EP 4371130 A1 20240522 - METHOD OF EVALUATING TEXT SIMILARITY FOR DIAGNOSIS OR MONITORING OF A HEALTH CONDITION
Title (en)
METHOD OF EVALUATING TEXT SIMILARITY FOR DIAGNOSIS OR MONITORING OF A HEALTH CONDITION
Title (de)
VERFAHREN ZUR BEWERTUNG VON TEXTÄHNLICHKEIT ZUR DIAGNOSE ODER ÜBERWACHUNG EINES GESUNDHEITSZUSTANDS
Title (fr)
PROCÉDÉ D'ÉVALUATION DE SIMILARITÉ DE TEXTE PERMETTANT LE DIAGNOSTIC OU LA SURVEILLANCE D'UN ÉTAT DE SANTÉ
Publication
Application
Priority
- US 202117377612 A 20210716
- EP 21199570 A 20210928
- EP 2022069936 W 20220715
Abstract (en)
[origin: WO2023285688A1] The invention relates to a computer implemented method of training a machine learning model to evaluate the similarity of a candidate text to a reference text for determining or monitoring a health condition, where the model takes a text comparison pair comprising a reference text and a candidate text, each comprising data encoding a text sequence, the method comprising: pre-training an edit encoder to learn to generate an edit-space representation of an input text comparison pair, where the edit-space representation encodes information for mapping the reference text to the candidate text, the edit encoder comprising a machine learning model; and performing task-specific training by adding a task-specific network layer and training the task-specific network layer to map an edit- space representation generated by the pre-trained edit encoder to an output associated with a health condition. Edit-space representations learned in this way are able to encode a greater range of changes in language use than known metrics used to evaluate machine translations.
IPC 8 full level
G16H 50/20 (2018.01); G06N 20/00 (2019.01); G16H 50/70 (2018.01)
CPC (source: EP)
G06N 3/045 (2023.01); G06N 3/0455 (2023.01); G06N 3/088 (2013.01); G06N 3/096 (2023.01); G16H 50/20 (2018.01); G16H 50/70 (2018.01)
Designated contracting state (EPC)
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR
Designated extension state (EPC)
BA ME
Designated validation state (EPC)
KH MA MD TN
DOCDB simple family (publication)
DOCDB simple family (application)
EP 2022069936 W 20220715; EP 22754021 A 20220715