Referat PASTAS:
Present: Øystein, Arezoo, Rune, Ingrid
Each student make a personal space in Confluence and add a link to it in https://www.ntnu.no/wiki/display/pastas/Student+Projects|../../../../../../../../../../display/pastas/Student+Projects|||||||||||||||||\. Upload your introduction/resources here.
Meetings are now available at:
Arezoo:
format of data: patient number, date, group, icd-10, procedure.
Group patients by icd-10, patients with same diagnosis, try to follow trajectory for all patients with one diagnosis. Follow procedure for one group/trajectory.
Prob: How do we know if a new diagnosis is new or if the diagnosis changed. Separate chronic deseases, temporary diagnosis, diseases, trauma, accidents - Clustering dependent on type/group. Think about how we can group icd-10 codes. Different coding exist for trauma an chronic diseases.
Group patients by icd-10, patients with same diagnosis, try to follow trajectory for all patients with one diagnosis. Follow procedure for one group/trajectory. Prob: How do we know if a new diagnosis is new or if the diagnosis changed. Separate chronic deseases, temporary diagnosis and other events.
ICD-10 - classification of diseases - (example: no code for fear of having cancer)
ICPC2 - categorization of patients - why someone is a patient (two-dimensions: letter - organ/type of problem and letter 01-29 is a symptom. from 30 are preventive interventions, then therapeutic intervention, test results, adm interventions, 70- red ones are diseases ), wheather it is a intervention, a d:iagnosis or something else
finnkodekith.no
CCI - chronic condition indicator
suggestion: look at req for representing different temporal aspects of patient state. breaking leg and having diagnosis are different things - keep them apart. represent and reason about what is related and what is not.
Separation into Subtrajectories/related events.
Cooccurance of icd-10 codes. can assume that they are related. can be used to decide if events are related. for all pairs of events look at relatedness, make a measure of nearness, probability