A novel application of machine learning and zero-shot classification methods for automated abstract screening in systematic reviews

Zero-shot classification refers to assigning a label text (sentence, paragraph, whole paper) without prior training. This is possible by teaching the system how codify question and find its answer in text. In many domains, especially health sciences, systematic reviews are evidence-based syntheses of information related specific topic. Producing them demanding time-consuming terms collecting, filtering, evaluating synthesising large volumes literature, which require significant effort performed experts. One most steps abstract screening, requires scientists sift through various abstracts relevant papers include or exclude based on pre-established criteria. process subjective consensus between scientists, may not always be possible. With recent advances machine learning deep research, natural language processing, it becomes automate semi-automate this task. paper proposes novel application traditional zero-shot methods for automated screening reviews. Extensive experiments were carried out using seven public datasets. Competitive results obtained accuracy, precision recall across all datasets, indicate that burden human mistake might reduced.