<![CDATA[Chronic kidney disease is a global health problem that requires early detection to prevent the progression of more serious conditions. This study aims to design and implement a web-based early detection information system for chronic kidney disease integrated with a prediction model. The system was developed using the Prototyping method with TabNet as the classification model and Class-Conditional Conformal Prediction (CCP) to provide prediction confidence information. The study used secondary dummy data representing clinical attributes and risk factors for chronic kidney disease. Data were processed using median imputation, soft class weighting, and train–validation–calibration–test splitting. Prediction labels were determined using a CKD probability threshold of 0.75. Evaluation results showed accuracy of 0.8614, precision of 0.9448, recall of 0.9013, and F1-score of 0.9226. CCP enables the system to display a prediction set, making early detection results more informative and structured.Keywords: Chronic kidney disease; Conformal prediction; Early detection; Information systems; TabNet algorithmAbstrakPenyakit ginjal kronis merupakan masalah kesehatan global yang memerlukan deteksi dini untuk mencegah perkembangan kondisi yang lebih serius. Penelitian ini bertujuan merancang dan mengimplementasikan sistem informasi deteksi dini penyakit ginjal kronis berbasis web yang terintegrasi dengan model prediksi. Sistem dikembangkan menggunakan metode Prototyping dengan TabNet sebagai model klasifikasi dan Class-Conditional Conformal Prediction (CCP) untuk menyajikan informasi keyakinan prediksi. Data penelitian menggunakan data sekunder berbentuk data dummy yang merepresentasikan atribut klinis dan faktor risiko penyakit ginjal kronis. Data diproses menggunakan median imputation, soft class weighting, dan pembagian train–validation–calibration–test. Label prediksi ditentukan berdasarkan threshold probabilitas CKD sebesar 0,75. Hasil pengujian menunjukkan accuracy 0,8614, precision 0,9448, recall 0,9013, dan F1-score 0,9226. Penerapan CCP memungkinkan sistem menampilkan prediction set, sehingga hasil deteksi dini menjadi lebih informatif dan terstruktur.Kata kunci: penyakit ginjal kronis; conformal prediction; deteksi dini; sistem informasi; algoritma TabNet]]>
