BloodGlucose-Prediction

Code for Blood Glucose Level Prediction

1. Task : BG Prediction

• ML 혹은 DL 모델을 사용해서 30분 (6 point 이후)를 예측하는 regression model 개발
• Input Data : { $X_{t-7}, X_{t-6}, X_{t-5}, X_{t-4}, X_{t-3}, X_{t-2}, X_{t-1}, X_t$ } -> Output value : $X_{t+6}$

2. Data : T1DMS Virtual Patients BG Data

• BG level of 20 Virtual Patients
• 14 patients : Train Set / 6 patients : Test Set
• 데이터 1 point 당 5분, 5분마다 data point가 찍힌 것
• (연구실 내부 데이터로 공유는 어렵습니다)

3. Model

• MLP
• LSTM
• Support Vector Regression
• Random Forest Regression
• K-Nearest Neighbors

4. Metrics

• RMSE(Root Mean Squared Error)
• MAPE(Mean Absolute Percentage Error)
• Time Gain

5. Data Smoothing

• Moving Average

6. Result