Prediksi dan Analisis Hubungan Jumlah Kasus Infeksi Saluran Pernapasan Atas (ISPA) Berdasarkan Data Kualitas Udara Menggunakan Long Short-Term Memory (LSTM) dan Multiple Linear Regression (MLR) (Studi Kasus: Kota Surabaya)

Prastyka, Rania (2025) Prediksi dan Analisis Hubungan Jumlah Kasus Infeksi Saluran Pernapasan Atas (ISPA) Berdasarkan Data Kualitas Udara Menggunakan Long Short-Term Memory (LSTM) dan Multiple Linear Regression (MLR) (Studi Kasus: Kota Surabaya). Other thesis, Instittut Teknologi Sepuluh Nopember.

Text 5026211042-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Infeksi Saluran Pernapasan Akut (ISPA) merupakan salah satu penyakit yang umum terjadi dan erat kaitannya dengan kualitas udara, terutama di kota besar seperti Surabaya. Penelitian ini bertujuan untuk memprediksi jumlah kasus ISPA serta menganalisis hubungannya dengan parameter kualitas udara (PM2.5, PM10, NO2, SO2, CO, dan O3) menggunakan metode Long Short-Term Memory (LSTM) dan Multiple Linear Regression (MLR). Data time-series dari tahun 2019 hingga 2024 digunakan dalam pemodelan. Hasil tuning menunjukkan bahwa konfigurasi terbaik pada model LSTM diperoleh dengan train-test split 80:20, lookback 60 hari, dan 128 unit neuron, yang mampu memberikan performa prediksi akurat terutama pada parameter NO₂ (R² = 0.89, RMSE = 0.78) dan SO₂ (R² = 0.72, RMSE = 2.47). Model LSTM juga menunjukkan performa yang lebih unggul dalam memprediksi jumlah kasus ISPA dibanding MLR, khususnya saat menggunakan data hasil prediksi kualitas udara sebagai input. Sebaliknya, model MLR menunjukkan performa yang kurang optimal dengan nilai R² sebesar 0.0469 (data aktual) dan 0.1131 (data prediksi LSTM), serta MAPE masing-masing sebesar 84.45% dan 79.91%. Analisis korelasi antara kualitas udara dan jumlah kasus ISPA menunjukkan hubungan yang lemah, dengan korelasi tertinggi hanya sebesar 0.32 (CO). Temuan ini mengindikasikan bahwa ISPA merupakan fenomena multifaktor dan kompleks, sehingga pendekatan prediktif berbasis deep learning seperti LSTM lebih tepat digunakan daripada regresi linier sederhana.
================================================================================================================================
Acute Respiratory Infections (ARI) are among the most common illnesses and are closely related to air quality, especially in major urban areas such as Surabaya. This study aims to predict the number of ARI cases and analyze their relationship with air quality parameters (PM2.5, PM10, NO2, SO2, CO, and O3) using Long Short-Term Memory (LSTM) and Multiple Linear Regression (MLR) methods. Time-series data from 2019 to 2024 were used for modeling. The best LSTM configuration was achieved with an 80:20 train-test split, a 60-day lookback window, and 128 LSTM units, yielding strong predictive performance—particularly for NO₂ (R² = 0.89, RMSE = 0.78) and SO₂ (R² = 0.72, RMSE = 2.47). The LSTM model also outperformed MLR in predicting ARI cases, especially when using air quality data predicted by LSTM as input. In contrast, the MLR model yielded suboptimal results with R² values of 0.0469 (actual data) and 0.1131 (LSTM-predicted data), and MAPE scores of 84.45% and 79.91%, respectively. Correlation analysis revealed only a weak relationship between air quality and ARI cases, with the highest correlation coefficient being 0.32 (CO). These findings indicate that ARI is a multifactorial and complex phenomenon, and that deep learning approaches such as LSTM are better suited for capturing its nonlinear and time-dependent patterns compared to traditional linear regression methods.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ISPA, Kualitas Udara, LSTM, MLR, Acute Upper Respiratory Infection, Air Quality, LSTM, MLR
Subjects:	R Medicine > RC Internal medicine > RC771 Pneumonia. T Technology > TJ Mechanical engineering and machinery > TJ217.6 Predictive Control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Rania Prastyka
Date Deposited:	14 Jul 2025 08:01
Last Modified:	14 Jul 2025 08:01
URI:	http://repository.its.ac.id/id/eprint/119703

Actions (login required)

View Item