Conclusion & Future Directions

Conclusion

Our analysis reveals a clear ~28-day glucose variability cycle in young diabetic females, aligning with the menstrual cycle and absent in males and older females. Key daily glucose features, such as the 95th percentile, number of peaks, and time spent out of range, highlight this pattern and align with previous studies (e.g., Herranz et al., 2016) showing phase-related glucose changes.

We hypothesize that hormonal shifts, particularly the drop in estrogen during menstruation, influence both metabolic processes and behaviors like increased hunger, contributing to observed glucose fluctuations. However, limitations of this study include data processing delays and a lack of direct hormonal measurements. Future research should address these gaps and examine how glucose-hormone dynamics differ across female subgroups.

These insights have important implications for personalized diabetes care. Understanding menstrual cycle-driven glucose changes could improve treatment strategies for cyclic women through tailored medication timing, lifestyle adjustments, and overall care. Our findings underscore the need to integrate hormonal rhythms into diabetes management for better outcomes.

Future Directions

Currently, the code used to process and plot the data has not been streamlined, so one future direction includes packaging our code into an accessible tool, enabling other researchers to process CGM data from various sources. We also envision developing an interactive dashboard where users can explore research findings or upload personal CGM data for individualized visualizations. This would make our insights more accessible and support more personalized diabetes care. 

Future analyses could deepen our findings by exploring differences between T1D and T2D, day vs. night patterns, treatment types, and more. A major limitation remains the lack of longitudinal hormonal data, which is critical for confirming menstrual cycle effects on glucose levels. Studies on conditions like PCOS, which elevate diabetes risk, should also be prioritized.

Lastly, future work will include clustering analysis to explore individual variation in hormonal cycles and expand to more diverse populations. Our current dataset, limited to CGM users with financial access, may not be fully representative. Greater inclusivity is essential to ensure both accuracy and equity in future diabetes research. 

Leader: Sally Ha