first-icu-stay

To identify the first ICU stay in MIMIC-IV, you derive sequence fields like icustay_seq and first_icu_stay to isolate incident admissions. This prevents correlated observations by ensuring only the initial ICU stay is included per patient.

Readmission exclusion removes subsequent hospital and ICU visits to maintain patient-level independence in cohort construction. By isolating the first hospital admission using first_hosp_stay, you prevent correlated data points from skewing your analysis.

You exclude readmissions by applying sequential ranking to hospital and ICU stays, then filtering for first_hosp_stay and first_icu_stay. This removes repeat visits from mimiciv_hosp and mimiciv_icu to ensure independent observations.

Yes, you can apply age shifting and duration-based filters alongside first_icu_stay flags to select incident ICU patients. This allows you to build clean adult cohorts while handling mortality flags within the MIMIC-IV dataset.

Extracting the first hospital admission requires the mimiciv_hosp and mimiciv_hosp.patients tables. These tables provide the necessary data to generate hospstay_seq and first_hosp_stay fields for isolating incident hospitalizations.

Correlated observations occur when multiple ICU stays per patient are included. You resolve this by applying readmission exclusion logic to retain only the first_icu_stay, ensuring each patient contributes a single independent event to the cohort.