ANALISIS POLA PENYAKIT DIABETES MELITUS MENGGUNAKAN ALGORITMA APRIORI (STUDI KASUS: PUSKESMAS CIGUGUR TENGAH)

Abstrak

Indonesia is included in the top 10 countries with the highest number of diabetes mellitus sufferers in the world. In Cimahi City, the Cigugur Tengah Health Center also recorded quite high cases of diabetes mellitus. However, data management is still limited to simple analysis, which makes response policies less effective. Therefore, better methods are needed so that the policies taken are more targeted and have a real impact in reducing this disease. This study aims to identify symptom patterns in Diabetes Mellitus (DM) patients at Cigugur Tengah Health Center using the Apriori algorithm. The rising prevalence of DM calls for in-depth data analysis to enhance treatment and prevention strategies. This study analyzed 516 patient medical records with 10 symptom attributes, including frequent urination, constant hunger, tingling sensations, and age ≥ 45 years. Through data mining with a minimum support of 3%, this research identified 5 (five) symptom candidate itemsets. Based on the analysis of data from 516 patients using the Apriori algorithm in RapidMiner, the combination of symptoms "Always Hungry," "Frequent Urination," and "Tingling" has a high occurrence frequency (0.723) and stands out as a dominant pattern for early detection of diabetes mellitus, with individual symptoms like "Always Hungry" (0.959) and "Frequent Urination" (0.864) serving as primary indicators. This combination, especially when "Fatigue" is added (support 0.516), shows strong correlation, while symptoms like "Always Thirsty" and age ≥ 45 contribute less to this pattern. The combination of "Fatigue" and "Always Thirsty" has the highest lift value (1.383), indicating a significant association with other symptoms, thus strengthening the predictive potential for diabetes-related health conditions.

Keywords: Diabetes Mellitus, Apriori Algorithm, RapidMiner, Symptom Patterns, Community Health Center.