Machine-learning–driven biomarker discovery for the discrimination between allergic and irritant contact dermatitis

Abstract

Contact dermatitis tremendously impacts the quality of life of suffering patients. Currently, diagnostic regimes rely on allergy testing, exposure specification, and follow-up visits; however, distinguishing the clinical phenotype of irritant and allergic contact dermatitis remains challenging. Employing integrative transcriptomic analysis and machine-learning approaches, we aimed to decipher disease-related signature genes to find suitable sets of biomarkers. A total of 89 positive patch-test reaction biopsies against four contact allergens and two irritants were analyzed via microarray. Coexpression network analysis and Random Forest classification were used to discover potential biomarkers and selected biomarker models were validated in an independent patient group. Differential gene-expression analysis identified major gene-expression changes depending on the stimulus. Random Forest classification identified CD47, BATF, FASLG, RGS16, SYNPO, SELE, PTPN7, WARS, PRC1, EXO1, RRM2, PBK, RAD54L, KIFC1, SPC25, PKMYT, HISTH1A, TPX2, DLGAP5, TPX2, CH25H, and IL37 as potential biomarkers to distinguish allergic and irritant contact dermatitis in human skin. Validation experiments and prediction performances on external testing datasets demonstrated potential applicability of the identified biomarker models in the clinic. Capitalizing on this knowledge, novel diagnostic tools can be developed to guide clinical diagnosis of contact allergies.

Development of a Multi-Target Strategy for the Treatment of Vitiligo via Machine Learning and Network Analysis Methods

“…With the framework, we used an interpretable machine learning method to extract important transcriptomic features for vitiligo. It was a practicality for most biological studies (Fortino et al, 2020;Shen et al, 2020;Unger et al, 2020). Among these important transcriptomic features, several top-ranked variables were closely related to the melanogenesis and immune characteristics, which was the main phenotypes of vitiligo (Picardo et al, 2015;Niu and Aisa, 2017;Pu et al, 2021;Zhang et al, 2021).…”

Weighted Gene Co-Expression Network Analysis and Machine learning Identify Critical Genes in the Development of Osteomyelitis after Staphylococcus aureus Infections

“…The gene expression datasets usually contain many different noises, and random forest (RF) has good anti-noise ability. Because it introduces randomness, it is better at control over-tting [25,26] to avoid false variables. RF uses a random method to build a forest composed of many decision trees, and there is no correlation between each decision tree in the random forest.…”

Development of a Deep Learning Model for Early Alzheimer’s Disease Detection from Structural MRIs and External Validation on an Independent Cohort

“…These methods have been shown to outperform traditional techniques based on predefined features in most image processing and computer vision tasks 20,21 . In the biomedical field, CNN-based methods also have the potential to reveal new imaging biomarkers 22,23 based on 3D AlexNet, 3D Resnet, patch based models, Siamese networks, auto-encoder based models, among others [24][25][26] . Based on systematic reviews and survey studies 27,28 , many of previous approaches had major limitations in their design or validation: Most of these studies focus on distinguishing Alzheimer's disease dementia patients from normal controls.…”

“…These methods have been shown to outperform traditional techniques based on predefined features in most image processing and computer vision tasks 20,21 . In the biomedical field, CNN-based methods also have the potential to reveal new imaging biomarkers 22,23 . Multiple studies have addressed mild Alzheimer's disease dementia detection from MRI via deep learning, with notable examples of 3D convolutional neural networks based on 3D AlexNet, 3D Resnet, patch based models, Siamese networks, auto-encoder based models, among others [24][25][26] .…”

“…Based on the deep learning model,23.02% (95% CI: 21.43% -24.62%) of subjects in group A (blue line) progress to AD, compared to 8.81% (95% CI: 8.09% -9.53%) of subjects in group B (red line). For the ROI-volume/thickness model, 20.22% (95% CI: 18.49% -21.95%) of subjects in group A (blue line) progress to AD, compared to 11.11% (95% CI: 10.32% -11.89%) of subjects in group B (red line).Figure 2: Progression analysis for MCI subjects.…”