoglobin depicted in Table 4, LOS (OR � 1.10, 95% CI: 1.03–1.17), (11.02 versus 10.61 t � 2.03), MCHC (33.05 10 g/L versus temperature (OR � 0.74, 95% CI: 0.63–0.88), heart rate 32.59 10 g/L, t � 2.78), and the proration of patients with (OR � 1.03, 95% CI: 1.01–1.04), SpO (OR � 1.06, 95% CI: hyperlipidemia (60.59% versus 34.19%, χ � 4.282), CKD 1.01–1.11), Scr (OR � 0.83, 95% CI: 0.69–0.98), RDW (70.44% versus 60.65%, χ (OR � 1.25, 95% CI: 1.10–1.42), SAPS-II (OR � 1.02, 95% CI: � 3.771), diabetic retinopathy (21.18% versus 12.26%, χ � 4.888), and insulin use (94.58% 1.01–1.05), hyperlipidemia (OR � 0.43, 95% CI: 0.27–0.70), versus 89.68%, χ � 3.031) in the survival group were higher and eGFR-CKD-EPI (OR � 0.97, 95% CI: 0.94–0.99) were than those in the death group. )e proportion of patients predictors associated with the risk of mortality in with different marital status was statistically different be- elderly patients with DN within 2 years. )e final model was tween the survival group and the death group (χ �10.722) Log (p/1 − p) � 0.09 × LOS − 0.29 × temperature − 0.19 × (Table 3). creatinine + 0.03 × heart rate + 0.05 × SpO + 0.22 × RDW + 2 6 Journal of Healthcare Engineering Table 2: Baseline data of the participants in the training set and the testing set. Statistical Variable Total (n � 511) Testing set (n � 153) Training set (n � 358) P magnitude Gender, n (%) χ � 0.012 0.911 Male 292 (57.14) 88 (57.52) 204 (56.98) Female 219 (42.86) 65 (42.48) 154 (43.02) Marital status, n (%) χ � 5.188 0.393 Divorced 34 (6.65) 10 (6.54) 24 (6.70) Married 247 (48.34) 73 (47.71) 174 (48.60) Separated 3 (0.59) 0 (0.00) 3 (0.84) Single 85 (16.63) 32 (20.92) 53 (14.80) Unknown 14 (2.74) 4 (2.61) 10 (2.79) Widowed 128 (25.05) 34 (22.22) 94 (26.26) Ethnicity, n (%) χ � 3.443 0.632 Asian 18 (3.52) 3 (1.96) 15 (4.19) Black 93 (18.20) 28 (18.30) 65 (18.16) Hispanic or Latino 12 (2.35) 5 (3.27) 7 (1.96) Others 11 (2.15) 2 (1.31) 9 (2.51) Unknown 43 (8.41) 12 (7.84) 31 (8.66) White 334 (65.36) 103 (67.32) 231 (64.53) LOS, M (Q , Q ) 2.60 (1.37, 4.79) 2.93 (1.41, 5.02) 2.41 (1.35, 4.38) Z � 1.135 0.256 1 3 Age, M (Q , Q ) 74.39 (69.69, 80.12) 73.98 (69.44, 79.66) 74.60 (69.92, 80.30) Z � −0.983 0.326 1 3 Respiratory rate, mean± SD 19.08± 5.76 18.76± 6.00 19.22± 5.66 t � −0.84 0.404 Temperature, mean± SD 36.46± 0.95 36.35± 0.97 36.51± 0.94 t � −1.74 0.083 Heart rate, mean± SD 82.59± 17.99 82.61± 17.29 82.58± 18.30 t � 0.01 0.989 SBP, mean± SD 126.43± 28.02 126.34± 29.08 126.47± 27.60 t � −0.05 0.962 DBP, mean± SD 58.35± 16.25 57.85± 13.00 58.56± 17.48 t � −0.51 0.611 MAP, mean± SD 78.60± 18.90 79.54± 16.45 78.19± 19.87 t � 0.79 0.428 SpO , mean± SD 96.98± 4.73 97.16± 3.84 96.91± 5.06 t � 0.62 0.534 WBC, M (Q , Q ) 9.70 (7.30, 12.70) 9.40 (7.00, 12.00) 9.70 (7.40, 12.70) Z � −1.160 0.246 1 3 RBC, mean± SD 3.65± 0.71 3.62± 0.74 3.66± 0.70 t � −0.59 0.555 Sodium, mean± SD 137.70± 4.71 138.10± 4.17 137.53± 4.92 t � 1.33 0.183 Potassium, mean± SD 4.64± 0.97 4.67± 0.96 4.64± 0.98 t � 0.34 0.737 Phosphate, M (Q , Q ) 4.00 (3.30, 4.90) 4.00 (3.30, 4.70) 4.00 (3.30, 4.90) Z � −0.253 0.800 1 3 Calcium, mean± SD 8.69± 0.95 8.66± 0.96 8.71± 0.94 t � −0.58 0.562 216.00 (169.00, 218.50 (173.00, PLT, M (Q , Q ) 208.00 (159.00, 269.00) Z � −1.352 0.176 1 3 288.00) 289.00) Lactate, M (Q ,Q ) 1.60 (1.22, 2.20) 1.60 (1.20, 2.30) 1.60 (1.26, 2.20) Z � 0.520 0.603 1 3 INR, M (Q ,Q ) 1.20 (1.10, 1.50) 1.20 (1.10, 1.50) 1.20 (1.10, 1.40) Z � 0.507 0.612 1 3 MCV, mean± SD 90.89± 7.56 91.00± 7.42 90.85± 7.63 t � 0.20 0.838 Magnesium, mean± SD 2.05± 0.45 2.08± 0.52 2.04± 0.42 t � 0.97 0.334 166.00 (125.00, 168.50 (125.00, Glucose, M (Q , Q ) 162.00 (124.00, 230.00) Z � −0.668 0.504 1 3 242.00) 249.00) Creatinine, M (Q , Q ) 2.70 (1.70, 4.30) 2.80 (1.80, 4.40) 2.65 (1.70, 4.30) Z � 1.156 0.248 1 3 BUN, M (Q , Q ) 45.00 (31.00, 68.00) 42.00 (32.00, 69.00) 46.00 (31.00, 68.00) Z � −0.179 0.858 1 3 Bicarbonate, mean± SD 24.37± 5.39 24.41± 5.25 24.36± 5.46 t � 0.09 0.927 Hematocrit, mean± SD 32.91± 6.02 32.63± 6.23 33.03± 5.93 t � −0.69 0.493 Hemoglobin, mean± SD 10.81± 1.95 10.73± 2.02 10.84± 1.93 t � −0.62 0.538 MCHC, mean± SD 32.85± 1.60 32.84± 1.66 32.85± 1.58 t � −0.09 0.932 RDW, mean± SD 15.81± 1.89 15.77± 1.74 15.82± 1.95 t � −0.28 0.782 COPD, n (%) χ �1.254 0.263 No 419 (82.00) 121 (79.08) 298 (83.24) Yes 92 (18.00) 32 (20.92) 60 (16.76) AF, n (%) χ �1.665 0.197 No 286 (55.97) 79 (51.63) 207 (57.82) Yes 225 (44.03) 74 (48.37) 151 (42.18) Liver cirrhosis, n (%) χ � 0.269 0.604 No 488 (95.50) 145 (94.77) 343 (95.81) Yes 23 (4.50) 8 (5.23) 15 (4.19) Respiratory failure, n (%) χ � 3.798 0.051 No 355 (69.47) 97 (63.40) 258 (72.07) Journal of Healthcare Engineering 7 Table 2: Continued. Statistical Variable Total (n � 511) Testing set (n � 153) Training set (n � 358) P magnitude Yes 156 (30.53) 56 (36.60) 100 (27.93) Hyperlipidemia, n (%) χ � 0.956 0.328 No 267 (52.25) 85 (55.56) 182 (50.84) Yes 244 (47.75) 68 (44.44) 176 (49.16) Malignant cancer, n (%) χ �1.571 0.210 No 405 (79.26) 116 (75.82) 289 (80.73) Yes 106 (20.74) 37 (24.18) 69 (19.27) SAPS-II score, mean± SD 42.79± 11.78 43.12± 12.47 42.66± 11.48 t � 0.41 0.686 SOFA score, M (Q , Q ) 6.00 (4.00, 8.00) 6.00 (4.00, 8.00) 5.00 (4.00, 7.00) Z � 2.131 0.033 1 3 Insulin, n (%) χ � 2.326 0.127 No 33 (6.46) 6 (3.92) 27 (7.54) Yes 478 (93.54) 147 (96.08) 331 (92.46) Metformin, n (%) Fisher 0.768 No 497 (97.26) 148 (96.73) 349 (97.49) Yes 14 (2.74) 5 (3.27) 9 (2.51) 652.00 (87.00, 770.00 (103.00, 584.00 (80.00, Survival time, M (Q , Q ) Z � 0.813 0.416 1 3 3650.00) 3650.00) 3650.00) eGFR-MDRD, M (Q , Q ) 21.80 (11.97, 34.05) 18.92 (11.86, 32.84) 22.16 (12.12, 34.48) Z � −1.191 0.234 1 3 eGFR-CKD-EPI, M (Q , Q ) 21.44 (12.93, 33.35) 20.47 (11.85, 32.05) 21.98 (13.41, 33.94) Z � -1.418 0.156 1 3 CVD, n (%) χ � 0.328 0.567 No 122 (23.87) 34 (22.22) 88 (24.58) Yes 389 (76.13) 119 (77.78) 270 (75.42) CKD, n (%) χ � 0.564 0.453 No 178 (34.83) 57 (37.25) 121 (33.80) Yes 333 (65.17) 96 (62.75) 237 (66.20) Myocardial infarction, n (%) χ � 0.185 0.667 No 341 (66.73) 100 (65.36) 241 (67.32) Yes 170 (33.27) 53 (34.64) 117 (32.68) Hypertension, n (%) χ � 0.274 0.601 No 400 (78.28) 122 (79.74) 278 (77.65) Yes 111 (21.72) 31 (20.26) 80 (22.35) Peripheral vascular disease, n (%) χ � 0.000 0.994 No 481 (94.13) 144 (94.12) 337 (94.13) Yes 30 (5.87) 9 (5.88) 21 (5.87) Survival time within 2 years, M (Q , 730.00(87.00, 730.00) 730.00(103.00, 730.00) 730.00(80.00, 730.00) Z � 0.964 0.335 Q ) Death within 2 years, n (%) χ � 0.516 0.473 No 295 (57.73) 92 (60.13) 203 (56.70) Yes 216 (42.27) 61 (39.87) 155 (43.30) LOS: length of stay, SBP: systolic blood pressure, DBP: diastolic blood pressure, MAP: mean arterial pressure, SpO : peripheral oxygen saturation, WBC: white blood cells, RBC: red blood cells, INR: international normalized ratio, MCV: mean corpuscular volume, MCHC: mean corpuscular hemoglobin con- centration, RDW: red cell distribution width, COPD: chronic obstructive pulmonary disease, AF: atrial fibrillation, eGFR-CKD-EPI: the Chronic Kidney Disease Epidemiology Collaboration equation for estimated glomerular filtration rate, eGFR-MDRD: the Modification of Diet in Renal Disease equation for estimated glomerular filtration rate, CKD: chronic kidney disease, CVD: cardiovascular diseases, SOFA: Sequential Organ Failure Assessment, SAPS-II: the simplified acute physiology score-II. 0.02 × SAPS-II-0.84 × hyperlipidemia − 0.03 × eGFR-CKD- shown in Figure 2. For the model in the testing set, the EPI. sensitivity was 0.90 (95% CI: 0.83–0.98), the specificity was 0.47 (95% CI: 0.37–0.57), the PPV was 0.53 (95% CI: 0.43–0.62), the NPV was 0.88 (95% CI: 0.79–0.97), the AUC 3.5. Predictive Value of the Model. According to the data in was 0.72 (95% CI: 0.63–0.80), the accuracy was 0.64 (95% Table 5, for the model in the training set, the sensitivity was CI: 0.56–0.72). )e ROC, KS, and calibration curves in the 0.85 (95% CI: 0.80–0.91), the specificity was 0.59 (95% CI: testing set are exhibited in Figure 3. )e nomogram was 0.52–0.65), the PPV was 0.61 (95% CI: 0.55–0.68), the NPV plotted and a sample was selected, which showed that the was 0.84 (95% CI: 0.78–0.90), the AUC was 0.78 (95% CI: total score of the patient was 284, and the predicted 0.73–0.83), and the accuracy was 0.70 (95% CI: 0.65–0.75). mortality probability was 0.155, which was lower than the )e ROC, KS, and calibration curves in the training set are cut-off, 0.33 (Figure 4). )e predicted outcome of the 8 Journal of Healthcare Engineering Table 3: Comparisons between the characteristics of patients in the survival group and death group in the training set. Variable Survival within 2 years (n � 203) Death within 2 years (n � 155) Statistical magnitude P Gender, n (%) χ �1.495 0.221 Male 110 (54.19) 94 (60.65) Female 93 (45.81) 61 (39.35) Marital status, n (%) χ �10.722 0.057 Divorced 17 (8.37) 7 (4.52) Married 98 (48.28) 76 (49.03) Separated 0 (0.00) 3 (1.94) Single 33 (16.26) 20 (12.90) Unknown 3 (1.48) 7 (4.52) Widowed 52 (25.62) 42 (27.10) Ethnicity, n (%) Fisher 0.134 Asian 6 (2.96) 9 (5.81) Black 44 (21.67) 21 (13.55) Hispanic or Latino 4 (1.97) 3 (1.94) Others 7 (3.45) 2 (1.29) Unknown 14 (6.90) 17 (10.97) White 128 (63.05) 103 (66.45) LOS, M (Q , Q ) 2.15 (1.22, 3.66) 3.01 (1.61, 6.50) Z � 3.734 <0.001 1 3 Age, M (Q , Q ) 73.59 (69.32, 80.24) 76.03 (70.54, 80.87) Z � 1.770 0.077 1 3 Respiratory rate, mean± SD 18.87± 5.16 19.69± 6.23 t � −1.33 0.185 Temperature, mean± SD 36.59± 0.87 36.41± 1.02 t � 1.75 0.081 Heart rate, mean± SD 80.12± 17.07 85.81± 19.38 t � −2.95 0.003 SBP, mean± SD 127.97± 27.43 124.51± 27.80 t � 1.18 0.240 DBP, mean± SD 58.17± 16.54 59.08± 18.68 t � −0.49 0.625 MAP, mean± SD 77.33± 17.95 79.32± 22.14 t � −0.91 0.363 SpO , mean± SD 96.52± 5.86 97.42± 3.73 t � −1.77 0.077 WBC, M (Q , Q ) 9.50 (7.30, 12.40) 10.20 (7.50, 13.00) Z � 1.211 0.226 1 3 RBC, mean± SD 3.72± 0.74 3.58± 0.64 t � 1.85 0.064 Sodium, mean± SD 137.45± 4.91 137.63± 4.94 t � −0.34 0.733 Potassium, mean± SD 4.63± 1.00 4.64± 0.95 t � −0.03 0.979 Phosphate, M (Q , Q ) 3.90 (3.20, 4.80) 4.00 (3.40, 5.10) Z � 1.506 0.132 1 3 Calcium, mean± SD 8.82± 0.96 8.57± 0.91 t � 2.48 0.014 PLT, M (Q , Q ) 218.00 (174.00, 273.00) 220.00 (170.00, 303.00) Z � 0.113 0.910 1 3 Lactate, M (Q , Q ) 1.58 (1.20, 2.10) 1.70 (1.30, 2.30) Z � 1.454 0.146 1 3 INR, M (Q , Q ) 1.20 (1.10,1.40) 1.30 (1.10,1.60) Z � 2.767 0.006 1 3 MCV, mean± SD 90.36± 7.57 91.48± 7.67 t � −1.38 0.167 Magnesium, mean± SD 2.03± 0.38 2.05± 0.46 t � −0.39 0.700 Glucose, M (Q , Q ) 178.00 (125.00, 253.00) 163.00 (125.00, 239.00) Z � −0.995 0.320 1 3 Creatinine, M (Q , Q ) 2.30 (1.60, 4.20) 2.90 (1.90, 4.30) Z � 2.100 0.036 1 3 BUN, M (Q ,Q ) 43.00 (30.00,61.00) 50.00 (32.00,72.00) Z � 2.447 0.014 1 3 Bicarbonate, mean± SD 24.20± 4.97 24.56± 6.05 t � −0.60 0.548 Hematocrit, mean± SD 33.37± 6.28 32.59± 5.43 t � 1.24 0.215 Hemoglobin, mean± SD 11.02± 2.01 10.61± 1.79 t � 2.03 0.043 MCHC, mean± SD 33.05± 1.56 32.59± 1.58 t � 2.78 0.006 RDW, mean± SD 15.40± 1.74 16.37± 2.07 t � −4.71 <0.001 COPD, n (%) χ �1.320 0.251 No 173 (85.22) 125 (80.65) Yes 30 (14.78) 30 (19.35) AF, n (%) χ � 0.612 0.434 No 121 (59.61) 86 (55.48) Yes 82 (40.39) 69 (44.52) Liver cirrhosis, n (%) χ � 0.069 0.792 No 194 (95.57) 149 (96.13) Yes 9 (4.43) 6 (3.87) Respiratory failure, n (%) χ � 4.282 0.039 No 155 (76.35) 103 (66.45) Yes 48 (23.65) 52 (33.55) Hyperlipidemia, n (%) χ � 24.505 <0.001 No 80 (39.41) 102 (65.81) Journal of Healthcare Engineering 9 Table 3: Continued. Variable Survival within 2 years (n � 203) Death within 2 years (n � 155) Statistical magnitude P Yes 123 (60.59) 53 (34.19) Malignant cancer, n (%) χ � 0.604 0.437 No 161 (79.31) 128 (82.58) Yes 42 (20.69) 27 (17.42) SAPS-II score, mean± SD 40.28± 10.74 45.77± 11.71 t � −4.62 <0.001 SOFA score, M (Q , Q ) 5.00 (3.00, 7.00) 6.00 (4.00, 8.00) Z � 4.397 <0.001 1 3 Insulin, n (%) χ � 3.031 0.082 No 11 (5.42) 16 (10.32) Yes 192 (94.58) 139 (89.68) Metformin, n (%) Fisher 0.309 No 196 (96.55) 153 (98.71) Yes 7 (3.45) 2 (1.29) eGFR-MDRD, M (Q , Q ) 25.42 (12.48, 39.27) 20.41 (11.55, 31.45) Z � −2.266 0.023 1 3 eGFR-CKD-EPI, M (Q , Q ) 25.60 (13.67, 36.52) 19.68 (12.99, 29.23) Z � −2.705 0.007 1 3 CVD, n (%) χ � 0.590 0.442 No 53 (26.11) 35 (22.58) Yes 150 (73.89) 120 (77.42) CKD, n (%) χ � 3.771 0.052 No 60 (29.56) 61 (39.35) Yes 143 (70.44) 94 (60.65) Myocardial infarction, n (%) χ � 0.022 0.881 No 136 (67.00) 105 (67.74) Yes 67 (33.00) 50 (32.26) Hypertension, n (%) χ � 0.027 0.870 No 157 (77.34) 121 (78.06) Yes 46 (22.66) 34 (21.94) Diabetic retinopathy, n (%) χ � 4.888 0.027 No 160 (78.82) 136 (87.74) Yes 43 (21.18) 19 (12.26) Peripheral vascular disease, n (%) χ � 0.750 0.386 No 193 (95.07) 144 (92.90) Yes 10 (4.93) 11 (7.10) LOS: length of stay, SBP: systolic blood pressure, DBP: diastolic blood pressure, MAP: mean arterial pressure, SpO : peripheral oxygen saturation, WBC: white blood cells, RBC: red blood cells, INR: international normalized ratio, MCV: mean corpuscular volume, MCHC: mean corpuscular hemoglobin con- centration, RDW: red cell distribution width, COPD: chronic obstructive pulmonary disease, AF: atrial fibrillation, eGFR-CKD-EPI: the Chronic Kidney Disease Epidemiology Collaboration equation for estimated glomerular filtration rate, eGFR-MDRD: the Modification of Diet in Renal Disease equation for estimated glomerular filtration rate, CKD: chronic kidney disease, CVD: cardiovascular diseases, SOFA: Sequential Organ Failure Assessment, SAPS-II: the simplified acute physiology score-II. Table 4: Predictors for mortality of elderly patients with DN. Character β SE z P> |z| OR OR (lower (95%)) OR (upper (95%)) LOS 0.09 0.03 2.90 0.004 1.10 1.03 1.17 Temperature −0.29 0.08 −3.57 <0.001 0.74 0.63 0.88 Heart rate 0.03 0.01 3.49 <0.001 1.03 1.01 1.04 SpO 0.05 0.03 2.00 0.046 1.06 1.01 1.11 Creatinine −0.19 0.09 −2.16 0.031 0.83 0.69 0.98 RDW percent 0.22 0.07 3.36 0.001 1.25 1.10 1.42 SAPS-II 0.02 0.01 1.97 0.049 1.02 1.01 1.05 Hyperlipidemia −0.84 0.25 −3.40 0.001 0.43 0.27 0.70 eGFR-CKD-EPI −0.03 0.01 −2.45 0.014 0.97 0.94 0.99 LOS: length of stay, SpO : peripheral oxygen saturation, eGFR-CKD-EPI: the Chronic Kidney Disease Epidemiology Collaboration equation for estimated glomerular filtration rate, SAPS-II: the simplified acute physiology score-II. 10 Journal of Healthcare Engineering Table 5: )e predictive value of the model. Data set Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI) AUC (95% CI) Accuracy (95% CI) Training set 0.85 (0.80–0.91) 0.59 (0.52–0.65) 0.61 (0.55–0.68) 0.84 (0.78–0.90) 0.78 (0.73–0.83) 0.70 (0.65–0.75) Testing set 0.90 (0.83–0.98) 0.47 (0.37–0.57) 0.53 (0.43–0.62) 0.88 (0.79–0.97) 0.72 (0.63–0.80) 0.64 (0.56–0.72) CI: confidence interval, AUC: area under the curve, NPV: negative predictive value, PPV: positive predictive value. ROC Curve KS Curve Calibration Curve 1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.4 0.6 0.6 0.2 0.4 0.4 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Predicted values 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 40 140 0 20 60 80 100 120 False Positive Rate 0.0 0.2 0.4 0.6 0.8 1.0 Mean predicated value AUC (95%CI) = 0.78 (0.73-0.83) Tpr Diff KS=0.44 Fpr Figure 2: )e AUC, KS, and calibration curves of the model in the training set. ROC Curve KS Curve Calibration Curve 1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.4 0.6 0.6 0.2 0.4 0.4 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Predicted values 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 False Positive Rate N 0.0 0.2 0.4 0.6 0.8 1.0 Mean predicated value AUC (95%CI) = 0.72 (0.63-0.80) Tpr Diff KS=0.40 Fpr Figure 3: )e AUC, KS, and calibration curves of the model in the testing set. patient was survival, which was consistent with the actual the female group, the sensitivity was 0.91 (95% CI: outcome. 0.79–1.00), the specificity was 0.56 (95% CI: 0.41–0.71), the PPV was 0.51 (95% CI: 0.36–0.67), the NPV was 0.92 (95% CI: 0.82–1.00), the AUC was 0.78 (95% CI: 0.65–0.91), and 3.6. *e Predictive Value of the Model concerning Different the accuracy was 0.68 (95% CI: 0.56–0.79) (Table 6). Subgroups 3.6.1. Gender. In the male group, the sensitivity was 0.90 3.6.2. Age. In patients >75 years group, the sensitivity was (95% CI: 0.80–0.99), the specificity was 0.39 (95% CI: 0.88 (95% CI: 0.75–1.00), the specificity was 0.36 (95% CI: 0.25–0.52), the PPV was 0.54 (95% CI: 0.42–0.66), the NPV 0.22–0.50), the PPV was 0.43 (95% CI: 0.30–0.57), the NPV was 0.83 (95% CI: 0.67–0.98), the AUC was 0.66 (95% CI: was 0.84 (95% CI: 0.68–1.00), the AUC was 0.65 (95% CI: 0.55–0.78), and the accuracy was 0.61 (95% CI: 0.51–0.72). In 0.52–0.78), and the accuracy was 0.54 (95% CI: 0.43–0.66). True Positive Rate True Positive Rate Value Value Count Actual values Count Actual values Journal of Healthcare Engineering 11 Points 0 102030405060708090 100 SPO 55 75 SAPSII 20 50 80 Temperature 39 37 35 33 31 Creatinine 12 8 4 0 RDW 12 16 20 23 Yes Hyperlipidemia No Heart Rate 40 80 120 150 eGFR -CKD-EPI 100 80 60 40 20 0 LOS 0 5 10 15 20 25 30 35 40 45 50 55 60 Total points 240 260 280 300 320 340 360 380 400 0.155 Pr( ) 0.01 0.025 0.06 0.15 0.4 0.6 0.8 0.94 0.975 0.99 0.996 Figure 4: )e nomogram of the prediction model. Table 6: )e predictive value of the model in different subgroups. Subgroup Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI) AUC (95% CI) Accuracy (95% CI) Gender Male 0.90 (0.80–0.99) 0.39 (0.25–0.52) 0.54 (0.42–0.66) 0.83 (0.67–0.98) 0.66 (0.55–0.78) 0.61 (0.51–0.72) Female 0.91 (0.79–1.00) 0.56 (0.41–0.71) 0.51 (0.36–0.67) 0.92 (0.82–1.00) 0.78 (0.65–0.91) 0.68 (0.56–0.79) Age >75 years 0.88 (0.75–1.00) 0.36 (0.22–0.50) 0.43 (0.30–0.57) 0.84 (0.68–1.00) 0.65 (0.52–0.78) 0.54 (0.43–0.66) ≤75 years 0.92 (0.83–1.00) 0.57 (0.43–0.72) 0.62 (0.49–0.75) 0.90 (0.79–1.00) 0.78 (0.68–0.88) 0.72 (0.63–0.82) CKD Yes 0.90 (0.80–1.00) 0.51 (0.39–0.63) 0.47 (0.34–0.59) 0.92 (0.83–1.00) 0.74 (0.64–0.84) 0.64 (0.54–0.73) No 0.90 (0.79–1.00) 0.37 (0.19–0.55) 0.61 (0.47–0.76) 0.77 (0.54–1.00) 0.67 (0.52–0.82) 0.65 (0.53–0.77) CVD Yes 0.90 (0.82–0.98) 0.46 (0.34–0.58) 0.57 (0.46–0.67) 0.86 (0.75–0.97) 0.71 (0.61–0.80) 0.66 (0.57–0.74) No 0.89 (0.68–1.00) 0.48 (0.28–0.68) 0.38 (0.17–0.59) 0.92 (0.78–1.00) 0.71 (0.50–0.92) 0.59 (0.42–0.75) CI: confidence interval, AUC: area under the curve, NPV: negative predictive value, PPV: positive predictive value, CKD: chronic kidney disease, CVD: cardiovascular diseases. In patients≤75 years group, the sensitivity was 0.92 (95% CI: the accuracy was 0.64 (95% CI: 0.54–0.73). In patients not 0.83–1.00), the specificity was 0.57 (95% CI: 0.43–0.72), the complicated with CKD group, the sensitivity was 0.90 (95% PPV was 0.62 (95% CI: 0.49–0.75), the NPV was 0.90 (95% CI: 0.79–1.00), the specificity was 0.37 (95% CI: 0.19–0.55), CI: 0.79–1.00), the AUC was 0.78 (95% CI: 0.68–0.88), and the PPV was 0.61 (95% CI: 0.47–0.76), the NPV was 0.77 the accuracy was 0.72 (95% CI: 0.63–0.82) (Table 6). (95% CI: 0.54–1.00), the AUC was 0.67 (95% CI: 0.52–0.82), and the accuracy was 0.65 (95% CI: 0.53–0.77) (Table 6). 3.6.3. Accompanied with CKD or Not. In patients accom- panied with CKD group, the sensitivity was 0.90 (95% CI: 3.6.4. Accompanied with CVD or Not. In patients accom- 0.80–1.00), the specificity was 0.51 (95% CI: 0.39–0.63), the panied with CVD group, the sensitivity was 0.90 (95% CI: PPV was 0.47 (95% CI: 0.34–0.59), the NPV was 0.92 (95% 0.82–0.98), the specificity was 0.46 (95% CI: 0.34–0.58), the CI: 0.83–1.00), the AUC was 0.74 (95% CI: 0.64–0.84), and PPV was 0.57 (95% CI: 0.46–0.67), the NPV was 0.86 (95% 12 Journal of Healthcare Engineering CI: 0.75–0.97), the AUC was 0.71 (95% CI: 0.61–0.80), and 1.0 the accuracy was 0.66 (95% CI: 0.57–0.74). In patients not 0.9 accompanied with CVD group, the sensitivity was 0.89 (95% CI: 0.68–1.00), the specificity was 0.48 (95% CI: 0.28–0.68), 0.8 the PPV was 0.38 (95% CI: 0.17–0.59), the NPV was 0.92 (95% CI: 0.78–1.00), the AUC was 0.71 (95% CI: 0.50–0.92), 0.7 and the accuracy was 0.59 (95% CI: 0.42–0.75) (Table 6). )e comparisons of the AUCs of different subgroups 0.6 delineated that the model had good predictive values for female DN patients, DN patients≤75 years, and DN patients 0.5 accompanied with CKD. )e predictive values of the model 0.5 for DN patients accompanied with CVD and DN patients 0.0 not accompanied with CVD were similar (Figure 5). 4. Discussion )is study extracted the data of 511 DN patients aged ≥65 Figure 5: )e comparisons of the AUCs of the model for different years and screened the predictors to establish a prediction subgroups. model for the mortality of DN patients within 2 years. )e results revealed that the model had good predictive ability for the mortality of DN patients within 2 years. Additionally, the predictive values of female DN patients, DN patients ≤75 patients with different gender, age, being accompanied with years, DN patients accompanied with CKD, and patients with CKD or not, and being accompanied with CVD or not. )e or without CVD were also good. )e findings of our study results revealed that the model had better predictive values for might offer a tool for identifying DN patients with high risk of female DN patients, DN patients ≤75 years, and DN patients death within 2 years and the clinicians should provide timely accompanied with CKD. )e predictive values of the model interventions to those patients to improve their outcomes. for DN patients accompanied with CVD and DN patients not )is study established a prediction model for the mor- accompanied with CVD were similar. )is indicated that the tality of elderly DN patients within 2 years. In previous model might be more suitable for female DN patients, DN prediction models for the mortality of DN patents, many patients≤75 years, and DN patients accompanied with CKD. studies were focused on evaluating the risk of renal survival in )ese results suggested that the model could benefit specific DN patients [9, 16].Our study constructed a model and patients with DN. evaluated its predictive value for all-cause mortality in DN )e impaired glomerular filtration rate (GFR) was patients. DN patients were associated with various compli- regarded as a marker of DN in DM patients [18]. A previous cations and the all-cause mortality of DN patients was high meta-analysis revealed that the impaired GFR was an in- and should be brought to attention [17]. Sato et al. [10] dependent risk factor for progressive CKD, end-stage renal established a prediction model for all-cause mortality in DN failure, and all-cause mortality in general population [19]. patients, but this model was based on only one laboratory )e eGFR-CKD-EPI is an extensively used equation for index (predialysis neutrophil-lymphocyte ratio) and the estimating GFR [20]. )e decline of eGFR-CKD-EPI was sample size was small (n = 78). In addition, internal validation associated with renal hyperfiltration and impaired GFR in was also not performed to verify the performance of the DM patients [21]. )ese supported the results in our study, model [10]. In our study, the prediction model was con- which revealed that the eGFR-CKD-EPI was a predictor for structed based on the predictors including LOS, temperature, the mortality of DN patients within 2 years. Patients with heart rate, SpO , Scr, RDW, the simplified acute physiology rapid decline of eGFR-CKD-EPI should be brought to the score-II (SAPS-II), hyperlipidemia, and eGFR-CKD-EPI, forefront and special treatments should be provided to which presented a better predictive ability compared to the prevent the mortality of DN patients. DN was associated model involving one predictor. )e sample size in this study with higher Scr levels in patients, and high Scr levels in- was larger than that in the previous study. Additionally, dicated a declining renal function [22, 23]. )is allied with the results in this study, which indicated that the Scr level internal validation was performed and it was found that the predictive value of the model for the mortality of DN patents was an important predictor for the mortality of elderly DN patients within 2 years. Clinicians should pay special at- within 2 years was good. )e prediction model in our study might provide a tool for the clinicians for quickly identifying tention to DN patients with high level of Scr. SpO is an DN patients with high risk of death and timely interventions index for oxygenation status of people and tissue hypoxia is should be provided in those patients for improving their an important contributor to diabetic complications [24]. outcomes. We also plotted a nomogram of the prediction Frequent abnormal blood oxygen in patients was reported to model based on the results from the logistic regression. )e be associated with elevated inflammation in patients [25]. nomogram can quickly and intuitively obtain the probability Herein, SpO was a predictor for the mortality of elderly DN of mortality of each patient. Meanwhile, subgroup analysis patients within 2 years. In this study, RDW was another was also conducted to evaluate the predictive values for predictor for the mortality of elderly DN patients within 2 AUC Male Female >75 years ≤75 years With CKD Without CKD With CVD Without CVD Journal of Healthcare Engineering 13 years. )is was supported by several previous studies. Zhang Conflicts of Interest et al. [26] identified that patients with DN were found to be )e authors declare that they have no conflicts of interest. with high level of RDW and RDW was associated with increased risk of progression to ESRD in patients with DN [26]. Another study also demonstrated that high level of Supplementary Materials RDW was an indicator of prognosis in DN patients and high Supplementary Table 1: sensitivity analysis of the data before level of RDW in T2D patients indicated a poor prognosis for and after the manipulation of the missing value. (Supple- DN [27]. SAPS-II is an indicator evaluating the outcomes of mentary Materials) patients in ICUs and estimating their risk of mortality [28]. SAPS-II has good power to predict the deaths in ICU, which References has been recommended for the identification and mortality prognostication of patients in ICUs [29]. In our study, SAPS- [1] K. X. Li, M. J. Ji, and H. J. Sun, “An updated pharmacological II was found to be a predictor for the mortality in ICU insight of resveratrol in the treatment of diabetic nephrop- patients with DN. High-risk patients were associated with athy,” Gene, vol. 780, Article ID 145532, 2021. longer LOS in ICUs and with higher hospital mortality [30]. [2] M. Haneda, K. Utsunomiya, D. Koya et al., “A new classifi- )e prolonged LOS in ICUs has been reported to be a risk cation of diabetic nephropathy 2014: a report from joint factor for infections, which might also increase the risk of committee on diabetic nephropathy,” Clinical and Experi- mental Nephrology, vol. 19, no. 1, pp. 1–5, 2015. death in patients [31]. )ese gave evidence to the findings in [3] M. Cao, Z. Tian, L. Zhang, R. Liu, Q. Guan, and J. Jiang, this study, showing that LOS in ICUs was a predictor for the “Effects of CCR5 59029G/A polymorphism on the risk to mortality of DN patients in ICUs. diabetic nephropathy,” Oncotarget, vol. 8, no. 63, Several limitations existed in our study. Firstly, this study pp. 106926–106934, 2017. extracted the data from MIMIC-III database, which lacked [4] Y. Wang, T. Zhou, Q. Zhang et al., “Poor renal and cardio- several important variables including the medications of DN vascular outcomes in patients with biopsy-proven diabetic patients, as well as the control of blood glucose of the nephropathy,” Kidney & Blood Pressure Research, vol. 45, subjects, and these were closely associated with the outcomes no. 3, pp. 378–390, 2020. of these patients. Secondly, external validation of the pre- [5] T. Takao, M. Suka, H. Yanagisawa, and M. Kasuga, “Com- dictive value of the model was not performed. In the future, bined effect of diabetic retinopathy and diabetic kidney dis- studies with large scale of sample size were required to ease on all-cause, cancer, vascular and non-cancer non- vascular mortality in patients with type 2 diabetes: a real- validate the findings in our study. Currently, there were world longitudinal study,” J Diabetes Investig, vol. 11, no. 5, numerous machine learning algorithms that can be used for pp. 1170–1180, 2020. predicting the mortality of elderly patients with DN. Some [6] M. Nakayama, T. Sato, M. Miyazaki et al., “Increased risk of recent studies have also used principal component analysis- cardiovascular events and mortality among non-diabetic (PCA-) firefly based deep learning model for predicting the chronic kidney disease patients with hypertensive nephrop- occurrence or the detection of diabetic retinopathy [32–34]. athy: the Gonryo study,” Hypertension Research, vol. 34, )e predictive accuracy was evidently improved using these no. 10, pp. 1106–1110, 2011. methods. Diabetic nephropathy and DN are common mi- [7] K. Mise, J. Hoshino, T. Ueno et al., “Prognostic value of crovascular complications of diabetes mellitus. In our study, tubulointerstitial lesions, urinary N-Acetyl-β-d-Glucosami- we only used logistic regression model, and, in the future, nidase, and urinary β2-microglobulin in patients with type 2 diabetes and biopsy-proven diabetic nephropathy,” Clinical PCA-firefly based deep learning model might be applied in Journal of the American Society of Nephrology, vol. 11, no. 4, our further studies to improve the predictive ability for the pp. 593–601, 2016. mortality of DN patients and achieve a better tool for the [8] Y. Q. Huang, R. Gou, Y. S. Diao et al., “Charlson comorbidity clinicians to quickly and accurately identify those with high index helps predict the risk of mortality for patients with type risk of death. 2 diabetic nephropathy,” Journal of Zhejiang University - Science B, vol. 15, no. 1, pp. 58–66, 2014. 5. Conclusion [9] S. Jiang, T. Yu, Z. Zhang et al., “Prognostic nomogram and score to predict renal survival of patients with biopsy-proven )is study established a prediction model for the mortality of diabetic nephropathy,” Diabetes Research and Clinical Prac- DN patients within 2 years based on LOS, temperature, heart tice, vol. 155, Article ID 107809, 2019. [10] H. Sato, Y. Takeuchi, K. Matsuda et al., “Pre-dialysis neu- rate, SpO , Scr, RDW, SAPS-II, hyperlipidemia, and eGFR- trophil-lymphocyte ratio, a novel and strong short-term CKD-EPI. )e model had good predictive value for the predictor of all-cause mortality in patients with diabetic mortality of elderly patients with DN within 2 years. In nephropathy: results from a single-center study,” *erapeutic addition, the model showed good predictive values for fe- Apheresis and Dialysis, vol. 21, no. 4, pp. 370–377, 2017. male DN patients, DN patients ≤75 years, and DN patients [11] J. Y. Li, R. Q. Yao, S. Q. Liu, Y. F. Zhang, Y. M. Yao, and accompanied with CKD. Y. P. Tian, “Efficiency of monocyte/high-density lipoprotein cholesterol ratio combined with neutrophil/lymphocyte ratio Data Availability in predicting 28-day mortality in patients with sepsis,” Frontiers of Medicine, vol. 8, Article ID 741015, 2021. )e data used to support the findings of this study can be [12] Y. H. Chou, W. C. Lin, and Y. M. Chen, “Clinical outcomes in obtained from the corresponding author upon request. patients with biopsy-proved diabetic nephropathy compared 14 Journal of Healthcare Engineering to isolated lupus or crescentic glomerulonephritis,” Diabetes type 2 diabetic patients,” Scientific Reports, vol. 7, no. 1, p. 2709, 2017. Research and Clinical Practice, vol. 148, pp. 144–151, 2019. [13] A. E. Johnson, T. J. Pollard, L. Shen et al., “MIMIC-III, a freely [28] M. Pohlen, N. H. )oennissen, J. Braess et al., “Patients with acute myeloid leukemia admitted to intensive care units: accessible critical care database,” Scientific Data, vol. 3, no. 1, Article ID 160035, 2016. outcome analysis and risk prediction,” PLoS One, vol. 11, no. 8, Article ID e0160871, 2016. [14] M. )oresen, “[Logistic regression – applied and applicable],” Tidsskrift for den Norske laegeforening: Tidsskrift for den [29] I. Ka˛dziołka, R. Swistek, K. Borowska, P. Tyszecki, and W. Serednicki, “Validation of Apache II and SAPS II scales at Norske laegeforening: tidsskrift for praktisk medicin, ny the intensive care unit along with assessment of SOFA scale at raekkeidsskrift for Praktisk Medicin, Ny Raekke, vol. 137, the admission as an isolated risk of death predictor,” no. 19, 2017. Anaesthesiology Intensive *erapy, vol. 51, no. 2, pp. 107–111, [15] L. Chen and S. Chen, “Prediction of readmission in patients with acute exacerbation of chronic obstructive pulmonary [30] L. R. M. d A. Maciel, O. S. Franzosi, D. S. L. Nunes et al., disease within one year after treatment and discharge,” BMC “Nutritional risk screening 2002 cut-off to identify high-risk is Pulmonary Medicine, vol. 21, no. 1, p. 320, 2021. a good predictor of ICU mortality in critically ill patients,” [16] X. Tu, N. Luo, Y. Lv, B. Wang, and Y. Li, “Prognostic eval- Nutrition in Clinical Practice, vol. 34, no. 1, pp. 137–141, 2019. uation model of diabetic nephropathy patients,” Annals of [31] F. A. S. Aleidan, H. Alkhelaifi, A. Alsenaid et al., “Incidence Palliative Medicine, vol. 10, no. 6, pp. 6867–6872, 2021. and risk factors of carbapenem-resistant Enterobacteriaceae [17] A. K. Roumeliotis, S. K. Roumeliotis, S. A. Panagoutsos et al., infection in intensive care units: a matched case-control “Association of ALOX12 gene polymorphism with all-cause study,” Expert Review of Anti-infective *erapy, vol. 19, no. 3, and cardiovascular mortality in diabetic nephropathy,” In- pp. 393–398, 2021. ternational Urology and Nephrology, vol. 50, no. 2, pp. 321– [32] W. Cao, N. Czarnek, J. Shan, and L. Li, “Microaneurysm 329, 2018. detection using principal component analysis and machine [18] KDOQI, “KDOQI clinical practice guidelines and clinical learning methods,” IEEE Transactions on NanoBioscience, practice recommendations for diabetes and chronic kidney vol. 17, no. 3, pp. 191–198, 2018. disease,” American Journal of Kidney Diseases, vol. 49, no. 2 [33] E. Mohamed, M. A. Elmohsen, and T. Basha, “Improved Suppl 2, pp. S12–S154, 2007. automatic grading of diabetic retinopathy using deep learning [19] R. T. Gansevoort, K. Matsushita, M. van der Velde et al., and principal component analysis,”vol. 2021, pp. 3898–3901, “Lower estimated GFR and higher albuminuria are associated in Proceedings of the Annual International Conference of the with adverse kidney outcomes. A collaborative meta-analysis IEEE Engineering in Medicine and Biology Society IEEE En- of general and high-risk population cohorts,” Kidney Inter- gineering in Medicine and Biology Society Annual Interna- national, vol. 80, no. 1, pp. 93–104, 2011. tional Conference, vol. 2021, IEEE, Mexico, November 2021. [20] Y. Wang, J. Zhang, G. Teng et al., “Comparison of perfor- [34] C. Bhardwaj, S. Jain, and M. Sood, “Deep learning-based mance of equations for estimated glomerular filtration rate in diabetic retinopathy severity grading system employing Chinese patients with biopsy-proven diabetic nephropathy,” quadrant ensemble model,” Journal of Digital Imaging, vol. 34, Disease Markers, vol. 2019, no. 8, Article ID 4354061, 2019. no. 2, pp. 440–457, 2021. [21] P. Bjornstad, D. Z. Cherney, J. K. Snell-Bergeon et al., “Rapid GFR decline is associated with renal hyperfiltration and impaired GFR in adults with Type 1 diabetes,” Nephrology Dialysis Transplantation, vol. 30, no. 10, pp. 1706–1711, 2015. [22] M. R. Islam, N. Sultana, S. R. Sutradhar, and M. Asaduzzaman, “Prevalence of diabetic nephropathy in patients attending the endocrine department of mymensingh medical college hospital,” Mymensingh Medical Journal, vol. 29, no. 3, pp. 530–538, 2020. [23] X. Li, T. Zhang, J. Geng et al., “Advanced oxidation protein products promote lipotoxicity and tubulointerstitial fibrosis via CD36/β-catenin pathway in diabetic nephropathy,” An- tioxidants and Redox Signaling, vol. 31, no. 7, pp. 521–538, [24] S. Garg, S. Gupta, M. S. Mobeen, and S. V. Madhu, “Effect of obesity and glycated hemoglobin on oxygen saturation in ambulatory type 2 diabetic individuals: a pilot study,” Diabetes & Metabolic Syndrome: Clinical Research Reviews, vol. 10, no. 3, pp. 157–160, 2016. [25] P. Xue, N. Covassin, X. Ran et al., “Association of parameters of nocturnal hypoxemia with diabetic microvascular com- plications: a cross-sectional study,” Diabetes Research and Clinical Practice, vol. 170, Article ID 108484, 2020. [26] J. Zhang, R. Zhang, Y. Wang et al., “)e association between the red cell distribution width and diabetic nephropathy in patients with type-2 diabetes mellitus,” Renal Failure, vol. 40, no. 1, pp. 590–596, 2018. [27] X. F. Xiong, Y. Yang, X. Chen et al., “Red cell distribution width as a significant indicator of medication and prognosis in
http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png
Journal of Healthcare Engineering
Hindawi Publishing Corporation
http://www.deepdyve.com/lp/hindawi-publishing-corporation/construction-of-a-prediction-model-for-the-mortality-of-elderly-dWh00nkey3