Fully automated state of the art

Fully automated State of The Art

Fully automated State of The Art

Machine learning techniques for the automation of literature reviews (State of The Art).
State of The Art in Clinical Evaluation Reports (CER) is a highly structured process that takes several steps to be completed, often involving several reviewers working together over a long period of time.

We develop automated methods for acquisition and discovery of medical knowledge embedded in CER according to MDR 2017/745. A Natural Language Processing (NLP) system, is applied to extract and encode clinical entities from scientific literature (PubMed and Embase).

Once the obtained articles are transformed into a manageable input format, one application of Machine Learning Techniques is found in the construction of classifiers that can be employed to determine which of the retrieved articles are relevant for the study and which articles can be omitted in any further analysis.

Quantitative and qualitative analysis are based on statistical methods adjusted by volume tests. We focus on two types of entities, clinical performance and safety per indication. The automated method is generalizable and can be applied to detect other equivalent and benchmark devices. Therefore, it makes it easier for manufacturers to construct a bank of State of The Art that can be used for similar medical devices.