Ideas

DISCLAIMER: Some of the original content of the submitted ideas have been modified to fit the data available in MIMIC-II and to fit the time frame for a weekend-long hack-a-thon. If your idea is not present, you should receive an email with further discussion. Any questions, please feel free to contact jeffreydmarshall@gmail.com. Thank you all for your contributions and please continue to send your ideas.


The following are example cases demonstrating what can be explored within the MIMIC dataset.



1. Do dynamic patients variables create superior predictive models compared with static variables?


2. Is age an independent predictor of mortality or other comorbidities upon admission to the ICU?


3. Is acute kidney injury (AKI) an independent risk predictor for long-term mortality among patients who survive their initial ICU encounter?


4. Are serum magnesium concentrations clinically relevant to a patient’s blood pressure?


5. Is PPI use in the ICU associated with clinically important hypomagnesemia?


6. Are there ways to predict whether or not serial laboratory tests are likely to result in a clinically relevant change in care?


7. Can the addition of RDW to prediction scores improve their prognostic accuracy?


8. Are SSRIs/SNRIs independent indicators of mortality or comorbidities in the ICU?


9. What is the relationship between fluid resuscitation, vasoactive agents, or both, in short-term and long-term mortality risk among patients who have had a hypotensive episode?


10. What are the trends in severity of illness and mortality among the elderly after the addition of an 8-bed MICU?