Pathogen strain characterization from mixed strain samples

Overview

Development stage

In development or acquisitionSource: a) Pre-deployment – The use case is in a development or acquisition status.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

Not reported

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Science

Operational functions

Biometrics and identity, Clinical and medical Derived by keyword rules; see methodology

Problem the AI is intended to solve

We use the DNA sequences at specific places in pathogen genomes to create a "DNA fingerprint" that allows us to link cases of diarrheal illness and identify potential foodborne outbreaks. When a single patient has more than one strain of the same pathogen (e.g. two pathogenic E. coli), the pieces of the DNA fingerprint get mixed together in the sample and make the data unusable for outbreak surveillance. Our ML-based method is intended to sort the pieces of the DNA fingerprints into separate strains and make this data usable for foodborne outbreak surveillance.

Expected benefits

Surveillance for diarrheal foodborne outbreaks currently depends upon the availability of bacterial isolates obtained from patient stools to obtain the pathogen "genomic fingerprints" identifying pathogen strains. The availability of these isolates for fingerprinting is declining nationwide due to technological advancements that improve patient care. To maintain our ability to detect outbreaks without isolates, CDC is developing laboratory methods that obtain the pathogen genomic fingerprint directly from the patient stool specimen. However, patient stools frequently contain more than one strain of pathogen, so the ability to deploy these methods and maintain the sensitivity of foodborne outbreak surveillance is dependent upon development of this ML-based method to sort pathogen genomic fingerprint pieces from stool. Based on FoodNet data, we estimate that failure to implement these methods could lead to the loss of up to 75% of the samples currently captured by surveillance for some pathogens. Fewer surveillance samples will mean fewer outbreaks are detected and it will take longer to detect them, resulting in more people affected. For a sense of the scale of the challenge, NORS recorded ~300 outbreaks of Salmonella and E. coli in 2023 that were detected as a result of isolate-based surveillance. Economic impact evaluations have estimated that PulseNet surveillance alone prevents ~270,000 cases of foodborne illness in the US annually for a savings of at least $500,000,000 to the economy.

System outputs

Our ML-based method 1) predicts the number of strains of a pathogen found in a single sample, 2) reports the DNA fingerprint defining each strain, and 3) gives the likelihood that two samples contain the same pathogen strain.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Not reported

Public source code

Not reported

Sourcing

How it was built

Not reported

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Not reported

Privacy Impact Assessment

Not reported

Authorization to Operate

Not reported

Demographic features

Not reported

Training and evaluation data

Not reported

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported