Nurjoko, - (2026) IMPLEMENTASI METODE KECERDASAN BUATAN (ARTIFICIAL INTELLIGENCE) DENGAN INDIKATOR KESEHATAN MASYARAKAT DI PROVINSI LAMPUNG (STUDI KASUS: PEMETAAN DAN DETEKSI RISIKO STUNTING BERBASIS WILAYAH). [Dataset]
![[img]](http://digilib.unila.ac.id/style/images/fileicons/text.png) |
Text
FILE ABSTRAK_Revisi.pdf Restricted to Registered users only Download (434kB) | Request a copy |
|
|
Text
FILE DISERTASI FULL_Revisi.pdf Restricted to Repository staff only Download (4MB) | Request a copy |
|
|
Text
FILE DISERTASI TANPA BAB PEMBAHASAN_Revisi.pdf Download (1MB) | Preview |
Abstract
Stunting masih menjadi permasalahan kesehatan masyarakat global yang krusial, khususnya di negara berkembang seperti Indonesia, karena dampaknya yang signifikan terhadap pertumbuhan fisik, perkembangan kognitif, dan kesejahteraan jangka panjang anak. Meskipun berbagai intervensi telah dilakukan, tantangan dalam deteksi dini dan penanganan yang tepat sasaran masih menjadi kendala utama. Oleh karena itu, diperlukan pendekatan berbasis kecerdasan buatan (Artificial Intelligence/AI) yang mampu mengolah data kesehatan secara komprehensif, akurat, dan mudah diinterpretasikan. Penelitian ini mengintegrasikan tiga pendekatan utama, yaitu analisis tren penelitian AI melalui Systematic Literature Review (SLR), pemodelan prediksi berbasis machine learning, serta pengembangan sistem pakar fuzzy untuk klasifikasi risiko stunting. Tahap awal dilakukan dengan metode SLR menggunakan protokol PRISMA terhadap basis data ilmiah seperti PubMed, ScienceDirect, Scopus dan Google Scholar untuk mengidentifikasi tren, metode, serta tantangan dalam penerapan AI pada prediksi stunting. Hasil kajian menunjukkan bahwa algoritma yang paling banyak digunakan meliputi Random Forest, Support Vector Machine (SVM), Gradient Boosting (XGBoost, LightGBM) dan Artificial Neural Network (ANN) dengan evaluasi kinerja menggunakan metrik seperti AUC-ROC, akurasi, sensitivitas, dan spesifisitas. Namun demikian, tantangan utama yang dihadapi meliputi keterbatasan kualitas dan akses data, rendahnya interpretabilitas model, serta integrasi sistem dalam layanan kesehatan. Selanjutnya, dilakukan pemodelan prediksi status stunting menggunakan algoritma Support Vector Machine (SVM) dan Extreme Gradient Boosting (XGBoost) berbasis data Posyandu sebanyak 17.498 balita di Provinsi Lampung. Proses analisis meliputi praproses data, penyeimbangan kelas menggunakan metode Synthetic Minority Over-sampling Technique (SMOTE) serta evaluasi model menggunakan stratified 10-fold cross-validation. Hasil penelitian menunjukkan bahwa model XGBoost memiliki performa terbaik dengan nilai akurasi, presisi, recall, dan F1-score mencapai 0,9979, lebih tinggi dibandingkan dengan SVM yang mencapai akurasi 0,9949. Hal ini menunjukkan bahwa XGBoost lebih efektif dalam menangani data berdimensi tinggi, tidak seimbang, serta mampu menangkap pola nonlinier secara optimal. Tahap akhir penelitian mengusulkan pengembangan sistem pakar fuzzy berbasis inferensi Mamdani untuk memprediksi tingkat risiko stunting dengan mengintegrasikan indikator kesehatan masyarakat, seperti prevalensi stunting, cakupan desa UCI, cakupan ASI eksklusif, akses sanitasi sehat, dan pelayanan kesehatan balita. Model fuzzy diimplementasikan menggunakan Scikit-Fuzzy pada bahasa pemrograman Python dengan metode defuzzifikasi Center of Area (COA) untuk menghasilkan skor risiko yang presisi dan terinterpretasi. Data yang digunakan berasal dari laporan kesehatan nasional periode 2020–2024, dan hasil analisis divisualisasikan menggunakan pendekatan spasial-temporal berbasis GeoJSON untuk mengidentifikasi wilayah dengan tingkat risiko stunting di Provinsi Lampung. Hasil penelitian menunjukkan bahwa integrasi metode machine learning dan sistem pakar fuzzy dan ArcGIS tidak hanya meningkatkan akurasi prediksi, tetapi juga memberikan kemampuan interpretasi yang lebih baik melalui pendekatan Explainable AI (XAI). Sistem yang dikembangkan mampu mendukung pengambilan keputusan dengan menggunakan indikator kesehatan masyarakat untuk peningkatan program kesehatan masyarakat, khususnya terkait kasus stunting, secara lebih efektif, transparan, dan berbasis data. Penelitian ini diharapkan dapat menjadi kontribusi signifikan dalam pengembangan sistem prediksi dan deteksi dini stunting berbasis AI di Indonesia serta membuka peluang pengembangan lebih lanjut melalui integrasi teknologi seperti Internet of Things (IoT), multimodal AI, dan natural language processing (NLP) dalam ekosistem kesehatan digital. Kata kunci: Stunting, Kecerdasan Buatan, Machine Learning, XGBoost, SVM, Sistem Pakar Fuzzy, Explainable AI, Indikator Kesehatan Masyarakat. Stunting remains a critical global public health issue, particularly in developing countries such as Indonesia, due to its significant impact on children’s physical growth, cognitive development, and long-term well-being. Despite various interventions that have been implemented, challenges in early detection and targeted intervention remain major obstacles. Therefore, an approach based on Artificial Intelligence (AI) is required to process health data comprehensively, accurately, and in an interpretable manner. This study integrates three main approaches, namely the analysis of AI research trends through a Systematic Literature Review (SLR), predictive modeling using machine learning, and the development of a fuzzy expert system for stunting risk classification. The initial stage was conducted using the SLR method with the PRISMA protocol across scientific databases such as PubMed, ScienceDirect, Scopus, and Google Scholar to identify trends, methodologies, and challenges in applying AI to stunting prediction. The results indicate that the most commonly used algorithms include Random Forest, Support Vector Machine (SVM), Gradient Boosting methods (XGBoost, LightGBM), and Artificial Neural Networks (ANN), with performance evaluated using metrics such as AUC-ROC, accuracy, sensitivity, and specificity. However, several key challenges remain, including limitations in data quality and accessibility, low model interpretability, and difficulties in integrating these systems into healthcare services. Subsequently, predictive modeling of stunting status was conducted using Support Vector Machine (SVM) and Extreme Gradient Boosting (XGBoost) algorithms based on a dataset of 17,498 records of children under five collected from Posyandu data in Lampung Province. The analytical process included data preprocessing, class imbalance handling using the Synthetic Minority Over-sampling Technique (SMOTE), and model evaluation using stratified 10-fold cross-validation. The results show that the XGBoost model achieved superior performance, with accuracy, precision, recall, and F1-score reaching 0.9979, outperforming the SVM model, which achieved an accuracy of 0.9949. These findings indicate that XGBoost is more effective in handling high-dimensional and imbalanced data, as well as in capturing nonlinear patterns. In the final stage, this study proposes the development of a fuzzy expert system based on Mamdani inference to predict stunting risk levels by integrating key public health indicators, such as stunting prevalence, Universal Child Immunization (UCI) village coverage, exclusive breastfeeding coverage, access to proper sanitation, and child healthcare services. The fuzzy model was implemented using the Scikit-Fuzzy library in Python, with defuzzification performed using the Center of Area (COA) method to generate precise and interpretable risk scores. The data were obtained from national health reports covering the period 2020–2024, and the results were visualized using a spatio-temporal approach based on GeoJSON to identify regions with varying levels of stunting risk across Lampung Province. The findings of this study demonstrate that the integration of machine learning methods and fuzzy expert systems not only improves prediction accuracy but also enhances interpretability through an Explainable AI (XAI) approach. The developed system is capable of supporting decision-making in public health planning in a more effective, transparent, and data-driven manner. This research is expected to make a significant contribution to the development of AI-based stunting prediction and early detection systems in Indonesia, while also opening opportunities for future advancements through the integration of technologies such as the Internet of Things (IoT), multimodal AI, and natural language processing (NLP) within the digital health ecosystem. Keywords: Stunting, Artificial Intelligence, Machine Learning, XGBoost, SVM, Fuzzy Expert System, Explainable AI, Public Health Indicators.
| Item Type: | Dataset |
|---|---|
| Subjects: | ?? 500 ?? |
| Divisions: | ?? 21 ?? |
| Depositing User: | 2605752365 Digilib |
| Date Deposited: | 17 Jun 2026 04:40 |
| Last Modified: | 17 Jun 2026 04:40 |
| URI: | http://digilib.unila.ac.id/id/eprint/100640 |
Actions (login required)
 |
View Item |