Coded Data Record Tokenization for Early Anomaly Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing systems fail to accurately predict anomalies in data records, such as computing system failures or health conditions, often requiring invasive and costly procedures, with low accuracy and high resource consumption.

Innovation Solution

A data processing system that uses machine learning models to tokenize data records, generating high-fidelity predictions of anomalies by screening for patterns indicative of future issues, allowing for timely mitigation actions with minimal computing resources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing systems use traditional methods to detect anomalies in data records, then they can identify some anomalies, but the accuracy is low and false negatives occur frequently

Engineering Contradiction:

Improveanomaly prediction accuracyVSAvoidfalse negative rate

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary tokenization and pattern matching on data records before anomalies actually manifest. By tokenizing data into standardized units and comparing against known anomaly patterns in advance, the system predicts potential anomalies before they become critical failures, thereby improving detection accuracy and reducing false negatives

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces token sequences as an intermediary representation between raw data records and anomaly detection. These token sequences serve as a bridge that transforms complex data into a standardized format that can be efficiently compared against anomaly patterns, significantly improving measurement precision while maintaining reliability

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If invasive tests and resource-intensive procedures are used to detect anomalies, then detection capability improves, but resource consumption and costs increase

Engineering Contradiction:

Improveanomaly detection capabilityVSAvoidcomputing resource consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The system extracts only the essential token sequences from data records that are relevant to anomaly detection, rather than processing entire data sets. This extraction approach maintains high detection capability by focusing on critical patterns while dramatically reducing computing resource consumption and operational costs

Inventive Principle:

Principle #2Taking out (Extraction)

3Loss of time

If traditional anomaly detection methods are used, then some anomalies can be detected, but the detection occurs too late for effective prevention

Engineering Contradiction:

Improvetime to detect anomalyVSAvoidability to prevent anomaly

Core Design Contradiction:

Loss of timeVSReliability

Solution Approach 1:

The system performs preliminary tokenization and pattern matching operations on data records before anomalies fully develop. By identifying anomalous token sequences in advance and comparing them against known failure patterns, the system detects potential issues early enough to enable preventive actions, reducing time loss and improving reliability

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12430586B1Generating high-fidelity predictions of changes to coded data records

Publication Date: 2025.09.30 GEORGETOWN UNIV

AI summary

Systems and processes described herein are for monitoring data records and predicting changes to coded values in the data records. More specifically, this disclosure relates to predicting and potentially preventing anomalies represented in data records by predicting values of entries to be added to the data records and causing a prophylactic response to the predicted anomaly to prevent the anomaly from occurring.