fig3

Modeling treatment effect heterogeneity in prophylactic lumbar drainage: a Double Machine Learning reanalysis of EARLYDRAIN

Figure 3. Conditional average treatment effect (CATE) distributions for prophylactic lumbar drainage (LD) across six clinical outcomes. Each panel displays the estimated treatment effect (risk difference) from the Causal Forest model. Outcomes include: vasospasm, secondary infarction, and infection during ICU course (top row); and modified Rankin Scale (mRS ≤ 2), Glasgow Outcome Scale-Extended (GOS-E ≥ 5), and shunt dependency at six months (bottom row). These distributions suggest heterogeneity in treatment effects across patients. ICU: intensive care unit.

Artificial Intelligence Surgery
ISSN 2771-0408 (Online)
Follow Us

Portico

All published articles will be preserved here permanently:

https://www.portico.org/publishers/oae/

Portico

All published articles will be preserved here permanently:

https://www.portico.org/publishers/oae/