Machine Learning Mold Detection via DNA Sequencing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting mold growth due to water damage in buildings are inaccurate, subjective, and fail to account for the diversity of fungal communities, leading to ineffective assessment of remediation needs.
Innovation Solution
A computer-implemented method using machine learning estimators, specifically trained on DNA sequences from dust samples, to distinguish between structures with and without mold growth due to water damage, incorporating techniques like Random Forest classifiers and Amplicon Sequence Variants analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If visual inspection, culturing, and microscopic assessment methods are used, then the assessment process is simple and accessible, but the accuracy and reliability of mold detection is insufficient
Solution Approach 1:
The patent replaces manual visual inspection, culturing, and microscopic assessment with automated DNA sequencing and machine learning analysis. The system extracts DNA from environmental samples, sequences fungal genomic regions, and uses trained machine learning models to automatically identify mold presence and types, eliminating subjective human judgment while maintaining operational simplicity through automated workflows.
Solution Approach 2:
The patent introduces DNA sequencing as an intermediary between sample collection and mold assessment. Instead of directly observing mold or culturing it, the system sequences fungal DNA from environmental samples and uses machine learning models as intermediaries to interpret the genetic data, bridging the gap between raw biological material and actionable assessment results.
2Measurement precision
If qPCR-based methods like ERMI are used, then the ability to distinguish mold types is improved, but the number of mold identification possibilities is limited
Solution Approach 1:
The patent changes the fundamental parameter of detection from targeted qPCR assays to broad-spectrum DNA sequencing. Instead of using a fixed panel of 36 specific qPCR assays, the system sequences fungal genomic regions (such as the internal transcribed spacer region) from all fungi present in the sample, enabling identification of any fungal species within the sequenced regions and dramatically expanding mold identification possibilities.
Solution Approach 2:
The patent makes the detection system universal by using DNA sequencing that can identify any fungal species within the targeted genomic regions, rather than being limited to specific pre-selected mold types. The machine learning models are trained on diverse fungal genomic data and can adapt to identify various mold types, making the system applicable to a wide range of fungal communities without requiring assay-specific modifications.
3Speed
If active air sampling is used, then real-time mold detection is achieved, but the system becomes costly and difficult to install
Solution Approach 1:
The patent uses disposable environmental sampling materials (such as dust collection surfaces or swabs) that can be easily deployed and discarded, replacing complex active air sampling systems. These simple, low-cost sampling materials collect fungal DNA from the environment, which is then analyzed through DNA sequencing and machine learning, achieving accurate mold detection without requiring expensive, complex installation infrastructure.
4Ease of operation
If specific genera like Aspergillus/Penicillium are used as indicators, then the assessment method is simple, but the diversity of fungal communities is not adequately accounted for
Solution Approach 1:
The patent changes the assessment parameter from monitoring specific indicator genera to comprehensive fungal community analysis. Instead of measuring only Aspergillus or Penicillium abundances, the system sequences DNA from all fungi in the sample and uses machine learning models trained on diverse fungal genomic data to assess the entire fungal community, capturing both common and rare species while maintaining computational efficiency.
Data Source
AI summary
A computer-implemented method includes: receiving a set of DNA sequences extracted from one or more dust samples collected from a structure; analyzing the sequences using a machine learning estimator, where the machine learning estimator has been trained to distinguish structures with mold growth due to water damage from structures without mold growth due to water damage; and determining if the structure has mold growth due to water damage.