Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Modelling of classification rules on metabolic patterns including machine learning and expert knowledge.

Christian Baumgartner1, Christian Böhm, Daniela Baumgartner

  • 1Research Group for Biomedical Data Mining, Institute for Information Systems, University for Health Sciences, Medical Informatics and Technology, Innrain 98, A-6020 Innsbruck, Austria. christian.baumgartner@umit.at

Journal of Biomedical Informatics
|March 31, 2005
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A new flow chip in combination with multiphoton microscopy as a protocol for longitudinal 3D imaging of tissue calcification under shear stress.

FEBS open bio·2026
Same author

Cancer: A bioelectric disease?

Clinical and translational medicine·2026
Same author

Timing and outcome prediction of intravenous thrombolysis in posterior circulation stroke: Insights from the Austrian Stroke Unit Registry.

European stroke journal·2026
Same author

Miniaturized device for assessing calcification propensity of biohybrid implants under continuous flow.

Lab on a chip·2025
Same author

Stereo cell: A new approach to the next generation of clinical precision medicine.

Clinical and translational medicine·2025
Same author

Problem Gambling and its Relation to Delayed Reward Discounting: A Secondary Analysis of a Randomized Controlled Trial.

Journal of gambling studies·2025

Machine learning effectively identifies markers for metabolic disorders like phenylketonuria, medium-chain acyl-CoA dehydrogenase deficiency, and 3-methylcrotonyl-CoA carboxylase deficiency. Logistic regression analysis with established flags improved diagnostic specificity for these conditions.

Area of Science:

  • Biochemistry
  • Computational Biology
  • Genetics

Background:

  • High-dimensional metabolic data presents challenges for identifying disease markers.
  • Machine learning offers a powerful approach for marker discovery without prior assumptions.
  • Severe metabolic disorders require accurate and efficient diagnostic tools.

Purpose of the Study:

  • To investigate metabolic patterns associated with phenylketonuria (PAHD), medium-chain acyl-CoA dehydrogenase deficiency (MCADD), and 3-methylcrotonyl-CoA carboxylase deficiency (3-MCCD).
  • To develop and compare machine learning classification models for disease screening and diagnosis.
  • To evaluate the utility of established diagnostic flags in enhancing model specificity.

Main Methods:

  • Utilized machine learning, specifically decision tree and logistic regression analysis (LRA), to analyze metabolic data.

Related Experiment Videos

  • Constructed classification models for PAHD, MCADD, and 3-MCCD.
  • Assessed the relevance of established biochemical diagnostic flags for LRA model refinement.
  • Compared classification accuracy, sensitivity, and specificity between the decision tree and LRA models.

    • Main Results:

    • Both decision tree and LRA models achieved high sensitivity (>95.2%) for disease classification.
    • LRA models incorporating established diagnostic flags demonstrated significantly enhanced specificity.
    • The false positive rate in the LRA models with flags was exceptionally low, not exceeding 0.001%.

    Conclusions:

    • Machine learning, particularly LRA with biochemical flags, is highly effective for screening and diagnosing metabolic disorders.
    • Established diagnostic flags significantly improve the specificity of machine learning-based diagnostic models.
    • These findings support the integration of machine learning into newborn metabolic screening protocols.