Method and system for automated text
AssigneeHarrison-AI Pty Ltd
InventorBenjamin Clayton Hachey
Life of this patent
Estimating the term…
Abstract
A system for automated text anonymisation of clinical text, the system including an AI pipeline module to configure symbolic AI pipeline components for detecting protected health information (PHI) in the clinical text; a masking module for masking the detected PHI in the clinical text and generating a de-identified clinical text output file as well as a corresponding label file with de-identified information. The pipeline components may include at least one non-symbolic AI pipeline component or machine learning model.
Claims18
A system for automated text anonymisation of documents, the system comprising:
- at least two symbolic artificial intelligence (AI) pipeline components including named-entity recognition (NER) processes that detect personal information in at least a first document of the documents, at least one of the symbolic artificial intelligence (AI) pipeline components generating at least one label that indicates a type of the personal information and indicates a position of the at least one label in the first document;
- a masking component receiving the at least one label and applying a mask to the first document based on the position of the at least one label with the first document, the masking component generating a de-identified first document;
- and
- a zoning component, connected to at least one of the symbolic artificial intelligence (AI) pipeline components or the masking component, that executes a trained machine learning model to identify key areas of the first document,
- wherein:
- the symbolic AI pipeline components comprise a time-shift component that receives a first date or time being a second text item of the personal information from at least the first document and generates a second date or time by adding or subtracting a predetermined time period from the first date or time, and
- the second date or time is included in the at least one label.
A method for automated text anonymisation of documents, the method comprising:
- processing at least one first document of the documents by applying symbolic artificial intelligence (AI) pipeline components including named-entity recognition (NER) processes that detect personal information in the at least one first document;
- processing the at least one first document by applying a zoning component, connected to at least one of the symbolic artificial intelligence (AI) pipeline components or a masking component, that executes a trained machine learning model to identify key areas of the first document;
- outputting the personal information or the key areas individually as at least one label to a label file associated with the at least one first document, each of the at least one label including a type of the personal information or key area and a position of the personal information or key area in the at least one first document;
- and
- generating, via the masking component, at least one first de-identified document based on the label file and the at least one first document, the at least one first de-identified document having a mask at a mask position corresponding to the position of personal information in each of the at least one label in the at least one first document,
- wherein the method further comprises:
- receiving at a time-shift component of the symbolic AI pipeline components a first date or time being a second text item of the personal information from the at least one first document;
- generating a second date or time by adding or subtracting a predetermined time period from the first date or time;
- and
- incorporating the second date or time into a second label of the at least one label corresponding to the second text item of the personal information.
Dependencies are inferred from the claim text.