Clinical endpoint adjudication is performed using data and documents that go far beyond the anonymized information collected in the case report form of patients during a clinical study. The material necessary for endpoint adjudication may include reports, images, printouts and a variety of other files collectively called “source documents”. The risk of inadvertently revealing patient identifiers during the transfer is high and so is the need to efficiently redact documents before presenting them to the adjudicators. AI can help automate this tedious task and preserve patient confidentiality as well as company compliance.
Source documents presented to reviewers during a clinical trial adjudication can be in various formats, may contain images, handwritten notes, stamps or other features. Personal information can appear in the body of a text where it is easily identified but can also “hide” in many other places such as headers and footers, hand-written notes or even rubber stamps. In addition, black marker overlay is usually insufficient to completely obliterate the text.
In order to prevent confidentiality breaches, investigational sites are systematically trained and regularly reminded of the obligation to redact any personal identifiers before transmitting patient data to anyone outside of the site, but the responsibility to fully comply with confidentiality rules and regulations ultimately falls on the sponsor.
Double-checking all documents for anonymization before they are submitted to the adjudication committee adds workload and difficulty to the already heavy clinical trial effort for the sponsor or the sponsor’s third-party intermediates.
During clinical adjudication with the support of a web-based system, the investigational sites and the coordinator have access to the system for uploading source documents and can use the redaction functionality provided in the tool to save time and improve compliance.
How does the redaction technology work in Ethical’s eAdjudication®?
eAdjudication® integrates a Natural Language Processing (NLP) module that scans through documents uploaded in the platform as part of an event data package and detects items that could contain personal information. The tool is able to search in text data files but can also inspect scanned documents by converting the image into text with Optical Character Recognition (OCR) technology. When the automated search finds potential personal identifiers, the application highlights the suspected text. After checking the text, the user can confirm the parts that need redaction and mark the correction as done. Redaction is then complete and irreversible.
Image review has become an important component of endpoint adjudication in clinical trials and DICOM is the standardized format for medical images. eAdjudication® incorporates the necessary elements to allow manual or automatic redaction of DICOM metadata.
These features allow unbiased review by keeping the Adjudication committee totally blinded to personal information, thus ensuring that no bias is introduced in their judgement.
DOWNLOAD NOW THE FREE ENDPOINT ADJUDICATION HANDBOOK
The Complete Manual / Reference Book (34 pages) with all the topics related to the Independent Endpoint Adjudication Committees Management
