Ditemukan 8 dokumen yang sesuai dengan query :: Simpan CSV
ABSTRAK Nama : Marta Butar Butar Program Studi : Epidemiologi Judul : Prediktor Kejadian Infeksi Sifilis Pada Populasi Lelaki Suka Seks dengan Lelaki (LSL) di 10 Kabupaten/Kota Di Indonesia (Analisis Data STBP 2015) Latar Belakang : Berdasarkan angka kejadian sifilis pada kelompok LSL yang tercatat pada STBP Tahun 2011 cenderung meningkat sebesar 9 % (dari 4% menjadi 13%) dibandingkan STBP Tahun 2007. Tujuan penelitian ini adalah menganalisis prediktor kejadian sifilis pada populasi LSL yaitu umur, tingkat pendidikan, status HIV, penggunaan kondom, konsumsi Napza/Napza suntik, konsumsi alkohol, jumlah pasangan seks dan pemeriksaan IMS. Metode : Desain Penelitian cross sectional menggunakan data sekunder dari STBP 2015. Data di analisis dengan Cox regresion. Populasi penelitian yaitu kelompok LSL yang berasal dari 10 kabupaten/kota dengan jumlah sampel responden yaitu 1495 orang. Hasil : Proporsi infeksi Sifilis pada kelompok LSL pada 10 kabupaten/kota di Indonesia adalah 15,7%. Ada hubungan yang bermakna antara status HIV (PR 2,05 (95% CI 1,58-2,66), Umur (20-24 tahun (PR 2,45, 95% CI 1,07-5,64), 25-29 tahun (PR 3,01, 95% CI 1,30-6,95), > 30 tahun (PR 2,42, 95% CI 1,04-5,65) dibandingkan LSL umur 15-19 tahun) dengan kejadian infeksi sifilis pada LSL dan ada interaksi antara alkohol dan pendidikan (LSL berpendidikan rendah yang minum alkohol (PR Interaksi 0,47 95% CI 0,23-0,96), LSL berpendidikan rendah tidak minum alkohol (PR Interaksi 1,34 95% CI 0,94-1,90) dan LSL berpendidikan tinggi yang minum alkohol (PR Interaksi 1,4 95% CI 1,03-1,90) dibandingkan LSL yang berpendidikan tinggi yang tidak minum alkohol) dengan kejadian infeksi sifilis pada LSL sedangkan penggunaan kondom, Napza/Napza suntik, jumlah pasangan seks lelaki dan pemeriksaan IMS tidak berhubungan secara statistik dengan nilai p > 0,05 dengan kejadian sifilis. Kata kunci : Prediktor, sifilis, Lelaki Suka Seks dengan Lelaki
ABSTRACT Name : Marta Butar Butar Program Major : Epidemiology Title : Predictors of Syphilis Infections In Population of Male Who Have Sex With Men (MSM) in 10 Districts / Cities In Indonesia (Data analysis STBP 2015) Background : Based on the incidence of Syphilis in delayed groups of MSM in STBP 2011 the symptoms increased by 9% (from 4% to 13%) compared to STBP Year 2007. The purpose of this study was predictors of syphilis infection in MSM population, age, education level, HIV status, Condoms, intake / drug consumption, alcohol consumption, number of sex partners and STI examination. Method: The cross sectional study design used secondary data from STBP 2015. The data were analyzed by Cox regression. The population of the study were MSM group from 10 districts / cities with 1495 respondents Results: The proportion of Syphilis infections in MSM in 10 districts / cities in Indonesia was 15.7%. There was a significant relationship between HIV status (PR 2.05 (95% CI 1.58-2.66), Age (20-24 years (PR 2.45, 95% CI 1.07-5.64), 25 29 years (PR 3.01, 95% CI 1.30-6.95),> 30 years (PR 2.42, 95% CI 1.04-5.65) compared with men aged 15-19 years) with syphilis infection in MSM and there is an interaction between alcohol and education (low educated MSM who drink alcohol (PR Interaction 0.47 95% CI 0.23-0.96), low educated MSM who not drink alcohol (PR Interaction 1.34 95 % CI 0.94-1.90) and high educated MSM who drink alcohol (PR Interaction 1,4 95% CI 1.03-1.90) than high educated MSM who not drink alcohol with syphilis infection in MSM while condom use, drug/ injecting drug, number of male sex partners and STI examination were not statistically correlated (p> 0,05) with syphilis infection. Keywords: Predictors, syphilis, Men Sex With Men
Human Immunodeficiency Virus (HIV) is still an issue in health sector in the world, particularly in Indonesia. Progression of disease is influenced by various factors including age, genetic, and other infectious diseases such as tuberculosis and hepatitis, nutritional factors, and immunological status. ARV therapy has not been able to cure the disease yet is able to control the progression of HIV/AIDS by suppressing viral replication which reduce the incidence of opportunistic infections. Although the program has been implemented, the deaths from HIV continue to occur, especially in the first year of ARV treatment. This study aims to investigate the predictors related to death in HIV-AIDS patients with ARV therapy in Dr. H. Marzoeki Mahdi Hospital in Bogor in 2008-2012. The study design was retrospective cohort using ART registration data and Medical Record. Number of samples were 396 HIV patients with ARV therapy. Data analysis was performed using Cox Regression. The multivariate analysis showed that the predictors of deaths in HIV-AIDS patients with ARV therapy were functional baring status (RR = 2.34, 95% CI: 1.32-4.11), heavy IO category (RR = 2.11, 95% CI : 1.26-3.54), and anemia status (RR = 2.56, 95% CI: 1.74-3.77). Special attention and monitoring are required for HIV/AIDS patients taking antiretroviral medications with functional status of baring, anemia, and having severe opportunistic infections. Keywords: ARV; HIV-AIDS; Retrospective cohort; Death; Predictors.
Latar belakang: Preeklamsia merupakan sindroma kompleks yang timbul pada ibu hamil yang disebabkan oleh perubahan fisiologi pembuluh darah pada saat konsepsi. Dampak kesehatan yang timbul dari kehamilan dengan preeklamsia sangat luas. Sejauh ini berbagai upaya telah ditempuh untuk dapat melakukan prediksi akan terjadinya preeklamsia. Keterbatasan kemampuan statistik mengolah berbagai jenis prediktor saat ini bisa ditingkatkan dengan menggunakan metode machine learning (ML).
Tujuan: Penelitian ini bertujuan untuk melakukan pemodelan prediksi terhadap kejadian preeklamsia menggunakan ML menggunakan fitur yang serupa dengan fitur yang ada di fasyankes primer
Metodologi: Penelitian ini menggunakan disain restrospektif kohort dengan menggunakan data sekunder dari ibu hamil yang melakukan pemeriksaan antenatal di RS Budi Kemuliaan Jakarta yang direkrut pada periode waktu Juli 2012 hingga April 2015 dan diikuti hingga terjadi persalinan. Data tersebut adalah data yang dikumpulkan pada penelitian sebelumnya untuk melihat berbagai faktor risiko dari terjadinya hipertensi dalam kehamilan. Faktor risiko yang diteliti meliputi riwayat diabetes melitus (DM) sebelumnya, riwayat hipertensi sebelumnya, riwayat DM dalam keluarga, riwayat hipertensi dalam keluarga, riwayat merokok, primigraviditas, rerata tekanan arteri (MAP), indeks masa tubuh (BMI) sebelum kehamilan, terhadap kejadian preeklamsia. Permodelan dilakuan dengan menggunakan beberapa model dalam pembelajaran mesin: Random Forest, Logistic Regression, Support Vector Machine (SVM), Decision Tree, dan K-Nearest Neighbour (KKN) untuk mendapatkan model terbaik.
Hasil: Model Suport Vector Machine (SVM) dengan pembelajaran mesin yang disertai dengan re-sampling (undersampling) pada kumpulan data dan perlakuan hyperparameter tuning berhasil mendapatkan akurasi 70,31 %, sensitifitas 67,5 %, spesifisitas 57,23%, dan AUC 0.68.
Kesimpulan: Model SVM menunjukkan kinerja paling baik di antara model prediksi lain pada kumpulan data dan fitur yang tersedia. Model tersebut menghasilkan akurasi 70% dan AUC sebesar 0,68. Model ini mampu mencapai sensitivitas 67,5% dan spesifisitas 57,23%, dengan presisi (positive predictive value) sebesar 74,98%. Artinya, dari seluruh prediksi positif, sekitar 75% adalah benar, sementara sisanya adalah positif palsu. Temuan ini menunjukkan bahwa model prediksi berbasis SVM memiliki potensi untuk dikembangkan lebih lanjut. Optimasi sensitivitas dan spesifisitas menjadi prioritas pada penelitian lanjutan agar model lebih akurat dalam prediksi dan lebih efektif dalam deteksi dini preeklampsia di layanan primer.
Background: Preeclampsia is a complex syndrome that arises in pregnant women caused by physiological changes in blood vessels at the time of conception. The health impacts arising from pregnancy with preeclampsia are widespread. So far, various efforts have been taken to be able to predict the occurrence of preeclampsia. The limitations of statistical ability to process various types of predictors today can be improved by using machine learning (ML) methods.
Objective: This study aims to model predictions for the incidence of preeclampsia using machine learning based on features found in primary health facilities
Methodology: This study uses a cohort retrospective design using secondary data from pregnant women who underwent antenatal care at Budi Kemuliaan Hospital Jakarta who were recruited in the period from July 2012 to April 2015 and followed until delivery. The data was collected in previous studies to look at various risk factors for the occurrence of hypertension in pregnancy. The risk factors studied included history of diabetes mellitus (DM), history of hypertension, a family history of DM, a family history of hypertension, a history of smoking, primigravidity, mean arterial pressure (MAP), body mass index (BMI) before pregnancy, and the incidence of preeclampsia. The modeling was carried out using several models in machine learning: Random Forest, Logistic Regression, Support Vector Machine (SVM), Decision Tree, and K-Nearest Neighbour (KKN) to get the best model.
Results: The Support Vector Machine (SVM) model with machine learning accompanied by re-sampling (undersampling) on the dataset and hyperparameter tuning treatment managed to obtain an accuracy of 70.31%, sensitivity of 67.5%, specificity of 57.23%, and AUC of 0.68.
Conclusion: The SVM model shows the best performance among other prediction models for the available datasets and features. The model produces 70% accuracy and an AUC of 0.68. This model was able to achieve sensitivity of 67.5% and specificity of 57.23%, with a precision of 74.98% which provide the positive predictions about 75% are true, while the rest are false positives. These findings suggest that SVM prediction models have the potential to be further developed. Sensitivity and specificity optimization are priorities in further research to achieve a more robust model with optimum accuracy to predict preeclampsia and more effective in early detection of preeclampsia in primary care.
