Remote Chemical Assay Classification via Out-of-Sample Extensions

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for classifying chemical reaction assays rely heavily on manual expert examination and computationally intensive techniques, which are inefficient and require significant resources, especially for remote or field laboratories lacking strong computational capabilities.

Innovation Solution

A system and method for remote chemical assay classification using a central computer to create a classification model, applying dimensionality reduction techniques like Diffusion Mapping, and out-of-sample extensions to enable classification on portable devices with limited resources, allowing for accurate classification without recalculating the entire embedded space.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual expert examination is used for classification, then classification accuracy is maintained, but productivity is low and loss of time is high

Engineering Contradiction:
Improveclassification accuracyVSAvoidclassification throughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies preliminary action by pre-processing the training data to extract relevant features and pre-computing the classification model parameters before actual classification tasks. This allows the system to quickly classify new assays without requiring manual expert examination, thereby maintaining accuracy while improving productivity

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical system of manual expert examination with an automated computational classification system. The system uses extracted features from assay data and applies pre-computed classification rules to automatically determine assay outcomes, eliminating the need for manual review while maintaining consistent accuracy

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If computationally intensive classification techniques are used, then classification accuracy is improved, but device complexity increases and use of energy increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidcomputational resource requirements
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and separates the computationally intensive steps from the classification process. Specifically, it extracts feature extraction and model training as preliminary steps that are performed once, and separates them from the actual classification of new assays. This allows portable devices to perform only the lightweight classification step locally while maintaining high accuracy

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs computationally intensive operations in advance by pre-training the classification model and pre-extracting features from training data. The pre-computed model parameters and extracted features are then transferred to portable devices, enabling them to perform accurate classifications without requiring significant computational resources during actual use

Inventive Principle:
Principle #10Preliminary action

3Device complexity

If dimensionality reduction techniques are applied, then computational complexity is reduced, but measurement precision may be lost

Engineering Contradiction:
Improvecomputational complexityVSAvoidclassification accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent extracts and applies dimensionality reduction techniques during the preliminary model training phase. By reducing the dimensionality of the feature space in advance, the system creates a simplified classification model that maintains accuracy while requiring fewer computational resources for actual classification tasks on portable devices

Inventive Principle:
Principle #2Taking out (Extraction)

4Measurement precision

If a large training set is used for classification, then classification accuracy is improved, but loss of time increases and use of energy increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidmodel training time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs the time-consuming model training on a large training set as a preliminary action in a centralized computing environment. Once the model is trained, it is deployed to portable devices where classification can be performed quickly without requiring access to the large training set, thus maintaining accuracy while eliminating the time and energy costs during field operations

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3301633B1Remote chemical assay classification
Publication Date: 2022.02.16 AZURE VAULT LTD
  • EP3301633B1 patent drawingFigure 1
  • EP3301633B1 patent drawingFigure 2
  • EP3301633B1 patent drawingFigure 3

AI summary

A portable device for remote chemical assay classification, comprising a computer processor, and an apparatus implemented on the computer processor, the apparatus comprising: an out-of-sample data receiver, configured to receive data defining an out-of-sample extension extracted on a remote computer from classifying test assays of a chemical reaction on the remote computer into at least two groups, and an assay classifier, in communication with the out-of-sample data receiver, configured to classify a new assay of the chemical reaction into one of the groups, using the data defining the out-of-sample extension.