A healthcare company hosts a SageMaker endpoint that predicts patient readmission risk. They prioritize high recall and will tolerate false positives. The current model degraded, so they trained and deployed a new model as a shadow variant and monitored live traffic for one month. The shadow has higher recall but lower precision than the existing model. What should the company do next?

The company prioritizes high recall and tolerates false positives. The shadow variant has higher recall, aligning with their business objective, even though its precision is lower. Therefore, promoting it to full production is the logical next step. Extending the shadow testing period is unnecessary because the model already meets the primary objective. Blue/green deployment is for testing with a small portion of live traffic, but the shadow variant has already been monitored with live traffic for a month. Disabling the shadow variant and rolling back would mean discarding a model that better meets the company's stated objective.