Towards the development of a diagnostic test for autism spectrum disorder: Big data meets metabolomics

Qureshi, Fatir; Hahn, Juergen

doi:10.1002/cjce.24594

Cited by 6 publications

(4 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Capping the number of input features at 5 maximizes the classifiers’ performance for this dataset while also minimizing potential overfitting concerns. This is also in line with the findings from Howsmon’s [ 30 ] and Qureshi’s study [ 42 ]. To validate the results of this feature selection, a similar procedure was carried out where the data were split the same way 1000 times and the combinations yielding the highest testing accuracy from each split were saved.…”

Section: Methodssupporting

confidence: 93%

“…Using that panel of seven metabolites, Howsmon was able to achieve a cross-validatory accuracy of 96.9% using LDA [ 30 ]. In a more recent study using the same dataset, Qureshi et al was able to achieve a cross-validatory accuracy of 97% using SVM and five metabolites [ 42 ]. Again, all features used in this classifier were present among the 11 identified by the factor analysis.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Framework for Testing Robustness of Machine Learning-Based Classifiers

Chuah

Krüger

Wang

et al. 2022

JPM

Self Cite

View full text Add to dashboard Cite

There has been a rapid increase in the number of artificial intelligence (AI)/machine learning (ML)-based biomarker diagnostic classifiers in recent years. However, relatively little work has focused on assessing the robustness of these biomarkers, i.e., investigating the uncertainty of the AI/ML models that these biomarkers are based upon. This paper addresses this issue by proposing a framework to evaluate the already-developed classifiers with regard to their robustness by focusing on the variability of the classifiers’ performance and changes in the classifiers’ parameter values using factor analysis and Monte Carlo simulations. Specifically, this work evaluates (1) the importance of a classifier’s input features and (2) the variability of a classifier’s output and model parameter values in response to data perturbations. Additionally, it was found that one can estimate a priori how much replacement noise a classifier can tolerate while still meeting accuracy goals. To illustrate the evaluation framework, six different AI/ML-based biomarkers are developed using commonly used techniques (linear discriminant analysis, support vector machines, random forest, partial-least squares discriminant analysis, logistic regression, and multilayer perceptron) for a metabolomics dataset involving 24 measured metabolites taken from 159 study participants. The framework was able to correctly predict which of the classifiers should be less robust than others without recomputing the classifiers itself, and this prediction was then validated in a detailed analysis.

show abstract

Section: Methodssupporting

confidence: 93%

Section: Resultsmentioning

confidence: 99%

Framework for Testing Robustness of Machine Learning-Based Classifiers

Chuah

Krüger

Wang

et al. 2022

JPM

Self Cite

View full text Add to dashboard Cite

show abstract

“…In ‘Towards the Development of a Diagnostic Test for Autism Spectrum Disorder: Big Data Meets Metabolomics,’ Fatir Qureshi and Jeurgen Hahn present their latest progress in their long‐term quest to improve autism spectrum disorder (ASD) diagnosing techniques. [ 2 ] This work is part of Prof. Hahn's personal quest to make ASD diagnostics more accurate, more specific, and more applicable in the earlier stages of a child's development. It is a challenging problem because ASD symptoms present differently from case to case.…”

Section: Using Pse Technology To Diagnose Autismmentioning

confidence: 99%

Special Series: Systems and Control Division

Adams

2022

Can J Chem Eng

View full text Add to dashboard Cite

Dear Readers of The Canadian Journal of Chemical Engineering, I am pleased to introduce our newest Special Series, containing invited papers from the Canadian Society for Chemical Engineering's (CSChE) Systems and Control Division. We hand-picked the first submissions in the series from the best submissions to the 2021 Canadian Chemical Engineering Conference (CCEC) held virtually from Montréal. The papers highlight some of the interesting diversity of research going on within the division, including how the field is starting to move into nontraditional application areas, such as medicine and global health. In fact, nearly half of the extended abstracts published in the Systems & Control division's CCEC 2021 conference proceedings [1] relate to biotech topics such as vaccine development, N95 mask sterilization, nutrient uptake from beverages, and other biotech applications.

show abstract

“…It is shown that one can determine with high accuracy if a sample came from a child with an ASD diagnosis, thereby offering the potential for biochemical tests to support an ASD diagnosis in the future. [ 1 ]…”

mentioning

confidence: 99%

Issue Highlights

2022

Can J Chem Eng

View full text Add to dashboard Cite

Autism spectrum disorder (ASD) is currently diagnosed only by observations. As such, the current practice does not look at differences in metabolites that are common between children with and without an ASD diagnosis. This work uses a machine learning-based approach applied to metabolomic data measured from blood samples to distinguish between samples taken from children with an ASD diagnosis versus their typically developing peers. It is shown that one can determine with high accuracy if a sample came from a child with an ASD diagnosis, thereby offering the potential for biochemical tests to support an ASD diagnosis in the future. [1]

show abstract

Towards the development of a diagnostic test for autism spectrum disorder: Big data meets metabolomics

Cited by 6 publications

References 48 publications

Framework for Testing Robustness of Machine Learning-Based Classifiers

Framework for Testing Robustness of Machine Learning-Based Classifiers

Special Series: Systems and Control Division

Issue Highlights

Contact Info

Product

Resources

About