Over the last few years, LCPH has been creating surgical patient cohorts. Specifically, the hospital was able to create cohorts with specific target lengths of stay for pacemaker surgeries, Aortic valve repair, Mitral valve repair, and other cardiovascular procedures. Target based care is incredibly important as it allows for both better clinical workflow design and provides information to the patients about what they should expect during their stay. Our goal throughout this project was to automate the process of creating these cohorts to make it faster and less manually intensive.
- Create a centralized dataset which contains all the necessary features to filter out patients who are considered to be outliers.
- Create a workflow for cohort creation that allows for an easier and faster process.
- Run the workflow on the available cohorts and compare the median LOS with the historical results.
- Target based care carries tremendous value for both the hospital and the patients. Existing cohorts show a significant reduction in LOS variability after the introduction of the targets.
- Necessary data is highly fragmented and is at times difficult to access which is the main reason for the manually intensive process currently in place.
- The filtering criteria applied in the workflow created do a good job at identifying outlier patients which should be excluded from the cohort. The LOS extracted from the results of the algorithm closely resembled the numbers achieved through the manual process.
- Recreate all existing cohorts using the same workflow methodology and compare the resulting
- Continue to develop more features that will serve as filtering criteria applied to patients
- Provide clinicians with a standardized way to create new cohorts by accessing the centralized dataset and applying the filtering process to identify outlier patients