The research drew on data from more than 2,000 athletes who recorded their menstrual cycle length and their experience of bodily changes.
Apps for tracking menstrual cycles are widely available and becoming increasingly sophisticated. Some are used to help generate an awareness of one’s own body while others might want to use them for contraceptive purposes. Others still might need the app for making plans for travel or sport.
An issue with this can be that many of these apps aren’t great at predicting menstrual cycle length – a factor that proves particularly problematic for athletes who are planning their training. This is what researchers from Insight, the Science Foundation Ireland research centre for data analytics, wanted to address.
The study aimed to design a statistical model that accounted for variability both between menstrual cycles and within a person’s own cycle to identify patterns. This could then be used to inform decisions on training and health.
A broad range of experiences
First, the researchers recruited 2,125 athletes to log their menstrual cycles and the bodily changes they experienced relating to their cycle. This resulted in 16,524 cycles being logged.
The participants used a mobile app called FitrWoman. The app’s intended audience is athletes who want to use an understanding of their individual cycle to improve their performance and proved a useful tool in the research.
The free, calendar-based app enabled the participants to record their menstrual cycle and provide relevant information relating to wellness, nutrition and exercise based on the athlete’s predicted menstrual cycle and length.
This meant the users also inputted 25 variables including flow, bloating, constipation, mood and feelings of weakness.
Using the right kind of statistics
The researchers based in NUI Galway noted how previous studies have classed menstrual cycles into two groups: ‘standard’ and ‘menstrual dysfunction’. A standard cycle is that which is less than 35 days in length, while those above 35 days are classified as dysfunctional.
This is important as while the standard menstrual cycles fit within classical statistical methodologies, accounting for non-standard cycles takes some extra work. This involves using Bayesian statistics – a form of statistics that uses prior data to increase accuracy.
From this, the study found that the overall menstrual length without any reported symptoms was 27.4 days. Importantly it also found that some of the variables reported had a significant effect on predicting the length of the cycle.
Injury, stomach cramps and flow amount were all associated with an increased menstrual length. Meanwhile, the reporting of tender breasts was associated with a shorter cycle. These data were used to accurately predict many of the cycle lengths of the participants, showing that continued development on these apps could prove useful for women who want to track their cycles more accurately.
The research does acknowledge that the data is observational and depends on its participants logging their data. It also doesn’t help understand the causal mechanisms of variables on cycle length. The study concludes that future research would benefit from collecting more future information, such as polycystic ovary presence, daily diet and country of origin.