Title Science and Technology Indonesia e-ISSN:2580-4391 p-ISSN:2580-4405 Vol. 8, No. 1, January 2023 Research Paper The Risk Cluster in Type 2 Diabetes Mellitus Based on Risk Parameters Using Fuzzy C-Means Algorithm Marhamah1, Sugiyarto Surono2*, Endang Darmawan1 1Department of Pharmacology and Clinical Pharmacy, Faculty of Pharmacy, Ahmad Dahlan University, Yogyakarta, 55164, Indonesia2Department of Mathematics, Faculty of Applied Science and Technology, Ahmad Dahlan University, Yogyakarta, 55191, Indonesia *Corresponding author: Sugiyarto@math.uad.ac.id AbstractThe prevalence of type 2 diabetes mellitus increases every year. In the long term, type 2 diabetes mellitus can lead to complicationsof other diseases. This study aimed to analyze the risk cluster for type 2 diabetes mellitus based on risk parameters using the FuzzyC-Means algorithm. The benefit of analyzing the risk cluster as an initial screening to prevent the occurrence of type 2 diabetesmellitus. This study used 905 subjects’ data consisting of 562 males and 343 females. After the data preprocessing, the optimalnumber of clusters was determined using a Fuzzy C-Means algorithm process. Subsequently, the Pearson correlation test wasconducted to determine the correlation between the risk parameters of type 2 diabetes mellitus and the cluster results. The studyresulted in 2 risk clusters, subjects in cluster 1 were older than 60 years (34.1%), had a family history of type 2 diabetes mellitus(62.7%), had hypertension (55.4%), routinely took medicines (73.5%), undertook physical activity for less than half an hour (40.5%),and had a high blood pressure level (53.5%). The Pearson correlation test found that age, regular medication use, hypertension andblood pressure level all seem to have significant correlations with cluster outcomes. The risk cluster of type 2 diabetes mellitus wasseparated into two clusters using Fuzzy C-Means algorithm, namely the high-risk cluster and the low-risk cluster. KeywordsType 2 Diabetes Mellitus, Cluster Analysis, Fuzzy C-Means Algorithm, Pearson Correlation Received: 10 August 2022, Accepted: 10 November 2022 https://doi.org/10.26554/sti.2023.8.1.17-24 1. INTRODUCTION Diabetes mellitus is a chronic condition that occurs due to increase blood glucose levels, resulting from insucient pro- duction of insulin hormones or insulin that does not work opti- mally due to damage to pancreatic beta cells (Sarría-Santamera et al., 2020). The risk factors are lack of physical activity, fam- ily history of diabetes mellitus, certain racial/ethnic groups, gestational diabetes history, hypertension history, high density lipoprotein (HDL) 150 mg/dL, smoking, prediabetes history, cardiovascular disease history, and age >45 years without the previously mentioned risk factors (Association, 2021). The symptoms include polyuria, polyphagia, polydipsia, and unex- plained weight loss (Virani et al., 2021). Type 2 diabetes mellitus has been identied as the world’s leading cause of mortality (Pham et al., 2020). In 2019, the International Diabetes Federation estimated that 463 million adults aged 20-79 years had diabetes, and this number was predicted to reach 578.4 million and 700.2 million in 2030 and 2045 worldwide, respectively (Saeedi et al., 2019). Type 2 diabetes mellitus could cause complications such as heart at- tack, stroke, neuropathy, nephropathy, retinopathy, and kidney failure (Mezil, 2021). Early screening for the risk of type 2 diabetes mellitus should be conducted to prevent the disease. Previous studies had predicted the risk of type 2 diabetes mellitus using classi- cation techniques (Logistic Regression, K-Nearest Neighbor, Support Vector Machine, Decision Tree Classication, and Random Forest Classication) (Tigga and Garg, 2020). This study used a dierent approach-cluster analysis-to identify the type 2 diabetes mellitus risk group. The ability of cluster anal- ysis to categorize substantial amounts of data and a variety of variables was one of its benets (Muslimatin, 2011). Cluster analysis divides similar objects into a group and diers from others based on distance measures (Madhulatha, 2012). One of the methods in cluster analysis was the Fuzzy C-Means algo- rithm, a clustering method in which the level of data presented was determined by the degree of membership introduced, the membership value allows it to be between 0 and 1 (Ramya, 2018). Fuzzy C-Means clustering have been used to predict the level of diabetes mellitus, so that it can assist medical per- https://crossmark.crossref.org/dialog/?doi=10.26554/sti.2023.8.1.17-24&domain=pdf https://doi.org/10.26554/sti.2023.8.1.17-24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Figure 1. Flow Chart of Methods sonnel in determining the right treatment therapy for patients (Jamuna, 2020). Kunwar et al. (2019) used Fuzzy C-Means method to assist clinical decision-making regarding kidney fail- ure which showed there were 7 clusters. The existence of this grouping is an eort to prevent early kidney failure so that timely treatment can be given and reduce the risk of death. Subsequently, Sanakal and Jayakumari (2014) performed a diabetes mellitus prognosis by comparing Fuzzy C-Means al- gorithm with Support Vector Machine (SVM), showing the Fuzzy C-Means algorithm accuracy of 94.3%, but was 59.5% for SVM. Lomo et al. (2021) used Fuzzy C-Means clustering to identify patients with type 2 diabetes mellitus, based on de- mographic factors, blood glucose levels, the patient’s surviving condition, and medication. It revealed that there were three clusters, each representing a patient’s condition that helped individuals with type 2 diabetes mellitus live longer. Therefore, this study aimed to perform a risk cluster analysis of type 2 diabetes mellitus by identifying the characteristics of the risk parameters using the Fuzzy C-Means algorithm. Compared to earlier studies, more and dierent variables were used in this cluster analysis study to assess the risk of type 2 diabetes mellitus. Some variables were used to test the risk of type 2 diabetes mellitus, as used by the Association (2021). 2. EXPERIMENTAL SECTION 2.1 Materials This study used secondary data obtained from kaggle datasets of diabetes owned by Tigga with 952 subjects aged >18 years old, female and male (Tigga and Garg, 2020). They were asked to answer a questionnaire containing questions about theirhealth, lifestyle, andfamilybackground, as showninTable 1 (Tigga and Garg, 2020). After the blank data were removed, 905 subjects were obtained, consisting of 562 males and 343 females. 2.2 Methods The data were analyzed using the Python program with the followed stages (See Figure 1) : 1. Read data The data is read used pandas library in Python. 2. Data preprocessing a. Data Conversion from categorical to numerical Data in the categorical form should rst be converted into numerical form. The categorical data contained in the parameters of age, gender, family history of diabetes, hypertension, physical activity, smoking, alcohol con- Figure 2. Flow Chart of Research Subject Selection sumption, regularly taking medication, consumption of fast food, stress, blood pressure level, gestational diabetes history, and frequency of urination. b. Blank data Elimination Blank data were excluded from this study, yielding 905 nal results (See Figure 2). c. Data normalization DatanormalizationwasconductedusingZ-score tomake each parameterhad adistribution in the same range. The data normalization formula is as follows (Gökhan et al., 2019): Z = (X − `) 𝜎 (1) Where: x = observed value ` = average 𝜎 = standard deviation Z = Z-score (Raw Value) d. The determination of the optimal number of clusters The elbow method was used to analyze and determine theoptimaldataset cluster. Itbeganbyplotting thevalues resulting from the function of the number of clusters and marking them at the elbow of the curve. Subsequently, the curve provided information about the number of clusters used. For example, when the value of the rst and second gave the angle on the graph or the value that had decreased the most, the number can be used (Kurniawan et al., 2020). 3. Data clustering using the Fuzzy C-Means algorithm. 4. Analyzing the results of clusters to determine the char- acteristics or distribution of each parameter from each obtained cluster. 5. Doing a correlation test on the cluster results to measure thecorrelationbetweentheparametersused inthecluster and the cluster results. The interpretation of Pearson correlation coecient is shown in Table 2 (Hinkle et al., 2003). © 2023 The Authors. Page 18 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Table 1. Risk Parameters of Type 2 Diabetes Mellitus of the Subjects (Tigga and Garg, 2020) Parameters Category Age < 40 years 40-49 years 50-59 years >60 years Gender Male Female Family history Yes/No Hypertension Yes/No Physical activity None Less than half an hour More than half an hour One hour or more Smoking Yes/No Alcohol consumption Yes/No Routinely taking drugs Yes/No Fast food consumption Occasionally Often Very Often Always Stress Not at all Sometimes Very often Always Blood pressure level Low/Normal/High Gestational diabetes history Yes/No Urination frequency Not much/Quite often Body Mass Index (BMI) Numeric (15-45 kg/m2) Sleep time Numeric (4-11 hours) Deep sleep Numeric (0-11 hours) Pregnancies Numeric (0-4) Table 2. The Interpretation of Pearson Correlation Coecient (Hinkle et al., 2003) Size of Correlation Interpretation .90 to 1.00 (-.90 to -1.00) Very high positive (negative) correlation .70 to .90 (-.70 to -.90) High positive (negative) correlation .50 to .70 (-.50 to -.70) Moderate positive (negative) correlation .30 to .50 (-.30 to -.50) Low positive (negative) correlation .00 to .30 (.00 to -.30) Negligible correlation © 2023 The Authors. Page 19 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Figure 3. The Results of the Cluster of Subjects at Risk of Type 2 Diabetes Mellitus Using the Fuzzy C- Means Algorithm (Cluster 1 is Depicted in Blue and Cluster 2 is Depicted in Red) 3. RESULT AND DISCUSSION 3.1 Strand Geometry The risk parameters of the subject in this study are described in Table 1. Figure 3 shows the data distribution plot of risk cluster results, cluster 1 is depicted in blue and cluster 2 is depicted in red. The optimal number of risk clusters of type 2 diabetes mellitus obtained was 2 clusters. Cluster 1 consisted of 370 subjects, while cluster 2 consisted of 535 subjects. The risk cluster results using the Fuzzy C-Means method can be seen in Table 3 and Table 4. Cluster 1 was dominated by subjects with 34.1% aged > 60 years, with a family history of type 2 diabetes mellitus (62.7%), who have hypertension (55.4%), who regularly take drugs (73.5%), who did physical activity for less than half an hour (40.5%), and who have a high blood pressure level (53.5%). In addition, Cluster 1 predominantly consisted of subjects with an average body mass index (BMI) of 27.84 kg/m2 and the average sleep time was 6.75 hours. Cluster 2 was dominated by subjects under the age of 40 (77.0%), who did not have a family history of type 2 diabetes mellitus (62.4%), had no history of hypertension (97.2%), did not frequently take medicines (90.5%), did more than half an hour of physical activity (31.0%), and had a normal blood pres- sure level category (92.7%). Furthermore, the average sleep time was 7.09 hours and the BMI was 23.9 kg/m2. The Pear- son correlation test showed the highest correlation value with the clusterresults on the age parameter (-0.67), regularly taking medication (-0.66), hypertension (-0.60), and blood pressure level (-0.59). Ourstudyshowed that cluster1 had asignicant probability of acquiring diabetes mellitus, because many of the subjects in this cluster had diabetes mellitus risk parameters. The risk parameters were being aged >60 years, having a family history of diabetes, having hypertension, taking medication regularly, doing physical activity for less than half an hour, having high blood pressure levels, having an average BMI of 27.84 kg/m2, and sleeping less than 6.75 hours. In this study, subjects aged more than 60 years old were in the cluster with a high-risk of developing diabetes mellitus based on Fuzzy C-Means clustering. These results are consis- tent with previous studies which reported that the prevalence of diabetes mellitus increased at 55- 64 years of age and de- creased after passing this range. Therefore, aged people were at high risk of developing diabetes mellitus (Association, 2021). Asiimwe et al. (2020) showed that the age group of 61–65 years was strongly inuenced by diabetes mellitus. The meta- analysis results also showed an increase in the likelihood of diabetes mellitus at age 40 years compared with those aged <40 years (Asamoah-Boaheng et al., 2019). Aging can increase the occurrence of chronic inammation. In addition, there is an increase in the concentration of free fatty acids in the blood due to impaired lipid metabolism in the elderly. This can lead to insulin resistance (Ismail et al., 2021). In several studies, a family history of diabetes mellitus has beenlinkedtoanincreasedriskofdevelopingtype2diabetes. A cross sectional study conducted by Zhang et al. (2015) showed that there was a signicant increase in the prevalence of dia- betes mellitus in study participants who had two generations of rst-degree relatives with diabetes with a history of diabetes by 32.7%, then only one generation of rst-degree relatives with a history of diabetes by 20.1% and no rst-degree relatives with diabetes at 8.4%. The rst-degree relatives of diabetes mainly had 𝛽-cell dysfunction, and that the higher the family history risk category, the more severe the 𝛽-cell dysfunction. Gopalakrishnan and Geetha (2017) reported that almost 68.8% of patients with diabetes mellitus had a family history of dia- betes in which mothers with diabetes had a greater inuence than fathers with diabetes. The meta-analysis conducted by Asamoah-Boaheng et al. (2019) showed that a family history of diabetes mellitus had a signicant relationship with the oc- currence of diabetes mellitus in a person. Based on hypertension parameters, cluster 1 was domi- nated by subjects with a history of hypertension. Study showed that the prevalence of hypertension in patients with type 2 diabetes mellitus is 59.5% higher in the 50-60-years age group (Akalu and Belsti, 2020). Wei et al. (2011) reported a sig- nicant relationship between the incidence of type 2 diabetes mellitus and hypertension compared to normal blood pressure in white. The study by Kim et al. (2015) showed that diabetes mellitus was higher in initial subjects with pre-hypertension and hypertension than those with normal blood pressure and subjects who had normal blood pressure at the beginning of the examination later. After eight years, prehypertension or hypertension was at a signicantly higher risk of developing diabetes mellitus in subjects compared to those with controlled blood pressure. Wangetal. (2018) reportedthatmoderate levelsofphysical activity (30 minutes/3 days) had a signicant relationship with the occurrence of type 2 diabetes mellitus. The study by Sim- © 2023 The Authors. Page 20 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Table 3. The Results of the Cluster of Subjects at Risk of Type 2 Diabetes Mellitus Using the Fuzzy C- Means Algorithm Based on Categorical Parameters Parameters Cluster Results Pearson correlation (r)Cluster 1 Cluster 2 Total n = 370 % n = 535 % n = 952 Age <40 years 50 13.5 412 77.0 462 -0.67 40-49 years 70 18.9 83 15.5 153 50-59 years 124 33.5 24 4.5 148 >60 years 126 34.1 16 3.0 142 Gender Male 207 55.9 355 66.4 562 0.11 Female 163 44.1 180 33.6 343 Family history Yes 232 62.7 201 37.6 433 0.25 No 138 37 334 62.4 472 Hypertension Yes 205 55.4 15 2.8 220 -0.60 No 165 44.6 520 97.2 685 Physical activity None 67 18.1 62 11.6 129 0.15 Less than half an hour 150 40.5 167 31.2 317 More than half an hour 85 23.0 166 31.0 251 One hour or more 65 18.4 140 26.2 208 Smoking Yes 40 10.8 66 12.3 106 0.023 No 330 89.2 469 87.7 799 Alcohol consumption Yes 99 26.8 88 16.4 187 -0.13 No 271 73.2 447 83.6 718 Routinely taking drugs Yes 272 73.5 51 9.5 323 -0.66 No 98 26.5 484 90.5 582 Fast food consumption Occasionally 290 78.4 343 64.1 633 0.1 Often 44 11.9 132 24.7 176 Very Often 20 5.4 32 6.0 52 Always 16 4.3 28 5.2 44 Stress Not at all 39 10.5 92 17.2 131 -0.34 Sometimes 168 45.4 362 67.7 530 Very often 20 23.0 73 13.6 158 Always 78 21.1 8 1.5 86 Blood pressure level Low 0 0 27 5.0 27 -0.59 Normal 172 46.5 496 92.7 668 High 198 53.5 12 2.2 210 Gestational diabetes history Yes 10 2.7 4 0.7 14 -0.078 No 360 97.3 531 99.3 891 Urination frequency Not much 208 56.2 434 81.1 642 -0.27 Quite often 162 43.8 101 18.9 263 © 2023 The Authors. Page 21 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Table 4. The Results of the Cluster of Subjects at Risk of Type 2 Diabetes Mellitus Using the Fuzzy C- Means Algorithm Based on Numeric Parameters Parameters Cluster 1 (n=370) (mean) Cluster 2 (n=535) (mean) Total subject (n=905) Pearson correlation (r) BMI (kg/m2) 27.8 23.9 905 -0.37 Sleep time (hours) 6.75 7.09 905 0.13 Deep sleep (hours) 5.05 5.88 905 0.22 Pregnancies 0.60 0.23 905 -0.19 bolon et al. (2020) showed that individuals with less physical activity were more likely to suer from type 2 diabetes mellitus than very active individuals. Suppose a person’s physical activ- ity was lacking; an imbalance in the amount of energy in the body occurred since the energy consumed was greater than the amount expended. The energy that entered the body and went unused was stored mainly as adipose tissue, triggering insulin resistance, causing type 2 diabetes mellitus. Furthermore, lack of physical activity coupled with the consumption of carbo- hydrates, proteins, and fats were obesity factors that resulted in an increase in free fatty acids. Therefore, it reduced the translocation of glucose transporters to the membrane plasma and causes insulin resistance. So, subjects with less physical activity were more at high-risk of developing type 2 diabetes mellitus. In this study, subjects taking medicine regularly were cate- gorized as having a higher risk of developing diabetes mellitus. Diabetes mellitus was sometimes associated with comsump- tion some drugs. Some drugs could increase the occurrence of type 2 diabetes mellitus associated with reduced insulin pro- duction, reduced insulin sensitivity, or the occurrence of both conditions. One example of a drug-associated with the risk of diabetes mellitus was glucocorticoids, with a mechanism that could reduce insulin production and sensitivity (Repaske, 2016). Based on the results of a study involving subjects with autoimmune diseases, the overall risk of developing diabetes afterone yearwas 0.9% when glucocorticoids were not received. In contrast, the risk increased to 2.1% for prednisolone <5 mg (daily dose) when glucocorticoids were administered and 5.0% for prednisone 25 mg (daily dose) (Wu et al., 2020). Further- more, the thiazide drug class could also increase the risk by reducing insulin production (Repaske, 2016). There was an increase in fasting blood glucose in hypertensive patients given thiazide diuretics compared to those that did not receive thi- azide diuretics. However, with low doses of thiazide diuretics (25 mg/day) given thiazide diuretics (hydrochlorothiazide or chlorthalidone), there were smaller changes in fasting blood glucose compared with higher doses in hypertensive patients Zhang and Zhao (2016) Another study showed that taking an- tibiotics within 90 days had a higher risk of developing diabetes mellitus than those who did not use them. The use of ve or more classes of antibiotics was also at higher risk than those us- ing only one class of antibiotics (Park et al., 2021). Patients are expected to discontinue the medication when diabetes mellitus is associated with the consumption of drugs (Repaske, 2016). The subjects with a high blood pressure level in this study were included in the high-risk of diabetes mellitus. The results of a prospective study showed that the systolic blood pressure level in the standard, prehypertension, mild hypertension and moderate/severe hypertension groups had a risk of 19%, 30%, 31%, and 49% of having diabetes mellitus during the follow-up period without being inuenced by risk factors such as BMI and others (Stahl et al., 2012). The higher average BMI indicates having a higher risk of developing diabetes mellitus. Pinidiyapathirage et al. (2013) reported that a high BMI is associated with an increase in fasting plasma glucose levels. The study by Liyanage (2018) showed a signicant relationship between BMI and diabetes mellitus. The majority (51%) of study participants who were overweight experienced diabetes mellitus, followed by 29.40% of normal weight and 7.80% of underweight. According to Tang et al. (2021) the risk ratio for diabetes mellitus was 2.13 times for BMI 22.5- <25.0 kg/m2, 2.14 times for BMI 25.0- <27.5 kg/m2, 3.17 times for BMI 27.5- <30.0 kg/m2, and 3.14 times for BMI ≥30,0 kg/m2. Cluster 1 has a higher risk of developing type 2 diabetes mellitus on sleep time and deep sleep parameters. The meta- analysis results showed that the risk of type 2 diabetes was low at 7-8 hours of sleep per day. Sleep durations shorter and longer than usual were associated with a signicantly in- creased risk of type 2 diabetes mellitus. A decrease in one hour was associated with 9%, and an increase of one hour was associated with a 14% increase in the risk of type 2 diabetes mellitus (Shan et al., 2015). Furthermore, lack of sleep also aect leptin and ghrelin hormones. The hormone leptin helps provide a feeling of fullness, while ghrelin helps increase ap- petite. Insucient sleepdurationdecreases leptinhormoneand increases ghrelin. Furthermore, lack of sleep can reduce pep- tide tyrosine-tyrosine and glucagon like peptide-1 hormones that help suppress hunger; hence, these changes increase ap- petite, cause obesity, and interfere with blood glucose control (Sakamoto et al., 2018). We found that the diabetes mellitus risk parameters corre- lated with the obtained risk cluster were age, regularly taking medication, hypertension, and blood pressure. The obtained Pearson correlation coecients between the risk parameters and the risk clusters are categorized as moderate correlation (Hinkle et al., 2003). © 2023 The Authors. Page 22 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 According to Steyn et al. (2004) the risk parameters of type 2 diabetes mellitus are divided into 2, namely modiable and non-modiable risk parameters. We can start prevention as earlyas possible to reduce the modiable risk parameters of the occurrence or development of type 2 diabetes mellitus, such as changing lifestyle (Uusitupa et al., 2019). These eorts include regular physical activity such as brisk walking for 30 minutes, which when done ve times a week will help increase insulin sensitivity in the body (Association, 2021; Steyn et al., 2004). Thenmaintainanidealbodyweightbymanagingmealportions and eating foods that have balanced nutrition, namely fruits, vegetables, and whole grains that can make you full longer, accompanied by physical activity, because obesity can increase the risk of type 2 diabetes mellitus (Uusitupa et al., 2019). Another eort is to lead a healthy lifestyle by not smoking and not consuming alcohol because both will reduce insulin sensitivity. Too little or too much sleep is also not good because it can increase the risk of type 2 diabetes mellitus, so sucient sleep duration is needed, which is 7-8 hours per day and is also benecial in improving concentration, memory, and mood. In addition, as an eort to prevent and detect the risk of diabetes mellitus, it can also be done bycontrolling blood glucose, blood pressure, and lipid prole at the nearest health care facility (Association, 2021). This study had some limitations. First, the data used was secondary data obtained from Kaggle. As a result, some sub- jects’ data were incomplete, and they were eventually dropped from the study. Second, we could not conrm more detailed information from the subject, for example, the risk parameter, which stated that the subject regularly consumes drugs. We did not know the type and amount of drugs consumed by the subjects. 4. CONCLUSION The Fuzzy C-Means clustering algorithm was used to cluster the results of 905 subjects, yielding clusters 1 (high risk cluster) and 2 (low risk cluster). Cluster 1 was more likely to acquire diabetesmellitusdueto thepresenceofmoreprevalentdiabetes risk parameters. Subjects who have risk parameters that are included in the high-risk cluster need to be aware and alert to prevent the occurrence of type 2 diabetes mellitus. Further research is needed to determine whether changing the risk parameters of type 2 diabetes mellitus through lifestyle changes can reduce the risk of developing type 2 diabetes mellitus. 5. ACKNOWLEDGMENT The authors are grateful to the Faculty of Pharmacy and com- putational math lab department of mathematics FAST UAD for their support. REFERENCES Akalu, Y. and Y. Belsti (2020). Hypertension and its Associ- ated Factors Among Type 2 Diabetes Mellitus Patients at Debre Tabor General Hospital, Northwest Ethiopia. Dia- betes, Metabolic Syndrome and Obesity: Targets and Therapy, 13; 1621 Asamoah-Boaheng, M., O. Sarfo-Kantanka, A. B. Tuour, B. Eghan, and J. C. Mbanya (2019). Prevalence and Risk Factors for Diabetes Mellitus Among Adults in Ghana: A Systematic Review and Meta-analysis. International Health, 11(2); 83–92 Asiimwe, D., G. O. Mauti, and R. Kiconco (2020). Prevalence and RiskFactors Associated With Type 2 Diabetes in Elderly Patients Aged 45-80 Years at Kanungu District. Journal of Diabetes Research, 2020; 1–5 Association, A. D. (2021). Improving Care and Promot- ing Health in Populations: Standards of Medical Care in Diabetes-2021. Diabetes Care, 44(1); S7–S14 Gökhan, A., C. O. Güzeller, and M. T. Eser (2019). The Eect of The Normalization Method Used in Dierent Sample Sizes on the Success Of Articial Neural Network Model. International Journal of Assessment Tools in Education, 6(2); 170–192 Gopalakrishnan, S. and A. Geetha (2017). Studyon the Impact of Family History of Diabetes Among Type 2 Diabetes Mel- litus Patients in an Urban Area of Kancheepuram District, Tamil Nadu. International Journal Community Medicine and Public Health, 4; 4151–4156 Hinkle, D. E., W. Wiersma, and S. G. Jurs (2003). Applied Statistics for the Behavioral Sciences. Houghton Miin College Division Ismail, L., H. Materwala, and J. Al Kaabi (2021). Association of Risk Factors With Type 2 Diabetes: A Systematic Review. Computational and Structural Biotechnology Journal, 19; 1759– 1785 Jamuna, . K. K. M., S. (2020). Prediction of Diabetes and Clustering Based on its Levels Using Fuzzy C Means Algo- rithm. International Journal of Scientic and Technology, 9(2); 3222–3225 Kim, M. J., N. K. Lim, S. J. Choi, and H. Y. Park (2015). Hypertension is an Independent Risk Factor For Type 2 Diabetes: the Korean Genome and Epidemiology Study. Hypertension Research, 38(11); 783–789 Kunwar, V., A. S. Sabitha, T. Choudhury, and A. Aggarwal (2019). Chronic Kidney Disease Using Fuzzy C-means Clustering Analysis. International Journal of Business Analytics (IJBAN), 6(3); 43–64 Kurniawan, M., R. R. Muhima, and S. Agustini (2020). Com- parison of Clustering K-means, Fuzzy C-means, and Link- age For Nasa Active Fire Dataset. International Journal of Articial Intelligence and Robotics (IJAIR), 2(2); 34 Liyanage, L. (2018). Diabetes Mellitus and its Risk Fac- tors. EPRA International Journal of Multidisciplinary Research (IJMR), 4(9); 114–117 Lomo, S. I., E. Darmawan, and S. Sugiyarto (2021). Cluster analysis of Type Ii Diabetes Mellitus Patients with the Fuzzy C-means Method. Annals of Mathematical Modeling, 13(1); 32–39 © 2023 The Authors. Page 23 of 24 Marhamah et. al. Science and Technology Indonesia, 8 (2023) 17-24 Madhulatha, T. S. (2012). An Overview on Clustering Meth- ods. IOSR Journal of Engineering, 2(4); 719–725 Mezil, . A. B. A., S. A. (2021). Complication of Diabetes Mellitus. Annals of the RomanianSociety forCellBiology, 25(3); 1546–156 Muslimatin, B. (2011). Perbandingan Metode K-means dan Metode Fuzzy C-means untuk Clustering Data: Studi Kasus Pada Harga Saham Harian Pada Pt. Astra, Tbk. Ph.D. thesis, Uni- versitas Islam Negeri Maulana Malik Ibrahim Park, S. J., Y. J. Park, J. Chang, S. Choi, G. Lee, J. S. Son, K. H. Kim, Y. H. Oh, and S. M. Park (2021). Association Between Antibiotics Use and Diabetes Incidence in A Nationally Rep- resentative Retrospective Cohort Among Koreans. Scientic Reports, 11(1); 1–10 Pham, T. B., T. T. Nguyen, H. T. Truong, C. H. Trinh, H. N. T. Du, T. T. Ngo, and L. H. Nguyen (2020). Ef- fects of Diabetic Complications on Health-related Quality of Life Impairment in Vietnamese Patients with Type 2 Diabetes. Journal of Diabetes Research, 2020; 1–8 Pinidiyapathirage, M., A. Kasturiratne, U. Ranawaka, D. Gu- nasekara, N.Wijekoon, K.Medagoda, S.Perera, F.Takeuchi, N. Kato, and T. Warnakulasuriya (2013). The Burden of Diabetes Mellitus and Impaired Fasting Glucose in an Urban Population of Sri Lanka. Diabetic Medicine, 30(3); 326–332 Ramya, T. B. (2018). Disease Prediction System Using Fuzzy C-means Algorithm. International Journal of Engineering Re- search & Technology (IJERT), 6(3); 1–4 Repaske, D. R. (2016). Medication Induced Diabetes Mellitus. Pediatric Diabetes, 17(6); 392–397 Saeedi, P., I. Petersohn, P. Salpea, B. Malanda, S. Karuranga, N. Unwin, S. Colagiuri, L. Guariguata, A. A. Motala, and K. Ogurtsova (2019). Global and Regional Diabetes Preva- lence Estimates For 2019 and Projections For 2030 and 2045: Results From The International Diabetes Federation Diabetes Atlas. Diabetes Research and Clinical Practice, 157; 107843 Sakamoto, R., T. Yamakawa, K. Takahashi, J. Suzuki, M. M. Shinoda, K. Sakamaki, H. Danno, H. Tsuchiya, M. Waseda, and T. Takano (2018). Association of Usual Sleep Quality and Glycemic Control in Type 2 Diabetes in Japanese: A Cross Sectional Study. Sleep and Food Registry in Kanagawa (SOREKA). PloS one, 13(1); 0191771 Sanakal, R. and T. Jayakumari (2014). Prognosis of Diabetes Using Data Mining Approach-fuzzyC Means Clustering and Support Vector Machine. International Journal of Computer Trends and Technology, 11(2); 94–98 Sarría-Santamera, A., B. Orazumbekova, T. Maulenkul, A. Gaipov, and K. Atageldiyeva (2020). The Identication of Diabetes Mellitus Subtypes Applying Cluster Analysis Techniques: A Systematic Review. International Journal of Environmental Research and Public Health, 17(24); 9523 Shan, Z., H. Ma, M. Xie, P. Yan, Y. Guo, W. Bao, Y. Rong, C. L. Jackson, F. B. Hu, and L. Liu (2015). Sleep Duration andRiskOfType2Diabetes: AMeta-analysisofProspective Studies. Diabetes Care, 38(3); 529–537 Simbolon, D., A. Siregar, and R. A. Talib (2020). Physiological Factors and Physical Activity Contribute to The Incidence of Type 2 Diabetes Mellitus in Indonesia. Kesmas: Jurnal KesehatanMasyarakat Nasional (NationalPublic HealthJournal), 15(3); 120–127 (In Indonesia) Stahl, C. H., M. Novak, G. Lappas, L. Wilhelmsen, L. Björck, P.-O. Hansson, and A. Rosengren (2012). High-normal Blood Pressure and Long-term Risk Of Type 2 Diabetes: 35-yearProspective Population Based Cohort Studyof Men. BMC Cardiovascular Disorders, 12(1); 1–8 Steyn, N. P., J. Mann, P. Bennett, N. Temple, P. Zimmet, J. Tuomilehto, J. Lindström, and A. Louheranta (2004). Diet, Nutrition and The Prevention of Type 2 Diabetes. Public Health Nutrition, 7(1a); 147–165 Tang, M., Y. Zhou, A. Song, J. Wang, Y. Wan, and R. Xu (2021). The Relationship Between Body Mass Index and Incident Diabetes Mellitus in Chinese Aged Population: A Cohort Study. Journal of Diabetes Research, 2021; 1–8 Tigga, N. P. and S. Garg (2020). Prediction of Type 2 Diabetes Using Machine Learning Classication Methods. Procedia Computer Science, 167; 706–716 Uusitupa, M., T. A. Khan, E. Viguiliouk, H. Kahleova, A. A. Rivellese, K. Hermansen, A. Pfeier, A. Thanopoulou, J. Salas-Salvadó, and U. Schwab (2019). Prevention of Type 2 Diabetes by Lifestyle Changes: A Systematic Review and Meta-analysis. Nutrients, 11(11); 2611 Virani, S. S., A. Alonso, H. J. Aparicio, E. J. Benjamin, M. S. Bittencourt, C. W. Callaway, A. P. Carson, A. M. Chamber- lain, S. Cheng, and F. N. Delling (2021). Heart Disease and Stroke Statistics—2021 Update: A Report From the Ameri- can Heart Association. Circulation, 143(8); e254–e743 Wang, Q., X. Zhang, L. Fang, Q. Guan, L. Gao, and Q. Li (2018). Physical Activity Patterns and Risk of Type 2 Dia- betes and Metabolic Syndrome In Middle-aged and Elderly Northern Chinese Adults. Journal of Diabetes Research, 2018; 1–8 Wei, G. S., S. A. Coady, D. C. Go Jr, F. L. Brancati, D. Levy, E. Selvin, R. S. Vasan, and C. S. Fox (2011). Blood Pressure and the Risk Of Developing Diabetes in African Americans and Whites: Aric, Cardia, and the Framingham Heart Study. Diabetes Care, 34(4); 873–879 Wu, J., S. L. Mackie, and M. Pujades-Rodriguez (2020). Glu- cocorticoid Dose Dependent Risk of Type 2 Diabetes in Six Immune-mediated Inammatory Diseases: A Population- based Cohort Analysis. BMJ Open Diabetes Research and Care, 8(1); 001220 Zhang, J., Z. Yang, J. Xiao, X. Xing, J. Lu, J. Weng, W. Jia, L. Ji, Z. Shan, and J. Liu (2015). Association Between Family History Risk Categories and Prevalence of Diabetes in Chinese Population. PLoS One, 10(2); e0117044 Zhang, X. and Q. Zhao (2016). Association of Thiazide-Type Diuretics With Glycemic Changes in Hypertensive Patients: A Systematic Review and Meta-Analysis of Randomized ControlledClinicalTrials. TheJournalofClinicalHypertension, 18(4); 342–351 © 2023 The Authors. Page 24 of 24 INTRODUCTION EXPERIMENTAL SECTION Materials Methods RESULT AND DISCUSSION Strand Geometry CONCLUSION ACKNOWLEDGMENT