RANCANG BANGUN SISTEM INFORMASI MATERNAL MENGGUNAKAN ALGORITMA MACHINE LEARNING RANDOM FOREST DI DINAS KESEHATAN PROVINSI LAMPUNG

Afifatul, Mukaromah (2026) RANCANG BANGUN SISTEM INFORMASI MATERNAL MENGGUNAKAN ALGORITMA MACHINE LEARNING RANDOM FOREST DI DINAS KESEHATAN PROVINSI LAMPUNG. Diploma thesis, UIN RADEN INTAN LAMPUNG.

	PDF Download (5MB)
	PDF Restricted to Repository staff only Download (5MB)

Abstract

Tingginya Angka Kematian Ibu (AKI) di Provinsi Lampung menjadi masalah serius, sementara pengelolaan data manual membuat identifikasi wilayah berisiko tinggi lambat dan rentan kesalahan. Belum adanya sistem informasi maternal yang mengintegrasikan analisis prediktif berbasis machine learning Random Forest di Dinas Kesehatan Provinsi Lampung menjadikan deteksi dini risiko kematian ibu belum dapat dilakukan secara sistematis. Penelitian ini merancang dan membangun sistem menggunakan model Waterfall, framework Laravel dan Python/scikit learn berbasis MySQL, dengan algoritma Random Forest sebagai basis analisis prediktif, menggunakan data agregat 15 kabupaten/kota periode 2020–2024 (75 data, dua belas variabel penyebab kematian ibu). Pengujian dilakukan menggunakan K-Fold Cross Validation (K=5), Black Box Testing, dan System Usability Scale (SUS) terhadap 16 responden. Hasil pengujian menunjukkan akurasi 61,3%, precision 58,1%, recall 52,9%, dan F1-Score 55,4%, dengan Perdarahan sebagai faktor paling dominan (Feature Importance 17,67%). Seluruh fungsi sistem dinyatakan valid melalui Black Box Testing, dan nilai SUS 82,50 (kategori Sangat Baik) menunjukkan sistem layak digunakan sebagai pendukung pengambilan keputusan di Dinas Kesehatan Provinsi Lampung. Kata Kunci: Sistem Informasi Maternal, Machine Learning, Random Forest, Angka Kematian Ibu, K-Fold Cross Validation, Feature Importance The high Maternal Mortality Rate (MMR) in Lampung Province remains a serious issue, while manual data management makes the identification of high-risk regions slow and prone to error. The absence of a maternal information system integrating predictive analysis based on Random Forest machine learning at the Lampung Provincial Health Office has hindered the systematic early detection of maternal mortality risk. This research designed and developed a system using the Waterfall model, the Laravel framework, and Python/scikit-learn with a MySQL database, employing the Random Forest algorithm as the basis for predictive analysis. The system utilized aggregate data from 15 regencies/cities over the 2020–2024 period (75 data points, twelve maternal mortality causal variables). Testing was conducted using K-Fold Cross Validation (K=5), Black Box Testing, and the System Usability Scale (SUS) involving 16 respondents. The test results showed an accuracy of 61.3%, precision of 58.1%, recall of 52.9%, and an F1-Score of 55.4%, with Hemorrhage identified as the most dominant factor (Feature Importance of 17.67%). All system functions were declared valid through Black Box Testing, and the SUS score of 82.50 (categorized as Excellent) indicates that the system is feasible for use as a decision support tool at the Lampung Provincial Health Office. Keywords: Maternal Information System, Machine Learning, Random Forest, Maternal Mortality Rate, K-Fold Cross Validation, Feature Importance

Actions (login required)

View Item