Systematic Literature Reviews are commonly used in medicine to judge the efficacy of medical interventions based on publications of clinical trials. However, systematic reviews can require months to years to complete if done manually. Therefore, there exists a significant interest in automating the systematic review process as much as possible. This talk will try to propose a few ideas on how machine learning models can help automate systematic reviews. We will focus on the step of citation screening, a very tedious and repetitive but also a time-consuming stage of systematic reviews. We will present research questions and the plan to approach them. In our experiments, we try to measure the quality of current state-of-the-art methods for using fewer manual annotations for training machine learning models in the citation screening step with the help of the eligibility criteria.