{"id":"W4410376957","doi":"10.1002/pds.70155","title":"How Effective Are Machine Learning and Doubly Robust Estimators in Incorporating High‐Dimensional Proxies to Reduce Residual Confounding?","year":2025,"lang":"en","type":"article","venue":"Pharmacoepidemiology and Drug Safety","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Centre for Advancing Health Outcomes; University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Confounding; Estimator; Propensity score matching; Logistic regression; Proxy (statistics); Lasso (programming language); Statistics; Observational study; Residual; Econometrics; Regression; Medicine; Computer science; Machine learning; Mathematics; Algorithm","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0985372,0.001158655,0.002417563,0.00166325,0.0006411386,0.00319699,0.002542338,0.002537169,0.00203647],"category_scores_gemma":[0.3018132,0.0006088931,0.002582487,0.001885754,0.002894154,0.005146882,0.002205752,0.002885722,0.0005648548],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001101777,"about_ca_system_score_gemma":0.002903862,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003436741,"about_ca_topic_score_gemma":0.002523823,"domain_scores_codex":[0.9399809,0.05075742,0.001691677,0.003834677,0.003121722,0.0006134994],"domain_scores_gemma":[0.7385535,0.2314491,0.01142329,0.01209113,0.005521079,0.0009618383],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001339187,0.0002900862,0.1303255,0.002620172,0.005294572,0.0002748142,0.0008186765,0.2772378,0.0009572338,0.07675354,0.01054263,0.4935458],"study_design_scores_gemma":[0.0005392215,0.0008538559,0.03093107,0.002408379,0.001593369,0.0005640624,0.0005050868,0.7366117,0.003343215,0.1991062,0.02327215,0.0002715833],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.07510596,0.01614698,0.8909568,0.0122322,0.0004261497,0.0002825576,0.001184388,0.0006197575,0.003045418],"genre_scores_gemma":[0.6769701,0.005533405,0.3106252,0.003441226,0.0006502185,0.0006506626,0.001092202,0.0003006273,0.0007364032],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9014628,"threshold_uncertainty_score":0.5211208,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06151136878612086,"score_gpt":0.4128848080506421,"score_spread":0.3513734392645213,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}