CLINICAL

Protocol deviations in clinical trials present ongoing challenges, impacting data quality, patient safety, and trial efficiency. Traditionally documented as unstructured text, these deviations are difficult to analyse across trials, obscuring patterns and masking systemic issues.

The Protocol Deviation Monitoring DHP enables automated standardisation of unstructured PD text to structured categories using NLP.

This standardisation allows a suite of visualisations to be created at both the site and study level enabling systematic monitoring of PDs facilitating the identification of recurring issues proactively, supporting higher data quality and patient safety, which in turn allows for improvements to processes making sites more attractive to sponsors.

The DHP has been created with clinicians and data managers to ensure it address the clinical challenge of monitoring PD data.

DATA

User data is formatted for standardisation using a spreadsheet which is exported to .csv for ingest into the application.  The system ingests the .csv file and automatically categorises the unstructured data using NLP via an API to the AI application.  The system also allows for users to manually categorise the data prior to ingest, by-passing the automated AI categorisation.

Data categories assigned to the data have been based on unpublished cdisc (https://www.cdisc.org/standards) protocol deviation categories which have been peer reviewed by a clinician and data managers for further additions and refinements.