1. Compare our 95% CI with Matlab 95% confidence bounds. It looks quite close but I need further analysis.
2. Validate our 95% CI approach. For a specific patient ID, calculate 95% CI for two points first to then compare with three points and four points. We guess with more points in, 95% CI should be smaller and smaller.
3. Calculate 95% CI for median trajectory
1. Look at “G. Polglase-reterm birth: causes, consequences, treatment and prevention” and find a goldenberg’s paper
2. Goldenberg “Mechanisms of Disease: Intrauterine Infection and Preterm Delivery”.
3. Possible approach: Disease patients may be easy to predict (Periodontal disease/intrauterine infection/inflammation)
1. Using “BootStrapping” to put the 95% CI for “Median Trajectory Approach” and “Birthing Surface Approach”
2. Birthing Surface needs to adjust the cut to get the high value of odds ratio. It includes:
1. cut curve up&down
2. cut curve rotate
3. choose different shape of cut curve
3. It also needs to test two of the best then three of the best:
Two of best:
1. CRH+Cortisol
2. CRH+Prog
3. CRH+AFP
4. CRH+ALP
Three of best:
1. CRH+Cortisol+AFP
