Automatic Information Extraction of Patients Molecular Parameters from Clinical Trial Eligibility Criteria

Eligibility criteria are represented in clinical trial descriptions in the form of unstructured text. In this regard, when referring patients to clinical trials, doctors have to manually review the descriptions for patient compliance with the selection criteria. Partial automation of this process would significantly reduce its time cost. The crucial part of such automation is the automatic information extraction of eligibility criteria from the clinical trial descriptions. The eligibility criteria are often molecular disease parameters, that include gene and chromosomal mutations and changes in the expression level of specific genes. In this Master’s thesis, an algorithm was implemented that allows information extraction of disease molecular parameters. The algorithm outputs this information in a structured way, which is convenient for further use in systems that search, filter and rank clinical trials corresponding to the characteristics of a particular patient.