Blood Glucose

Overview

Some text …………, backround……………

Objective

  1. The big Q: How many times on average is glucose level Optimal / Accepatable / Critical
  2. Can we predict glucose level using pattern over time ?

Import Data

A total of Xx…..

Trend analysis | overall

Let’s further dissect by:

  • “Before Breakfast”
  • “Before Lunch”
  • “Before Dinnner”
  • “2 Hrs Aft Breakfast”
  • “2 Hrs Aft Lunch”
  • “2 Hrs Aft Dinnner”

Before Breakfast

Before Lunch

Before Dinnner

2 Hrs Aft Breakfast

2 Hrs Aft Lunch

2 Hrs Aft Dinnner

Trend analysis | by Month

January Trend

Feb Trend

March Trend

April Trend

May Trend

Results | Overall

The big Q: How many times on average is glucose level Optimal / Accepatable / Critical

Results | Dissected by time of day

Let’s further dissect by:

  • “Before Breakfast”
  • “Before Lunch”
  • “Before Dinnner”
  • “2 Hrs Aft Breakfast”
  • “2 Hrs Aft Lunch”
  • “2 Hrs Aft Dinnner”

Before Breakfast

Before Lunch

Before Dinnner

2 Hrs Aft Breakfast

2 Hrs Aft Lunch

2 Hrs Aft Dinnner

Results | Dissected by Month

2021-01

2021-02

2021-03

2021-04

2021-05

Appendix | ML / AI

Time series is simply a series of data points ordered in time. In a time series, time is often the independent variable and the goal is usually to decipher patterns or make a forecast for the future.

EDA

Plotting the ACF estimation

An ACF measures and plots the average correlation between data points in a time series and previous values of the series measured for different lag lengths. For example, the correlation at the first lag is measured as the correlation between values of the time series measured at time t with all of the values for the series measured at time t − 1.

Plotting the PACF estimation

A PACF is similar to an ACF except that each correlation controls for any correlation between observations of a shorter lag length. Thus, the value for the ACF and the PACF at the first lag are the same because both measure the correlation between data points at time t with data points at time t − 1. However, at the second lag, the PACF measures the correlation between data points at time t with data points at time t − 2 after controlling for the correlation between data points at time t with those at time t − 1.

Prediction using ARIMA

Let’s try to model and predict

Prediction using ML

Conclusion on prediction

NOT predictable (only 55% accurate), more variables needed