PERANGKAT LUNAK UNTUK MENGETAHUI POLA PENYAKIT PADA PASIEN DIABETES MELITUS MENGGUNAKAN RAPIDMINER DI KLINIK PRATAMA TIARA MEDISTRA
Abstract
Diabetes Mellitus (DM) is a chronic non-communicable disease that requires more systematic use of clinical
and behavioral data. Tiara Medistra Primary Clinic stores medical-record data supplemented by lifestyle
information, but the relationships among patient attributes have not been explored adequately. This study
applies the Apriori algorithm with RapidMiner to identify association patterns among sex, age, family history
of DM, DM type, and lifestyle characteristics. The dataset contains 21 patients processed through selection,
cleaning, integration, and transformation into categorical transactions. The analysis uses 20% minimum support
and 60% minimum confidence. The results show frequent itemsets up to 8-itemsets and 38 association rules
focused on DM type. The rule “Family history of DM: Yes → Type 2 DM” obtains 47.62% support, 66.67%
confidence, and 1.400 lift. Several rules leading to Type 2 DM reach 75% confidence, while the highest lift of
2.625 appears in rules leading to unspecified DM with 100% confidence. Type 1 DM does not generate a rule
because it occurs in only 3 of 21 patients, or 14.29%, below the support threshold. The findings can support
patient education and characteristic mapping, but they should not be interpreted as a diagnosis or clinical causal
evidence.

