Diabetic Retinopathy Risk Prediction in Patients with Type 2 Diabetes Mellitus using a Nomogram Model

doi:10.21203/rs.3.rs-1565940/v1

PDF

Research Article

Diabetic Retinopathy Risk Prediction in Patients with Type 2 Diabetes Mellitus using a Nomogram Model

https://doi.org/10.21203/rs.3.rs-1565940/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted 26 Apr, 2022

You are reading this latest preprint version

Background: The primary purpose of this study is to develop diabetic retinopathy (DR) hazard nomogram in a Chinese sample of type 2 diabetes mellitus patients (T2DM).

Methods: From January 2018 to May 2020, we constructed a model containing 213 diabetic retinopathy patients, and we used the basic statistics and biochemical indicator test to assess the DR risk in T2DM patients. The patient data is used to evaluate diabetic retinopathy risk using R software and a minimal comparative contraction and selected operator (Lasso) predictive model. In Multivariable cox regression, examined the risk factors for developing diabetic retinopathy (DR) to reduce the Lasso penalty. The validation model, decision curve research, and C-index were all tested on the calibration plot. By counting probabilities, it utilized the bootstrapping methodology to check external validity.

Results:Lasso chose eight predictive variables from sixteen independent variables: illness duration, BMI, fasting plasma glucose (FPG), glycated hemoglobin Alanine aminotransferase (HbA1c), homeostatic model assessment-insulin resistance (HOMA-IR), triglyceride(TG), total cholesterol(TC), and Vitamin D3 (vitd-t3). Its c-index = 0.848 (95 percent confidence interval:0.798-0.898), indicating that the model can make accurate predictions. In interval validation, high scores (0.816) are still possible from an analysis of a diabetic retinopathy nomogram's decision curve to predict diabetic retinopathy of greater than 2%.

Conclusion: We are developing non-parametric techniques to predict the risk of diabetic retinopathy based on disease duration, BMI, FPG, HbA1c, HOMA-IR, TG, TC, and vitd-t3.

predict risk

nomogram model

type 2 diabetes mellitus

diabetic retinopathy

Diabetes mellitus (DM) is currently one of the world's fastest-growing chronic diseases. From 2015 to 2017, an epidemiological study of 31 cities and provinces in China indicated that DM was 11.%^[1]. According to the International Diabetes Federation's most recent data, the world's diabetic patients aged 20 ~ 79 year old totaled 463 million in 2019. The number of people affected by the disease is expected to reach 578 million by 2015, but even 700 million by 2045^[2]. Diabetes is known for the damage it causes to the body's microvasculature, of which high blood sugar is a crucial marker. Diabetic retinopathy is indeed a type of microangiopathy caused by diabetes. Diabetic retinopathy is becoming more common as the number of people with diabetes rises. By 2030, the number of non-proliferative diabetic retinopathy and progressive diabetic retinopathy would be 191 million and 56.3 million. Not only that, but due to a lack of awareness and understanding of diabetic retinopathy, this issue has become the leading cause of vision loss in adults aged 30–60 year, having a significant impact on people's quality of life and posing a health risk. Because diabetic retinopathy is silent at its inception, early detection of the disease's risk and early prevention and therapy are critical for lowering the disease's effect on life and social resources.

Many factors influence the onset and progression of diabetic retinopathy. Hyperglycemia, hypertension, dyslipidemia, and obesity are all risk factors that can be altered, as can smoking, anemia, a lack of health information, and poor treatment adherence. Ethnicity, family history or inheritance, diabetes onset age, type of diabetes, and diabetes duration are all constant risk factors^[3]. This paper aims to collect the essential clinical lab indicators in the clinic, develop a reliable and accurate model that can predict eye problems in T2DM patients through retroactive studies, and use it to assess the risk of disease occurrence to intervene earlier.

From January 2018 to May 2020, we studied type 2 diabetes patients whohave bee hospitalized and outpatients at Zunyi Medical University Affiliated

Hospital's Department of Endocrinology. All diabetes patients satisfied the

WHO diagnostic criteria, which included a fasting plasma glucose level of

greater than 126 mg/dl and/or an oral glucose level of 75mg two hours later. Blood glucose was greater than 200 mg/dl^[4]. Individuals who met the

following conditions were not allowed to participate: (1) people with T1DM

and other forms of diabetes, including Cushing syndrome. (2) Diabetes patientshave developed acute diabetic complications, such as ketoacidosis. (3) Pregnantand lactating women; (4)Patients who are unable to perform a fundus examination, such as those who have severe refraction in their eyes; myopia/hyperopia patients with a history of more than 3.0d; (5)Patients with a history of

fundus lesions, such as glaucoma, cataract, or other clients; (6) Patients who have used drugs that affect lipid metabolism and VitD in the previous six months. All individuals with type 2 diabetes were screened with CR-PGi (Canon).

It utilized the findings of the surgery plus evaluation by the same surgeon to predict the existence of retinopathy. It separated them into two groups based

on the fundus examination results: non-DR (n = 113) and DR (n = 100). Gender,

age, illness duration, systolic blood pressure (SBP), diastolic blood pressure (DBP), BMI, FPG, HbA1c, TG, TC, LDL-C, HDL-C, VitD-T3, and Cr were gathered from all individuals via medical records. HOMA-IR was used to assess

insulin resistance ([IR index = FPG (mmol/L) ×FINS (mU/L)/22.5 2.8) ^[5–6]。

Statistical Analysis

R (version 3.6.1; http://www.R-project.ORG) is a programming language. To forecast DR, using smallest evident shrunken choice operator (Lasso) was used to identify the risk factors. Gender, age, length of illness, SBP, DBP, BMI, FPG, HbA1c, HOMA-IR, TG, TC, LDL, HDL, VitD-T3, and Cr were being used to evaluate the data in the training set in order to select the relevant predictors in the present risk factors. The proposed methodology is then built using heterogeneous logistic regression and the cable regression model's properties. Characteristics also included frequency or P-value the with 95% confidence interval (CI). It was decided to employ bidirectional statistically significant results. If the p-value was greater than 0.05, it was statistically significant. To predict DR infection, the cohort used all of the data to build a prediction model with appropriate adjustments and graphical representations. The Harrell index was used to evaluate a DR nomogram's accuracy^[7–8]. The c-index is validated through bootstrapping. Preference curve analysis was used in the T2DM cohort to determine predicted results utilizing received a copy based on projected benefit at scene probabilities ^[9].

Characteristics of the patient: A total of 213 t2dm patients with diabetes were gathered and studied. The patients were 55.25 ± 9.34 [range 29–83 years]. Males made up 59.15 percent of the total, while females made up 40.85 percent. The findings of the fundus examination were divided into two groups: non-DR (53.05 percent) and DR (46.95 percent ). The Table 1 illustrates the essential characteristics, including clinical indicators of patients and the data for 16 hands.

Table 1

Differences Between Demographic and Clinical Characteristics of non-DR and DR Groups
Demographic characteristics	non-DR(113)	DR(100)	Total (213)
Age (years)
<50 50–70 >70	4(44.44) 107(55.15) 2(0.20)	5(55.56) 87(44.85) 8(0.80)	9(4.23) 194(91.08) 10(4.69)
Gender
Male Female	68(53.97) 45(51.72)	58(46.03) 42(48.28)	126(59.15) 87(40.85)
Disease duration(years)
<5 >=5 < 10 >=10	65(76.47) 46(40.00) 2(15.38)	20(23.53) 69(60.00) 11(84.62)	85(39.91) 115(53.99) 13(6.10)
BMI
<24 >=24 < 28 >=28	66(72.53) 37(38.14) 10(40.00)	25(27.47) 60(61.86) 15(60.00)	91(42.72) 97(45.54) 25(11.74)
SBP(mmHg)
<140 >=140	102(52.04) 11(64.71)	94(47.96) 6(35.29)	196(92.02) 17(7.98)
DBP(mmHg)
<90 >=90	99(51.83) 14(63.64)	92(48.17) 8(36.36)	191(89.67) 22(10.33)
Clinical characteristics
FPG(mmol/L)
<7.0 >=7.0	41(95.35) 72(42.35)	2(4.65) 98(57.65)	43(20.19) 170(79.81)
HbA1c(%)
<7 >=7	34(87.18) 79(45.40)	5(12.82) 95(54.60)	39(18.32) 174(81.69)
FINs(mU/L)
<5 >=5 < 20 >=20	11(78.57) 98(50.78) 4(66.67)	3(21.43) 95(49.22) 2(33.33)	14(6.57) 193(90.61) 6(2.82)
HOMA-IR
<2.8 >=2.8	33(80.49) 80(46.51)	8(19.51) 92(53.49)	41(19.25) 172(80.75)
TG(mmol/L)
<1.7 >=1.7	48(66.67) 65(46.10)	24(33.33) 76(53.90)	72(33.80) 141(66.20)
TC(mmol/L)
<5.2 >=5.2	69(53.91) 44(51.76)	59(46.09) 41(48.24)	128(60.09) 85(39.91)
HDL(mmol/L)
>1.0 <=1.0	45(50.56) 68(54.84)	44(49.44) 56(45.16)	89(41.78) 124(58.22)
LDL(mmol/L)
<2.6 >=2.6	42(58.33) 71(50.35)	30(41.67) 70(49.65)	72(33.80) 141(66.20)
VitD-T3(ng/ml)
>=30 >=15 < 30 <15	4(80.00) 81(62.31) 28(35.90)	1(20.00) 49(37.69) 50(64.10)	5(2.35) 130(61.03) 78(36.62)
Cr(umol/L)
>=90 <90	104(55.03) 9(37.50)	85(44.97) 15(52.50)	189(88.73) 24(11.27)

Variables to be chosen

The lasso model includes the 16 indexes listed above (Table 1). There was a link between the non-DR group's duration of illness, BMI, FPG, HbA1c, HOMA-IR, TG, TC, and vitd-t3 and the DR group's duration, BMI, FPG, HbA1c, HOMA-IR, TG, TC, and vitd-t3. (a and b in Fig. 1)

Notes: (a)The optimum parameter in the Lasso Regression model is chosen using a threshold of sixteen cross-validations. The partial probability variance curves and the logarithm turn are shown. To get the best estimate, use a 1-SE technique with minimum and minimum standards. (b)For each of the 16 characteristics, the LASSO logistic regression coefficients were calculated. The logarithmic of lambda is used to determine the profile. The ideal lambda is tested using a cross-validation technique, with the optimal lambda producing eight coefficients.

Model Development for Prediction

Disease duration, BMI, FPG, HbA1c, HOMA-IR, TG, TC, and vitd-t3 were examined using multivariate logistic regression analysis. Table 2 displays the results. The aforesaid independent predictors were included in a model that was created then presented in a nomogram (

Figure 2). The following were some of the DR risk factors: Disease duration( > = 5 < 10:OR = 0.3913,>=10:OR = 0.8789),BMI( > = 24 < 28,OR = 0.4057,>=280.5618),FPG( > = 7.0:OR = 0.8209),HbA1c( > = 7.0:0.6870),HOMA-IR( > = 2.8:0.5494),TG( > = 1.7:OR = 0.3887),TC( > = 5.2:OR = 0.3910),vitd-t3( > = 15 < 30:OR = 0.3746,>=30:1.4488).Each

patient's clinical index is assigned to a score, and the overall score is then

calculated to determine their risk of DR.

Table 2

Prediction Factors for DR in T2DM Patients
Intercept and Variable	β	Prediction Model Odds Ratio	P value
Intercept	-4.6498	1.0446	< 0.0001
Disease duration
< 5	Reference
>=5 < 10	1.5818	0.3913	< 0.0001
>=10	2.4730	0.8789	0.0049
BMI
< 24	Reference
>=24 < 28	0.4317	0.4057	0.2873
>=28	0.7702	0.5618	0.1704
FPG
<7.0	Reference
>=7.0	2.7807	0.8209	0.0007
HbA1c
<7.0	Reference
>=7.0	0.5691	0.6870	0.4074
HOMA-IR
< 2.8	Reference
>=2.8	0.4423	0.5494	0.4208
TG
<1.7	Reference
>=1.7	0.3866	0.3887	0.3200
TC
< 5.2	Reference
>=5.2	0.5409	0.3910	0.1665
VitD-T3
<15	Reference
>=15 < 30	-0.8729	0.3746	0.0198
>=30	-0.9169	1.4488	0.5268

The accuracy of cohort DR exposure nomogram: The C-index in evaluating the occurrence of macular degeneration in type 2 diabetes individuals was 0.848 (95 percent confidence interval: 0.798–0.898) in this group, demonstrating a high validity (Fig. 3). The result of the bootstrap verification is 0.816. according to the results,the model has a good prediction accuracy.

Notes: On the x-axis, the possibility for hyperhidrosis is calculated. The diagnostics of High-Z-Diagnosis are represented on the x-axis. A perfect prediction model describes an ideal forecast. The graph depicts the fitness of the nomogram for forecasting outcomes, with a producing specific to a dotted line indicating a more reliable prediction.

Clinic Application

The DR nomogram is made up of scales for several variables used to calculate the likelihood of a given result. According to the judgment curve, the non-adherence nomogram's use raises the projected chance of DR infection if the patient and doctor thresholds are > 2 percent and 85 percent, respectively. The overlaps are compiled in this study to ensure an equal net gain. (Fig. 4).

Form the Nomograms model intends to predict tumor diagnostic outcomes and therapeutic effects. So this model is simple, quick, cheap, and noninvasive; it is already being used in various medical professions to better monitor patients and make correct clinical treatment decisions^[10]. A study of databases found that only a few studies have attempted to forecast the risk of DR patients. This study gathered clinical and demographic statistics for hospitalized T2DM patients to test a new prediction based on the probability of acquiring DR.

Because of its increased sensitivity and specificity, neither fundus photography is currently routinely employed in clinical settings for DR screening^[11–12]. On the other hand, screening is merely an evaluation of the outcome; it does not reveal the components that play a part in creating the impact. Almost all know that diabetic retinopathy is one of the long-term effects of the disease. Although it is widely accepted that diabetes is a potential risk for DR, are there any other factors that lead to the growth of DR? What is the significance of the relationship? By constructing the Nomograms model, all these may be studied and judged.

Overweight, lengthy diabetes, hypoglycemic, hypertension, high cholesterol, kidney disease, renal failure (DKD), pregnancy, and susceptibility genes are all common triggers for DR, according to studies like the DCCT and UKPDS^[13–14]. Gender, age, illness length, SBP, DBP, BMI, FPG, HbA1c, TG, TC, HOMA-IR LDL-C, HDL-C VitD-T3 and Cr, which are clinically available clinical indicators, were all correlated using the LASSO method. The six hands of illness time, as well as BMI, FPG, HbA1c, HOMA-IR, TG, TC, and Vitamin D, were all in agreement with the findings. Except for the length of the disease, all of these risk factors are modifiable; in other words, the value of our model is the identification and management of such modifiable risk factors.

The system that links risk for the pathogenesis of DR has recently been identified as hyperglycemia and disease duration^[15–16]. The prediction model we built backs this up, with greater risk levels for fasted glucose and disease progression. On the other hand, DR is a metabolic disorder that is difficult to treat. DR does not develop in all patients with reasonable glycemic control. Some with poor glycemic control, on the other hand, are more likely to create DR, implying that there are additional secondary contributing factors. According to specific research, the probability of having diabetic retinal development increases by around 64% per each 10percentage increase in HbA1c, and the two have a positive relationship^[17]. Lower higher serum levels were observed may decrease the risk of severe blindness by 47 percent when compared to the regular group after approximately 20 years of follow-up, according to some researchers^[18]. Not only that but "lipotoxicity," or the damage to the retinal barrier caused by excessive blood lipids, exceptionally high triglycerides, is also a vital part of DR^[19]. Controlling dyslipidemia, in addition to glycemic management, is critical for preventing and treating DR^[20]. HbA1c and lipids could be risk factors for DR, according to the independent variables assessed in our prediction model. Lower blood glucose and controlling lipids are two of the most critical preventive and therapy methods for DR.

A search of relevant databases showed that the nomogram model had been used to predict the risk of diabetic retinopathy^[9]. Still, this model yielded a higher C-index value, indicating a higher accuracy. The indicators included in this model are more comprehensive than the previous model, making it a more accurate predictor of risk.

Our research includes popular VD from recent years that other researchers have not used.Low VD levels are a specific and sensitive sign of proliferative disease. They are adversely connected with the intensity of DR. In DR situations, AUC recommends VD as a straightforward, sensitive, and specific laboratory test^[21]. According to a foreign cross-sectional investigation, patients with VD insufficiency were more likely than those with sufficient VD to acquire DR. Multidisciplinary ordinal regression analysis revealed a link between VD shortage and DR severity^[22]. This study also found that VD rates can be seen as a predictor of DR, and the incidence of DR increases with decreased VD levels, which is in line with the researchers' earlier findings. Because NO acts as a vascular and reduces vasoactivity, VD shields against the production of DR, which may be associated with NO modulation ^[23]. However, while VD deficit directly linked all-cause survival, it did not predict the development of microvascular complications, according to a prospective observational study^[24].

However, there are certain flaws in this research. This study had a tiny sample, but it was a long-term assessment to evaluate the risk of DR that didn't account for the bias caused by patients' medications as DR progressed. Furthermore, because lighting affects VitD-T3, our work could not collect data from patients during a continuous daytime period. As previously said, the risk or the occurrence of DR is not uniform, so we will conduct a study of type 2 diabetes patients and incorporate more indicators to derive more preventive methods for DR prevention.

By gathering clinical data widely obtainable in the field and creating a Nomograms model, this study assessed the risk to DR in Type 2 Diabetic patients. Colman's plot calculated the chances of developing DR in Patients with type 2 diabetes based on clinical indicators. It steered doctors and patients in developing personalized treatment plans based on these risk factors to eliminate the hazard of DR and avoid the onset and development of DR.

Acknowledgements

Not applicable.

Authors’ contributions

Q.W., N.Z. and H.T.contributed to the conception and design of the study. Q.W., N.Z.,X.Y., Q.Y.,L.Z.,H.Z.,Y.Z. and X.N.She contributed to data collection, analysis, or interpretation. Q.W. and N.Z.wrote the article. X.L. revised the manuscript. The final articles were read and approved by all participants before submission.Q.W. and N.Z. contributed equally to this work.

Funding

Thank you to our colleagues at Zunyi Medical College's Affiliated Hospital for helping me collect data. This research was funded in part by the Chinese National Science Foundation (82060273), as well as the Guizhou Technology and Science Program (Qian Ke He Base -ZK[2021]- General 407). The article's completion is entirely at the discretion of the funders.

Availability of data and materials

The supporting data can be acquired via correspondence author.

Ethics approval and consent to participate

All human research investigations (individuals, samples, or 213 data) are conducted according to the Declaration of Helsinki's standards. The Ethical Committee Institutional Review Board approved this study of the affiliated Zunyi Medical University Hospital (permission no [KLL-2021-329]).Since this study is a retrospective study and the data were analysed anonymously, informed consent from the subjects was not required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Li Y, Teng D, Shi X, et al. Prevalence of diabetes recorded in mainland China using 2018 diagnostic criteria from the American Diabetes Association: national cross sectional study[J]. BMJ, 2020, 369:m997. DOI:10.1136/bmj.m997
International Diabetes Federation, IDF Diabetes Atlas, International Diabetes 297 Federation, Brussels, Belgium, 9th edi
tion, 2019.
Jenkins AJ, Joglekar MV, Hardikar AA, Keech AC, O’Neal DN, Januszewski AS. Biomarkers in diabetic retinopathy. Rev Diabet Stud 2015; 12: 159–95.DOI:10.1900/RDS.2015.12.159.
Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: Diagnosis and
classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 1998; 15: 539–553.DOI:10.1002/(SICI)1096-9136(199807)15:7<539::AID-DIA668>3.0.CO;2-S
V. P. Balachandran, M. Gonen, J. J. Smith, and R. P. DeMatteo.Nomograms in oncology: more than meets the eye[J].The Lancet Oncology, 2015;16(4): e173–80.DOI: 10.1016/S1470-2045(14)71116-7
Jiang F, Wei K, Lyu W, et al. Predicting risk of insulin resistance in a Chinese population with polycystic ovary syndrome: designing and testing a new predictive nomogram[J]. Biomed Res Int. 2020;2020:1–9.DOI:10.1155/2020/8031497
Kramer AA, Zimmerman JE. Assessing the calibration of mortality benchmarks in critical care: the Hosmer-Lemeshow test revisited[J].
Crit Care Med.2007;35(9):2052–2056. DOI:10.1097/01.CCM.0000275267.64078.B0
Vickers AJ, Cronin AM, Elkin EB, et al. Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers[J].BMC Med Inform Decis Mak.2008;8:53. DOI:10.1186/1472-6947-8-53.
Ruohui Mo, Rong Shi, Yuhong Hu, and Fan Hu Nomogram-Based Prediction of the Risk of Diabetic Retinopathy: A Retrospective Study[J].Journal of Diabetes Research. 2020;2020:9–17.DOI:10.1155/2020/7261047.
Zeng N,Li Y,Wang Q,et al.Development and Evaluation of a New Predictive Nomogram for Predicting Risk of
Herpes Zoster Infection in a Chinese Population with Type 2 Diabetes Mellitus[J].Risk Manag Healthc Policy.2021;14:4789–4797.DOI: 10.2147/RMHP.S310938
Ahmed J, Ward TP, Bursell SE, et al. The sensitivity and specificity of nonmydriatic digital stereoscopic retinal imaging in detecting diabetic retinopathy[J]. Diabetes Care, 2006, 29(10):22052209. DOI: 10.2337/dc060295.
Solomon SD, Chew E, Duh EJ, et al. diabetic retinopathy: a position statement by the American Diabetes Association [J]. Diabetes Care, 2017, 40(3): 412418. DOI: 10.2337/ dc162641.
Song SJ, Wong TY. Current concepts in diabetic retinopathy[J]. Diabetes Metab J, 2014, 38(6): 416425. DOI: 10.4093/dmj.2014.38.6.416.
Kuo JZ, Wong TY, Rotter JI. Challenges in elucidating the genetics of diabetic retinopathy[J]. JAMA Ophthalmol, 2014, 132(1):96107. DOI: 10.1001/jamaophthalmol.2013.5024.
Ting DS, Cheung GC, Wong TY. Diabetic retinopathy: global prevalence,major risk factors, screening practices and public health challenges: areview[J]. Clin Exp Ophthalmol. 2016;44:260–77.DOI: 10.1111/ceo.12696
Tan GS, Ikram MK, Wong TY. Traditional and novel risk factors of diabetic retinopathy and research challenges[J]. Curr Med Chem. 2013;20:3189–99.DOI: 10.2174/09298673113209990023
Wat N, Wong R, Wong Y. Associations between diabetic retinopathy and systemic risk factors[J]. Hong Kong Academy of Medicine. 2016, 22(6):589–599.DOI: 10.12809/hkmj164869
Gong Q, Gregg EW, Wang J, et al. Longterm effects of a randomised trial of a 6year lifestyle intervention in impaired glucose tolerance on diabetesrelated microvascular complications: the China Da Qing Diabetes Prevention Outcome Study[J]. Diabetologia, 2011, 54(2): 300307. DOI: 10.1007/s0012501019489.
Renu A.Kowluru RA.Retinopathy in a Diet-Induced Type 2 Diabetic Rat Model and Role of Epigenetic Modifications
[J].Diabetes,2020,69(4):689–698. DOI: 10.2337/db19-1009
Fengqi Guo, Ting Zhou, Juan Tang, Mingxia Dong, Qianping Wei.Related Risk Factors between Subclinical Carotid Atherosclerosis and Diabetic Retinopathy in Newly Diagnosed Type 2 Diabetes Mellitus in China[J].Exp Clin Endocrinol Diabetes.2021;129(4):283–288.DOI:10.1055/a-0865-1754
Gauhar Nadri1, Sandeep Saxena1*, Abbas Ali Mahdi2, Apjit Kaur1, Md. Kaleem Ahmad2, Pragati Garg3 and Carsten H. Meyer4 Nadri et al.Serum vitamin D is a biomolecular biomarker for proliferative diabetic retinopathy[J]. Int J Retin Vitr,2019;5:31.DOI: https://doi.org/10.1186/s40942-019-0181-z.
Long M, Wang C, Liu D. Glycated hemoglobin A1C and vitamin D and their association with diabetic retinopathy severity[J]. Nutr Diabetes,2017;7:e281.DOI: 10.1038/nutd.2017.30.
Opatrilova R, Kubatka P, Caprnda M, Büsselberg D, Krasnik V,Vesely P, et al. Nitricoxide in the pathophysiology of retinopathy:evidences from preclinical and clinical researches[J]. Acta Ophthalmol,2018;96(3):222–31.DOI: 10.1111/aos.13384
Joergensen C, Hovind P, Schmedes A, Parving HH, Rossing P.Vitamin D levels, microvascular complications, and mortality in type 1 diabetes[J]. Diabetes Care. 2011;34:1081–5.DOI: 10.2337/dc10-2459

No competing interests reported.

PDF

Version 1

posted 26 Apr, 2022

You are reading this latest preprint version

Diabetic Retinopathy Risk Prediction in Patients with Type 2 Diabetes Mellitus using a Nomogram Model

Status:

Version 1

Abstract

Figures

Introduction

Patients And Methods

Statistical Analysis

Results

Discussion

Conclusion

Declarations

References

Competing Interests

Status:

Version 1