Present: Ingrid, Arezoo, Endre, Rune, Øystein
Topics: Who does what?
When?
Data: From Sintef.
Her følger en liten mengde konstruerte data (N=25000). See attachments
Variablene er:
- DAGB: Dummy hvorvidt det er dagbehandling
- POLK: Dummy hvorvidt det er poliklinisk konsultasjon
- INNL: Dummy hvorvidt det er innleggelse
- LOS: Liggetid
- DRG: Diagnose relatert gruppe
- Hoveddiag: Hoveddiagnose (ICD10)
- Bidiag1: Eventuell første bidiagnose
- Prosedyre: Eventuell prosedyre (kodeverk NCMP/NCSP)
- Inndato: Innskrivningsdato
- Kjønn: 1 menn, 2 kvinne
- Alder: Alder
Two common formats
1) Event-stream
2) I2B2 (Informatics for Integrating Biology and Bedside). Yearly Competitions (Natural Language Processing).
Collaboration
Øystein: CTex(t?) in Hexanord project?
Two types of patients:
Central Veins Catheter....
Sepsis... Infection in central vein leads to death of patients.
Find co-occurrences of Catheter and Sepsis (from natural language text).
Endre
-Wants to write his own thesis alone. Keep working on Anonymization. Will provide data for Arezoo and Ingrid.
Arezoo
-Simulator for temporal events... Produce data.
-Validator: Check consistences in (produced/real) data. Brain-surgery out-patients?! Home-care while Hospital-stay?!
(Øystein will contact: geoff.mcdonnell@unsw.edu.au, maybe simulation collaboration)
Ingrid is considering working on "Norwegian Named-Entity-Recognition", based on Thomas Brox Røst's Tagger. Collaborate with IFI's M.Sc. student: Rasch