OHDSI Home | Forums | Wiki | Github

ATLAS 기능 estimation 시 high correlation between covariates and treatment detected error 문제

안녕하세요 경희대 박석준 입니다.
Subgroup 분석시 high correlation between covariates and treatment detected error가 나오던데
log.txt에서는

2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs High correlation between covariate(s) and treatment detected:
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs covariateId covariateName analysisId conceptId
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42629079412 drug_era group during day 60 through 120 days relative to index: atezolizumab 412 42629079
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42629079404 drug_era during day 60 through 120 days relative to index: atezolizumab 404 42629079
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42629079411 drug_era group during day 1 through 120 days relative to index: atezolizumab 411 42629079
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 45775965411 drug_era group during day 1 through 120 days relative to index: pembrolizumab 411 45775965
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42629079403 drug_era during day 1 through 120 days relative to index: atezolizumab 403 42629079
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 45775965403 drug_era during day 1 through 120 days relative to index: pembrolizumab 403 45775965
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42918862304 drug_exposure during day 60 through 120 days relative to index: 250 ML Sodium Chloride 9 MG/ML Injectable Solution [ISOTONIC SODIUM CHLORIDE DAEHAN] 304 42918862
2023-04-17 12:11:15 [Thread 1] INFO CohortMethod createPs 42918862303 drug_exposure during day 1 through 120 days relative to index: 250 ML Sodium Chloride 9 MG/ML Injectable Solution [ISOTONIC SODIUM CHLORIDE DAEHAN] 303 42918862
2023-04-17 12:11:15 [Main thread] ERROR CohortMethod runCmAnalyses Thread 1 returns error: “High correlation between covariate(s) and treatment detected. Perhaps you forgot to exclude part of the exposure definition from the covariates?”

다음과 같이 나옵니다.
해당 로그에 나오는 컨셉들 때문에 high correlation이 나타나는걸까요?
맞다면 해당 컨셉들을 제외하면 high correlation이 나오지 않고 잘 돌아갈까요?

안녕하세요, 아주대학교 의료정보학과 김청수입니다.

해당 로그가 나타내는 concept들이 high correlation 을 유발하는 것이 맞습니다. 해당 concept 들을 ps model 에서 exclude 하시기를 추천드립니다.

t