Auto IV: Counterfactual Prediction via Automatic Instrumental Variable Decomposition

Instrumental variables (IVs), sources of treatment randomization that are conditionally independent of the outcome, play an important role in causal inference with unobserved confounders. However, the existing IV-based counterfactual prediction methods need well-predefined IVs, while it’s an art rather than science to find valid IVs in many real-world scenes. Moreover, the predefined hand-made IVs could be weak or erroneous by violating the conditions of valid IVs. These thorny facts hinder the application of the IV-based counterfactual prediction methods. In this article, we propose a novel Automatic Instrumental Variable decomposition (AutoIV) algorithm to automatically generate representations serving the role of IVs from observed variables (IV candidates). Specifically, we let the learned IV representations satisfy the relevance condition with the treatment and exclusion condition with the outcome via mutual information maximization and minimization constraints, respectively. We also learn confounder representations by encouraging them to be relevant to both the treatment and the outcome. The IV and confounder representations compete for the information with their constraints in an adversarial game, which allows us to get valid IV representations for IV-based counterfactual prediction. Extensive experiments demonstrate that our method generates valid IV representations for accurate IV-based counterfactual prediction.

Effects of Iron Supplementation on Hemoglobin Concentration in 12-18 Month Old Iron-Deficient Anemic Infants

A randomized double blind clinical trial was conducted to study the effects of iron supplementation on hemoglobin values of 97 iron-deficient anemic (DA) and iron-sufficient (IS) infants, 12 to 18 months of age. IDA (n=SO) infants were assigned randomly to either receive dietary ferrous sulfate or placebo for 4 months. Similar treatment randomization was done among IS (n=4 7) infants. Before intervention, the mean hemoglobin value of the IDA infants was significantly (p<0.01) lower than that of the IS group. After intervention, there was an increase in the hemoglobin values of the IDA infants who had received iron, and was significantly (p<0.01) higher than those of the placebo treated IDA group, lS group, as well as placebo treated IS group. Hemoglobin values of the IDA infants who had received iron, increased up to the normal level as refer to the WHO's criteria used for normal hemoglobin values.

Randomization in a Substance Abuse Treatment Study: Participants Who Consent vs Those Who Do Not

Objective: A mixed research design study compared patients who consented to treatment randomization with patients who opted for usual care. Methods: Patients in substance abuse rehabilitation who consented to randomization (that is, consenting group [CG]) were compared with those unwilling to undergo randomization (that is, nonconsenting group [NG]) but who, nevertheless, underwent the experimental assessment sessions, which spanned from intake to 6-month follow-up. Results: Patients in the CG exhibited a longer history of drug abuse, less occupational stability at intake, and following intensive treatment, tended not to do as well in terms of recovery, compared with those in the NG. Conclusion: Inclusion of data from partial participants in research protocols may offer a viable strategy to better appraise data generalizability.