Leveraging AI to understand Maternal Smoking Before & During Pregnancy and it’s correlation with Risk of Sudden unexpected Infant Death
The tragedy that never makes the headlines, yep Infant Sudden Unexpected Infant Death (SUID).
How do I know that? Well I haven’t heard of anything around SUID until I got the opportunity to work alongside Microsoft’s AI for good division and Seattle’s Children Hospital.
In the United States, >3700 infants die annually from sudden unexpected infant death (SUID), which includes sudden infant death syndrome (SIDS), accidental suffocation and strangulation in bed, and ill-defined causes.
Over the past few weeks I have been working on investigating the effects of maternal pre-pregnancy smoking, reduction during pregnancy, and smoking during pregnancy on SUID rates, but from a data side!
The experience has been a crazy one! When I first started working on AI projects I would be using Kaggle Datasets and now to be working with real hospital data and navigating just how to do that, it’s been a journey.
Let’s dive right into it!
The correlation of SUID and Smoking
It’s quite obvious that smoking has it’s concerns, with pregnancy and without, and most people know that it’s not a good idea to smoke while you’re pregnant, cause it can be harmful to a babies health.
What happens a lot of times is that women who realize they’re pregnant and have been smoking before, stop smoking right when they find out they’re not actually pregnant.
However what most people don’t realize is that smoking before and after pregnancy can still effect the health of the baby and the pregnancy it’s self.
Smoking during pregnancy can cause tissue damage in the unborn baby, particularly in the lung and brain, and some studies suggests a link between maternal smoking and cleft lip. During the first trimester the lungs and brain are beginning to develop so effect of smoking can be present during the first trimester.
Developing a Statistical Analysis
To understand the relationship between the reported average number of cigarettes smoked per day and risk of SUID, I got to develop both a logistic regression model and a generalized additive model (GAM).
Using 2017 data and a logistic regression model, we assessed the increased risk from pre-pregnancy smoking using a variable that identified the smoking habits before and during pregnancy. We then used a total of 3 log regression models to understand the effects of smoking in each trimester of a pregnancy.
We also examined the reduction in SUID risk when mothers quit or limited the amount they smoked compared with smokers who did not quit or limit smoking during pregnancy. I created a new categorical variable to identify the mothers who smoked in the first trimester and then quit, reduced, or continued the number of daily cigarettes in later trimesters.
If the number of cigarettes by the third trimester was 0, the mother was defined as having quit. If the number of total daily cigarettes in the second and third trimesters was less than the daily number of cigarettes in the first trimester multiplied by 2, the mother was categorized as a reduced smoker; those who continued to smoke the same amount (or more) were defined as continued smokers. In the model, we controlled for covariates and total number of cigarettes smoked during pregnancy.
To differentiate between SUID subcategories and non-SUID causes of death, we developed separate logistic regression models to estimate the risk of each cause of death independently.
**As this data/project is private, depth of data and code can not be shared.
It’s been found that the risk of SUID increases by double, when the mother is smoking. That’s CRAZY!
Smoking even one cigarette while pregnant doubles the risk of SUID. Any amount of smoking during pregnancy — even one cigarette — doubles the risk of SUID. For mothers who smoke 1–20 cigarettes per day, each additional cigarette increased the chance of SUID by 0.7 times.
Overall this has been an incredible experience and I have gotten to learn a ton about SUID and even just about how to work with real world data!
If you enjoyed this article: