Researchers at EPFL and Stanford have carried out an analysis of the largest datasets from fertility awareness apps. Analyzing data from 200,000 users of the apps Sympto and Kindara, they have been able to make population-level observations regarding user demographics, tracking behavior patterns and accuracy in measuring menstrual health and ovulation.
Image: Visualizing the data from the modelling part of the study. Each line is a user, each dot is a day. The dot colors are inferred biological states; Red dots are the menses; black dots are the most likely day of ovulation; blue dots are the follicular phase days; yellow dots are the luteal phase days. Credit: L. Symul (EPFL/Stanford)
For many women of reproductive age, the most common way of assessing their menstrual health and fertility means regular visits to a gynecologist or another clinician. When it comes to evaluating changes in fertility, menstrual health, and quality of life, these visits typically rely on remembering, which can lead to significant inaccuracies in evaluation.
On the other hand, many women today are turning to so-called "fertility awareness method" (FAM) apps to support them in tracking their menstrual cycles. There is a multitude of such apps available, which only demonstrates how popular they’ve become in recent years.
But how accurate are FAM apps’ What do users track? Can they help them and their gynecologists’ These are hard to answer for a simple reason: Neither visits nor app use have been systematically studied on a population level to determine and compare their accuracy in evaluating menstrual health and fertility in a meaningful way.
This is what Laura Symul at EPFL’s Digital Epidemiology Lab has done. Working with Stanford University ( which she has now joined ), Symul led a large-scale study on 200,000 users of two FAM apps, Sympto and Kindara. Both apps support the "Sympto-Thermal Method" and facilitate the identification of the fertile and infertile times of a woman’s menstrual cycle by taking into account recordings of cervical fluid, body temperature at wake-up, and other biological signs.
The scientists tracked more than 30 million days of observations from over 2.7 million menstrual cycles. The overall study had two aims: First, to see how and what users voluntarily track on FAM apps. Second, to find out if these records allow an accurate detection and estimation of ovulation timing.
Analyzing the data, the scientists found that women who were seeking pregnancy recorded Sympto-Thermal measurements every single day for up to 40% of their menstrual cycles. Then, by modeling the data, they found that the average duration and range of the follicular phase, which begins the menstrual cycle and ends at ovulation, were larger than previously reported. In fact, the modeling showed that only 24% of ovulations occur at days 14 to 15 of the cycle. On the other hand, the data showed that the duration and range of the luteal phase - the latter part of the menstrual cycle - matched previous studies.
The findings offer an affordable means for studying the interactions between the menstrual cycle and other physiological systems on a large scale. "Our study provides a common ground for users and their doctors to incorporate digital records in their visits, evaluate their own menstrual patterns and compare them with the statistics we report," says Laura Symul.
She continues: "New technologies, and in particular self-tracking, are changing the way we perceive our bodies and health. Both users and doctors wonder about the opportunities and the usefulness of digital self-tracking. Our study shows that users voluntarily track their menstrual cycle and fertility-related body signs very frequently, and what they track is aligned with what is expected in the vast majority of cases. While these measurements and observations are noisy and not perfectly regular, they provide valuable information for inferring the underlying hormonal changes and timing of ovulation in a way that is scalable both in time and in number of participants."
Marcel Salathé, director of the Digital Epidemiology Lab adds: "The digital epidemiology approach presented here can help to lead to a better understanding of menstrual health and its connection to women’s health overall, which has historically been severely under-studied."
University of Geneva