/Patent Tracker
AmazonGranted patentAI & softwareInterest 4/10

Amazon patents method to measure confidence in medical text negation

Amazon describes a system for analyzing clinical data that assigns a probability score to whether a medical condition is negated in a sentence, rather than just marking it as present or absent. The method uses statistical measures like pointwise mutual information to calculate how likely a specific word indicates a negation, providing a more nuanced confidence level than traditional binary checks.

Why it mattersThis improves the accuracy of automated medical record processing by allowing software to distinguish between strong and weak evidence of negation, which is useful for healthcare data analysis.

In the filing’s own words

Approaches for determining continuous confidence scores for rule-based negation detection systems overcome deficiencies of existing rule-based negation detection systems that provide only binary output (e.g., negated or not negated). The approaches encompass a pointwise mutual information (PMI)-based approach in which PMI is measured between negation cue words and gold-standard negation annotations in a reference dataset. The approaches also encompass a conditional probability-based approach in which the condition probabilities that gold-standard negations exist in a reference dataset for negation cue words are measured. In both approaches, a form of association between negation cue words and gold-standard negations are measured to provide a more accurate indication of the rule-based system's confidence in a negation determination compared to a binary output.

What exactly they claim (claim 1)

1. A method comprising: obtaining, by a provider network implemented by one or more computing devices, a set of clinical data; detecting, by the provider network, an entity in a context of the set of clinical data; using, by a natural language processing (NLP) pipeline implemented as NLP pipeline code executed by one or more processors of the one or more computing devices in the provider network, a machine learning-based system to determine that the entity in the context of the set of clinical data is not negated; obtaining, by the NLP pipeline, a first negation score reflecting a probabilistic confidence of the machine learning-based system that the entity in the context is negated; using, by the NLP pipeline, a rule-based system to determine that the entity in the context is negated; determining, by the NLP pipeline, a continuous negation score according to a pointwise mutual information approach, the continuous negation score reflecting a probabilistic confidence of the rule-based system that the entity in the context is negated; determining, by the NLP pipeline, an overall negation score based on the continuous negation score, the overall negation score reflecting a probabilistic confidence that the entity in the context is negated; and based on determining using the rule-based system that the entity in the context is negated, causing, by the NLP pipeline, a graphical user interface to be displayed that comprises an annotated result, the annotated result comprising an indication that the entity in the context is negated and comprising the overall negation score.

Facts

Company
Amazon (Amazon Technologies, Inc.)
Official title
Continuous confidence scoring for a rule-based negation detection system
Patent number
US12737649B1
Application number
18129760
Filed
Mar 31, 2023
Granted
Sep 15, 2026
Inventors
2
Claims
20

Classified under

G06N 5/025 Physics & computingG16H 50/20 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Amazon in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.