early prediction of preeclampsia with AI_Dr Sadia.pptx

early prediction of preeclampsia with AI

1. Supervisor: Dr. Maruf Haque Khan Assistant Professor Department of Public Health and Informatics Bangabandhu Sheikh Mujib Medical University Presented By: Dr. Sadia Farhana Nobi Epidemiology, Session: 2023-2024 Department of Public Health and Informatics Bangabandhu Sheikh Mujib Medical University Early Prediction of Preeclampsia Using Deep Learning Model

2. General Objective: Developing a predictive model using deep learning for early detection of late-onset preeclampsia. Specific Objectives: To assess the prevalence of pregnancy-induced hypertension in pregnant women attending a tertiary medical college hospital. To identify demographic factors associated with pregnancy-induced hypertension. To assess clinical factors associated with pregnancy-induced hypertension To identify obstetric factors associated with pregnancy-induced hypertension Objectives

3. Study Design: It will be a cross-sectional study. Place of the study: Mohammadpur Fertility Services and Training Centre and Azimpur and Lalkuthi, Mirpur branch of Maternal and Child Health Training Institute, Dhaka. Study Duration : One year. Study population: For Prevalence study: Pregnant w omen attending Mohammadpur Fertility Services and Training Centre and Azimpur and Lalkuthi, Mirpur branch of Maternal and Child Health Training Institute, Dhaka. For Application Development : An online available dataset includes 11,006 pregnant women who received antenatal care at Yonsei University Hospital. Materials and Methods

4. Sampling technique: Convenient sampling. Sample Size: 220 considering 14% prevalence of preeclampsia. Data Collection Tool: Face to face interview using structured questionnaire. Secondary online available dataset. Ethical Approval: Institutional Review board(IRB), BSMMU. Materials and Methods

5. Study Design with ML Models Missing data imputation, Feature Engineering Classical ML Models: MLP Classifier ElasticNet Linear Discriminant Analysis XGB Classifier Random Forest Classifier Logistic Regression ExtraTrees Classifier ADABoost Classifier KNN Classifier and Gradient Boosting Classifier Online available data Acquisition Data Preprocessing Training of Machine learning Models Best Model Selection Preeclampsia Positive Preeclampsia Negative Data Preprocessing Primary Data Collection Top-ranked Features Validation of The Model With Local Primary Data External Validation of Model Top-ranked Features Preeclampsia Positive Preeclampsia Negative Feature selection by Feature ranking Technique Data splitting

6. Study Design with ML Models Missing data imputation, Feature Engineering Classical ML Models: MLP Classifier ElasticNet Linear Discriminant Analysis XGB Classifier Random Forest Classifier Logistic Regression ExtraTrees Classifier ADABoost Classifier KNN Classifier and Gradient Boosting Classifier Online available data Acquisition Data Preprocessing Training of Machine learning Models Best Model Selection Preeclampsia Positive Preeclampsia Negative Data Preprocessing Primary Data Collection Top-ranked Features Validation of The Model With Local Primary Data External Validation of Model Top-ranked Features Preeclampsia Positive Preeclampsia Negative Feature ranking Technique Data splitting

7. Study Design with DL Models Online available data Acquisition Data Preprocessing Missing data imputation, Feature Engineering, Feature selection by Feature ranking Technique and Data splitting Training of Machine learning Models Preeclampsia Positive Preeclampsia Negative Best Model Selection Classical ML Models: MLP Classifier ElasticNet Linear Discriminant Analysis XGB Classifier Random Forest Classifier Logistic Regression ExtraTrees Classifier ADABoost Classifier KNN Classifier and Gradient Boosting Classifier

8. Study Design for Prevalence Primary Data Collection Data Processing Data Analysis Microsoft Excel and SPSS Prevalence Association with: Age Gravidity Sociodemographic variables like: Socioeconomic class Age Gravidity SBP DBP Sociodemographic variables like: Socioeconomic class Education Occupation