Perancangan Prediktor Gejala Berat dan Kematian Pada Penderita Pandemi

Rahmadianto, Ikhsan Heru (2024) Perancangan Prediktor Gejala Berat dan Kematian Pada Penderita Pandemi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311740000118-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (2MB) | Request a copy

Abstract

Diperlukan sebuah sistem yang dapat memprediksi kemungkinan munculnya gejala berat atau kematian dari faktor demografi dan faktor penyakit sebelumnya agar tenaga kesehatan dapat memprioritaskan penanganan untuk kelompok rentan saat pandemi. Perancangan prediktor gejala berat dan kematian pada penderita pandemi dilakukan dengan studi literatur terkait Random Forest Classifier (RFC), dan K-Modes untuk kemudian menjadi dasar dalam menyusun prediktor. Data berupa demografi dan riwayat penyakit pasien diambil sebanyak 1300 data dari Rumah Sakit Husada Utama Surabaya dengan rincian 780 data pengajar, 260 data validasi dan 260 data uji. Disusun algoritma pembuatan Machine Learning menggunakan data-data tersebut menjadi prediktor gejala berat dan kematian kemudian diuji tingkat akurasi dan prediksinya. Didapatkan 3 fitur dengan signifikansi yang sangat tinggi pada garis jumlah kumulatif sama dengan 70% Sehingga dapat disimpulkan bahwa faktor-faktor yang paling berpengaruh adalah obesitas, gagal ginjal dan komorbid diabetes Didapatkan performansi confusion matrix Random Forest Classifier dengan precision, recall, F1 Score, Accuracy dan specifity secara berturut-turut 85,6%, 99,8%, 92,2%, 85,6% dan 5,1% untuk validasi, 85,7%, 100%, 92,3%, 85,8% dan 5,1% untuk pengujian. Sedangkan dengan k-modes didapatkan performansi berupa didapatkan precision, recall, F1 Score, Accuracy dan specifity secara berturut-turut 87,2 %, 97,6%, 92,1%, 85,8% dan 15,9% untuk validasi, 86%, 96,3%, 90,8%, 83,8% dan 22,7%.
=================================================================================================================================
A system is needed that can predict the possibility of severe symptoms or death based on demographic factors and previous disease factors so that health workers can prioritize treatment for vulnerable groups during a pandemic. The design of predictors of severe symptoms and death in pandemic patients was carried out using literature studies related to Random Forest Classifier (RFC) and K-Modes to then become the basis for preparing predictors. Data in the form of demographics and disease history of patients was taken as much as 1300 data from the Husada Utama Hospital in Surabaya with details of 780 training data, 260 validation data and 260 test data. A Machine Learning algorithm was developed using this data to become a predictor of severe symptoms and death, then the level of accuracy and prediction was tested. There were 3 features with very high significance on the cumulative total line equal to 70%. So it can be concluded that the most influential factors are obesity, kidney failure and comorbid diabetes. The confusion matrix Random Forest Classifier performance was obtained with precision, recall, F1 Score, Accuracy and specificity respectively 85.6%, 99.8%, 92.2%, 85.6% and 5.1% for validation, 85 .7%, 100%, 92.3%, 85.8% and 5.1% for testing. Meanwhile, with k-modes, performance was obtained in the form of precision, recall, F1 Score, Accuracy and specificity respectively 87.2%, 97.6%, 92.1%, 85.8% and 15.9% for validation, 86%, 96.3%, 90.8%, 83.8% and 22.7%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Predictors, Random Forest Classifier, K-modes, Pandemic, Prediktor, Algoritma Random Forest, Pandemi.
Subjects:	R Medicine > RA Public aspects of medicine > RA644.C67 COVID-19 (Disease) R Medicine > RA Public aspects of medicine > RA653.5 Epidemics T Technology > T Technology (General) > T58.62 Decision support systems
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Ikhsan Heru Rahmadianto
Date Deposited:	01 Feb 2024 05:15
Last Modified:	01 Feb 2024 05:15
URI:	http://repository.its.ac.id/id/eprint/105878

Actions (login required)

View Item