Chapter 7 Conclusion

We successfully defined a metric quantifying the sleep quality of participants in MMASH dataset. We saw that bed time does not affect sleep quality but age and height do, although the relationship isn’t very strong. It was interesting to see that our metric captured sleep regulation by cortisol and melatonin in the case of some users. However, the results we got for screen usage were counter-intuitive, given that increased screen exposure is known to trigger inhibition of melatonin and release of cortisol.

A lot of improvements can be done in defining the metric. The biggest limitation was the small sample size and low diversity in the cohort. Due to these limitations the conclusions made possibly “overfit” the data i.e. they might not hold for a larger population. We believe that future work can expand on the following -

1. Increasing sample size by combining datasets with similar depth of sampling and variables
2. Imputing sleep quality value for user 11 having established all causal relationships
3. Include data corresponding to subjective matters like mental health, stress and anxiety in analysis to better quantify sleep quality
4. Replacing proportionality with equality along with the introduction of proportionality constants. More data is required to accurately determine the values of these constants and so for simplicity we assumed them to be equal to 1 when performing analysis