Machine Learning for Pore Electric Resistance Particle Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional pore electric resistance methods for identifying micron- to nano-sized particles, such as viruses and bacteria, face challenges in practical identification due to the use of pulse measurements from the same environment for learning and verification, which is not representative of real-world conditions where particles may vary in shape, surface charge, and host environment.
Innovation Solution
An apparatus and method that utilize a machine-learning program to analyze pulse waveforms from particles passing through pores, incorporating a storage system with host attribute information to differentiate between particles from various hosts, allowing for the calculation of optimized machine learning parameters using feature values from known particles from different hosts, enabling accurate identification of unknown particles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional pore electric resistance methods use pulse measurements from the same environment for both learning and verification, then the machine learning model can be trained and evaluated, but the identification accuracy deteriorates in practical clinical applications where particles vary in shape, surface charge, and host environment
Solution Approach 1:
The patent segments the learning dataset by host environment, creating separate learning datasets for each host type. This allows the machine learning model to learn host-specific characteristics while maintaining the ability to generalize across different hosts through multi-environment training
Solution Approach 2:
The patent changes the training parameters by incorporating host attribute information as additional input features to the machine learning model. This allows the model to adapt its identification based on the specific host environment, improving accuracy across diverse practical applications
2Ease of manufacture
If the machine learning model is trained using particles from a single host environment, then the training process is simplified, but the model fails to account for variations in particle characteristics across different hosts
Solution Approach 1:
The patent creates a universal machine learning model that can handle multiple host environments through a unified architecture. The model accepts host attribute information as input and produces reliable identification results across different hosts, making it universally applicable rather than host-specific
Solution Approach 2:
The patent performs preliminary action by collecting and organizing host attribute information before the machine learning training process. This pre-prepared host data is then integrated into the training dataset, allowing the model to learn host-specific patterns without complicating the overall training workflow
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables practical identification of particles applicable to clinical examinations by accounting for variations in host environments, improving the accuracy of particle type identification beyond conventional methods.
Implementation Method 1
measuring, by the pore electric resistance method, a transient change in ion current caused when target particles pass through pores
Implementation Method 2
The charged particles move by electrophoresis or the like and pass through the pores
Data Source
AI summary
An apparatus using a feature value extracted from a pulse waveform representing a transient change in ion current flowing between electrodes when a particle passes through a pore, as teacher data and data subject to analysis for machine learning. The apparatus includes a machine-learning program, a searcher, a host attribute table, and a feature value table, a host attribute table is searched using first host attribute information as a search key to extract a first host ID and a second host ID associated with the first host attribute information, a feature value table is searched using a first host ID as a search key to extract a first teacher feature value group obtained from first known particles of a first type, a feature value table is searched using a second host ID as a search key to extract a second teacher feature value group obtained from second known particles of the first type, learning is performed using the first teacher feature value group and the second teacher feature value group as teacher data and first particle type information representing the first type as a teacher label to calculate machine learning optimization parameters, and the machine learning optimization parameters with an input value that is a feature value group subject to analysis obtained from an unknown particle with a first host attribute are used to discriminate whether or not the unknown particle is of the first type.