Myeloproliferative Neoplasms (MPNs) are hematological malignancies that result from acquired driver mutations in hematopoietic stem cells (HSCs), causing overproduction of blood cells and an increased risk of thrombo-hemorrhagic events. The most common MPN driver mutation affects the JAK2 gene (JAK2V617F). Interferon alpha (IFNα) is a promising treatment against MPNs by inducing a hematological response and molecular remission for some patients. Mathematical models have been proposed to describe how IFNα targets mutated HSCs, indicating that a minimal dose is necessary for long-term remission. This study aims to determine a personalized treatment strategy. First, we show the capacity of an existing model to predict cell dynamics for new patients from data that can be easily obtained in clinic. Then, we study different treatment scenarios in silico for three patients, considering potential IFNα dose-toxicity relations. We assess when the treatment should be interrupted, depending on the response, the patient's age, and the inferred development of the malignant clone without IFNα. We find that an optimal strategy would be to treat the patients with a constant dose so that the treatment could be interrupted as fast as possible. Higher doses result in earlier discontinuation but also higher toxicity. Without knowledge of the dose-toxicity relationship, trade-off strategies can be found for each patient. A compromise strategy is to treat patients with medium doses (60-120 µg/week) for 10-15 years. Altogether, this work demonstrates how a mathematical model calibrated from real data can help build a clinical decision-support tool to optimize long-term IFNα therapy for MPN patients.