REM Behavior Disorder (RBD) is a sleep disorder characterized by the loss of normal muscle atony (loss of paralysis) during Rapid Eye Movement (REM) sleep, where sufferers act on dreams that can result in physical injury to individuals or their sleep partners. REM is a sleep stage characterized by cessation of eye movement, a decrease in body temperature, slow heart rate and no muscle activity in several parts of the body. One of the methods used to detect RBD is Electroencephalography (EEG). EEG is a method of recording or capturing electrical activity in the brain. The dataset used was sourced from PhysioNet.org which consisted of 2 classes, namely normal class and RBD class which were taken from 26 subjects with 6 normal subjects and 20 RBD subjects. This research was conducted to classify RBD sleep disorders based on EEG signals using the ELM algorithm and it is expected to determine the best algorithm for classifying RBD sleep disorders using the ELM algorithm which will be compared with the SVM and backpropagation algorithms based on the EEG signal in terms of the resulting accuracy value and also the time required. to create a model in the algorithm classification process. Classification of RBD sleep disorders based on EEG signals begins with data pre-processing, feature extraction and classification. Data pre-processing includes signal splitting per 30 seconds and data smoothing. The feature extraction process uses Discrete Wavelet Transformation. The RBD classification process based on EEG signals uses the Extreme Learning Machine (ELM) algorithm with the binary sigmoid activation function. Prior to the training process on the ELM algorithm, undersampling was first carried out to overcome the imbalance in the number of classes. Evaluation of the classification results is done by using k-fold cross-validation. The classification results of RBD sleep disorders based on EEG signals using the ELM algorithm show that the ELM algorithm can classify RBD and non-RBD sleep disorders based on EEG signals with an average accuracy value of 70.71% ± 5.44. The comparison result states that the backpropagation algorithm has the best average accuracy in RBD classification based on EEG signals, reaching 83.81% ± 1.40. However, based on the computation of time, the ELM algorithm is superior in the speed of the RBD classification process based on EEG signals, reaching 0.04 ± 0.06 seconds compared to the Support Vector Machine (SVM) algorithm and backpropagation.

Gagal ginjal kronis ialah suatu keadaan menurunnya kinerja ginjal yang bersifat kronik, progesif dan bertahan lama. Seseorang dengan gagal ginjal kronis akan memiliki efek samping seperti kerusakan pada sistem saraf serta kerusakan pada kekebalan tubuh dan aktivitas sehari-hari dapat terganggu, oleh karena itu sangat penting untuk mendiagnosis dini agar dapat mengurangi risiko dari dampak gagal ginjal kronis. Pada penelitian ini digunakan algoritma Support Vector Machine, K-Nearest Neighbor dan Multilayer Perceptron sebagai metode pengklasifikasian gagal ginjal kronis. Penelitian ini bertujuan agar dapat menentukan algoritma terbaik dalam klasifikasi gagal ginjal kronis berdasarkan nilai akurasi serta penggunaan waktu pembuatan model. Dataset yang digunakan yaitu dataset Chronic Kidney Disease yang bersumber dari UCI Machine Learning repository yang memiliki 2 kelas yaitu kelas gagal ginjal kronis dan kelas bukan gagal ginjal kronis yang diambil dari 400 subyek dengan 24 atribut. Sebelum proses klasifikasi, terlebih dahulu dilakukan normalisasi dan split data menggunakan validasi silang 10-fold. Berdasarkan hasil evaluasi, akurasi tertinggi didapatkan pada saat algoritma MLP dengan nilai rata-rata akurasi mencapai 99.48% ± 0.15. Namun berdasarkan waktu pembuatan model dalam proses klasifikasi, algoritma KNN memiliki waktu yang tercepat yaitu 0.0017 ± 0.0041 detik.