Big Data Study Provides Insight into Pregnancy Symptoms and How They Change Over Time

Fatigue, back pain or insomnia: during pregnancy, almost all women suffer from such symptoms. An interdisciplinary team of FAU researchers has now studied when such complaints are particularly common and how they progress. The team used an anonymized big data set from a pregnancy app.

Every pregnancy is unique, but almost all pregnant women experience similar pregnancy symptoms: they are tired, have back pain, suffer from constipation, trouble sleeping or shortness of breath.

“We have known about these symptoms for a long time. However, little research has been carried out so far into when they appear during pregnancy and how they influence each other,” explains Professor Björn Eiskofier. “We need to learn to better understand the context of these symptoms if we want to improve prenatal care and treatment options.”

The head of the FAU Machine Learning and Data Analysis Laboratory coordinates the interdisciplinary SMART Start research project together with Prof. Matthias W. Beckmann (Head of Department and Chair of Obstetrics and Gynecology) and Prof. Peter A. Fasching (Professor of Translational Gynecology and Obstetrics) from the Department of Obstetrics and Gynecology at the Universitätsklinikum Erlangen.

Also involved in the project are Professor Oliver Schöffski from the Chair of Health Management at the FAU and Professor Dr. Matthias Braun from the Chair of Systematic Theology and Ethics at the University of Bonn. Together, the researchers hope to encourage the digitalization of prenatal care in Germany based on a large dataset.

Tired in the first trimester

As part of the interdisciplinary research project, Michael Nissen, research associate and doctoral student at the Machine Learning and Data Analytics Lab, analyzed a big data set from German pregnancy app developer keleya. Expectant mothers can select their own symptoms in the keleya app. They then receive information and content tailored to their individual needs.

“The most common symptom among pregnant women is fatigue. It was selected by 92.9% of users. This is followed by back pain (92.6%), shortness of breath (81.0%) and sleep disorders (79.4%),” summarizes Nissen. “It is interesting to note that each symptom appears within a specific period of time,” explains the computer scientist.

Fatigue peaks during the first trimester of pregnancy, headaches begin particularly at the 15th week of pregnancy, and diarrhea tends to affect women mainly at the beginning and end of pregnancy, with an evident minimum around the 20th week. Sleep disturbances steadily increase as pregnancy occurs. the pregnancy is progressing.

There may be a link between sleep disorders and pregnancy complaints.

Some symptoms not only affect the expectant mother’s quality of life. They may also be linked to adverse consequences on pregnancy. For example, literature has proven that sleep disorders are linked to a higher risk of cesarean section, premature birth, and depression during pregnancy. It is therefore important to research the symptoms.

Large dataset available for research purposes

Keleya provided FAU with a large set of anonymized data from app users for research purposes, directly contributing to a better understanding of the topic, a good example of successful collaboration between industry and research. A total of 183,732 women tracked their pregnancy-related symptoms using the app’s symptom tracker.

They recorded more than 1.5 million symptoms. Researchers analyzed this huge dataset and compiled symptom progression curves with weekly curves for 15 different pregnancy-related symptoms. “The size of the dataset is significantly larger than previous work done in this area.”

Additionally, the dataset accurately reflects the real situation because it is based on real-world evidence. This could help reduce potential distortions and discrimination in medical research and provide insight beyond traditional medical studies.

User habits can make the use of health apps in scientific studies problematic. Some users only try the app once. “We were able to demonstrate that these data hardly differed from those of very active users,” says Nissen. As a result, one-time user data could be used for research purposes.

Overall, the study includes information on several previously unknown or controversial symptoms and how they change over time, and is considerably broader in scope than any other studies conducted to date. “Our study highlights the potential of secondary exploitation of sectoral data,” underlines the doctoral student. “Collaboration between science and industry can lead to new scientific discoveries.”

The results are published in the journal npj Digital Medicine.