Mean aortic pressure (MAP) is a primary measurement for monitoring blood and oxygen delivery to major organs. Prolonged periods of hypotension, low MAP, lead to low tissue perfusion and subsequent end organ damage. Patients on mechanical circulatory support (MCS) devices, such as the Impella CP, are managed to maintain sufficient MAP for end-organ perfusion. Forecasting MAP is important for early warning of clinically concerning events, including hypotension and instability. It would also provide clinicians with context when weaning an MCS device. However, patients on MCS have varying pathologic and hemodynamic profiles. Patients presenting with cardiogenic shock as a result of acute myocardial infarction (AMI/CGS) have increased hemodynamic instability when compared to patients undergoing high-risk percutaneous coronary interventions (HRPCI). Existing deep sequence models for forecasting often focus on the same patient cohort and cannot generalize to different cohorts. In this paper, we examine how deep sequence models respond to the distribution shift of the MAP across the MCS patient cohorts during forecasting. We measure the forecasting performance across cohorts and the following error profiles: increasing, decreasing, and stationary trends, as well as clinically relevant MAP regions (critical, managed, normal). In a benchmark analysis, we found significant generalization errors across cohorts. We propose conditional RNN, a deep sequence model that learns to adapt to a different cohort by conditioning on time-invariant cohort features. Our proposed model improves the forecasting performance, achieving a 5.2 mmHg - 6.1 mmHg RMSE for cross-cohort patients.