CHOP

the following paper summarizes the BHS dataset notes on the dataset can be found here

Datasets

• CHOP emergency department:
  • What does this dataset contain?
    • Behavioral Health screen - used by CHOP since 2014 to give staff in the ER a sense of mental health burden for each child. Questions include environmental adversity and other data that are difficult to collect. Demographics are included.
    • Regular EHR are also included (medical notes made by clinician).
    • Geocoded level data (zipcodes)
  • Dataset summary:
    • n = 20,000 children, ages 14-18
    • 50% of sample are Black
    • prevelence of suicide attempt is 50%
    • 15% of the cohort have longitudinal data
    • 2500 children with two timepoints, 1500 with 3 or more timepoints
    • data have also been collected before and during covid, so this is another variable that could be of interest
    • no guarantee that any information from the past will be include about developmental disorders (whatever is shared by adolescents)
  • Other:
    • With suicidality, comorbidities are the rule, not the exception
    • Classification algorithms are good because they have clinical application and are interpretable
    • Causal effects are now target for intervention
• MRI dataset
  • what does this dataset contain?
    • structural data is collected clinically but not functionally. in total, there are about 100,000 scans at CHOP but many of the participants in the clinical dataset likely don’t have MRI scans.
    • there are also neonatal scans: 0-2 years of age + preterm children.
    • large subset of children have 22Q (22q deletion syndrome) can affect any system of the body, however, most children wtih 22q have heart, immune, learning, speech, and/or behavior difficulties.