Deteksi Risiko Komplikasi Continuous Ambulatory Peritoneal Dialysis (CAPD) Melalui Image Classification Menggunakan Pendekatan Deep Learning pada Dataset Citra Effluent Dialysate

Sari, Fiqey Indriati Eka (2023) Deteksi Risiko Komplikasi Continuous Ambulatory Peritoneal Dialysis (CAPD) Melalui Image Classification Menggunakan Pendekatan Deep Learning pada Dataset Citra Effluent Dialysate. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111940000015-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (6MB) | Request a copy

Abstract

Continuous Ambulatory Peritoneal Dialysis (CAPD) merupakan alternatif terapi pengobatan untuk pasien dengan Penyakit Ginjal Tahap Akhir (PGTA) di Indonesia. CAPD memungkinkan pasien melakukan dialisis mandiri tanpa perlu ke rumah sakit dengan kualitas hidup yang lebih tinggi dibanding metode lainnya. Namun, risiko kematian pada pasien CAPD tinggi akibat komplikasi yang disebabkan oleh kelalaian dan kesalahan teknis. Salah satu indikator awal untuk mendeteksi komplikasi adalah perubahan effluent dialysate (cairan buangan) pada pasien CAPD. Penelitian terkait masih terbatas. Oleh karena itu, diusulkan pengembangan model pendeteksi risiko komplikasi CAPD melalui image classification dengan menggunakan deep learning terhadap dataset citra effluent dialysate. Dataset dikumpulkan secara langsung melalui rumah sakit di Surabaya dan berbagai jurnal karena ketersediannya yang sangat terbatas. Citra menjadi input model deep learning dengan menggunakan metode transfer learning terhadap model pre-trained VGGNet-16, InceptionV3, ResNet50, InceptionResNetV2, EfficientNetB7, dan MobileNetV2. Kemudian, model tersebut dilakukan fine-tuning dengan penambahan beberapa layer disertai eksperimen berbagai skenario resampling dataset, augmentasi, optimizer, dan learning rate. Hasil eksperimen menunjukkan bahwa model terbaik dicapai oleh InceptionResNetV2 dengan recall sebesar 80% dan F1-score sebesar 89% terhadap data test-II, yaitu data yang diperoleh dan divalidasi oleh dokter CAPD yang merupakan unseen dan real-world data. Dengan begitu, sistem ini diharapkan dapat membantu pendeteksian risiko komplikasi CAPD dalam sistem SahabatCAPD sehingga dapat mencegah lebih dini terjadinya komplikasi yang lebih serius pada pasien CAPD.
=================================================================================================================================
Continuous Ambulatory Peritoneal Dialysis (CAPD) is an alternative treatment therapy for patients with End-Stage Renal Disease (ESRD) in Indonesia. CAPD allows patients to perform self-dialysis without the need to go to the hospital, resulting in a higher quality of life compared to other methods. However, the risk of death for CAPD patients is high due to complications caused by negligence and technical errors. One early indicator to detect complications is the change in effluent dialysate (waste fluid) in CAPD patients. Research related to this is still limited; therefore, the author proposes the development of a risk detection system for CAPD complications through image classification using deep learning and a dataset of effluent dialysate images. The dataset was collected directly by researchers from hospitals in Surabaya and various journals due to its limited availability. The image serve as input to the deep learning model using transfer learning with pre-trained models VGGNet-16, InceptionV3, ResNet50, InceptionResNetV2, EfficientNetB7, and MobileNetV2. Subsequently, the model underwent fine-tuning by adding additional layers along with experiments involving various dataset resampling scenarios, augmentation, optimizers, and learning rates. The experimental results showed that the best model was achieved by InceptionResNetV2 with a recall of 80% and an F1-score of 89% on test-II data, which was obtained and validated by CAPD doctors and represents unseen and real-world data. Thus, it is expected that this system will help in detecting the risk of CAPD complications in the SahabatCAPD system, thereby preventing more serious complications in CAPD patients at an earlier stage.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Continuous Ambulatory Peritoneal Dialysis (CAPD), risiko komplikasi, deep learning, image classification, effluent dialysate.
Subjects:	R Medicine > RA Public aspects of medicine > RA0421 Public health. Hygiene. Preventive Medicine T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Fiqey Indriati Eka Sari
Date Deposited:	04 Aug 2023 03:45
Last Modified:	04 Aug 2023 03:45
URI:	http://repository.its.ac.id/id/eprint/100360

Actions (login required)

View Item