Menstrual Date Prediction Model Utilizing Daily post sentiment Data on Social Media for Modeling

Menstrual cycle prediction is a critical issue for many women, as it can help them plan their daily activities, and prepare for potential physical and emotional changes. However, current methods for predicting menstruation often rely on physiological data such as age, cycle length, and ovulation history, which may not be convenient or accurate for some women.This paper presents a comparative study of menstruation prediction models using daily social media sentiment data. The underlying hypothesis is that social media posts can reflect the user’s emotions and feelings, which may be related to the menstrual cycle. By tracking emotions from social media posts, it may be possible to monitor potential symptoms associated with the menstrual cycle, such as premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD).The proposed models are evaluated using a dataset of social media posts and self-reported menstruation data. The results demonstrate that the proposed models can effectively predict the onset of menstruation and identify potential symptoms of PMS and PMDD. This work has the potential to contribute to the development of new tools and interventions for women’s health and well-being.This experimental research utilized a dataset of posts on X from one woman over a 3-year period (2020-2023), covering both normal and abnormal menstrual cycles. The challenge of this research was to transform ordinary text that could be posted on social media but could not be used for prediction into text that could be used for in-depth prediction and analysis. This was achieved through two processes: 1) Sentiment Analysis using the WangchanBERTa model to determine the sentiment of the posts, which achieved an accuracy of 61.7%; 2) the Time Series Forecasting process for predicting the date of menstruation using the Random Forest model was the most appropriate if compared to Linear Regression and SVR, which is considered to be the most accurate. Challenge new faces This is because there has been no study of predicting the menstrual cycle from emotions using social media messages.This study demonstrates that sentiment data from social media posts can be used as a predictor for menstrual cycles. The Random Forest model can be used as a model for predicting menstruation dates without relying on physical data.