Prediction of oral food challenge outcomes via ensemble learning

Zhang, Zuopeng; Lee, Dong Hoon; Jungles, Kylie N.; Shaltis, Diane; Najarian, Kayvan; Ravikumar, Rajan; Sanders, Georgiana; Gryak, Jonathan

doi:10.1016/j.imu.2022.101142

Cited by 8 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Flare size was the most contributive feature in predicting OFC outcome as determined by SHAP for the LEAP cohort. This corroborates the finding in Zhang et al, 11 which utilized a data set distinct from that used in this study. As shown in Fig 3 , A , larger flare sizes are associated with adverse OFC outcomes, with all cases having a flare size of at least 17 mm resulting in reactive OFCs.…”

Section: Discussionsupporting

confidence: 93%

“…To our knowledge, this study is the first in FA to use machine learning on a well-reported, well-controlled, clinical trial data set to quantify OFC likelihood based on readily available clinical parameters. This study can be viewed in part as validating the prior work from our group in Zhang et al 11 and also as demonstrating how machine learning models trained on one FA data set may be made readily applicable to other data sets. We are aware of a similar report of machine learning used in cooked egg allergy 30 ; that study is notable as an analysis of a clinical retrospective cooked egg challenge cohort, although the predictive power of the machine learning model was relatively low, perhaps partly because of the total number of OFCs analyzed (n = 67).…”

Section: Discussionsupporting

confidence: 65%

“…In Zhang et al, 11 a machine learning methodology was developed to predict OFCs for patients with suspected peanut, egg, or milk allergies. Multiple machine learning models and cross-validation strategies were evaluated, with the best-performing predictive models constructed by using an ensemble of models obtained via the LUCCK algorithm.…”

Section: Methodsmentioning

confidence: 99%

“…No separate model was made for the consumption-only arm owing to the low percentage of failed OFCs (3%) in that arm. Unlike in our previous model, 11 no additional feature selection beyond those in Table I was used. The data set was shuffled and split participant-wise into a 5-fold cross-validation data set (83%) and test data set (16%) and then stratified to preserve the pass-to-fail ratio of approximately 8:1, after which the predictive models were constructed.…”

Section: Methodsmentioning

confidence: 99%

“…In prior work by members of our research group, 11 a machine learning methodology was developed on a retrospective data set from the University of Michigan (UM) containing features similar to those in prior work. 9 The work tested multiple machine learning models, such as Learning Using Concave and Convex Kernels (LUCCK) 12 and random forest, along with multiple feature selection and cross-validation strategies to create predictive models for peanut, egg, and milk challenges.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Prediction of pediatric peanut oral food challenge outcomes using machine learning

Gryak,

Georgievska,

Zhang

et al. 2024

Journal of Allergy and Clinical Immunology: Global

View full text Add to dashboard Cite

Section: Discussionsupporting

confidence: 93%

Section: Discussionsupporting

confidence: 65%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Prediction of pediatric peanut oral food challenge outcomes using machine learning

Gryak,

Georgievska,

Zhang

et al. 2024

Journal of Allergy and Clinical Immunology: Global

View full text Add to dashboard Cite

Artificial intelligence in pediatric allergy research

Lisik,

Basna,

Dinh

et al. 2024

Eur J Pediatr

View full text Add to dashboard Cite

Atopic dermatitis, food allergy, allergic rhinitis, and asthma are among the most common diseases in childhood. They are heterogeneous diseases, can co-exist in their development, and manifest complex associations with other disorders and environmental and hereditary factors. Elucidating these intricacies by identifying clinically distinguishable groups and actionable risk factors will allow for better understanding of the diseases, which will enhance clinical management and benefit society and affected individuals and families. Artificial intelligence (AI) is a promising tool in this context, enabling discovery of meaningful patterns in complex data. Numerous studies within pediatric allergy have and continue to use AI, primarily to characterize disease endotypes/phenotypes and to develop models to predict future disease outcomes. However, most implementations have used relatively simplistic data from one source, such as questionnaires. In addition, methodological approaches and reporting are lacking. This review provides a practical hands-on guide for conducting AI-based studies in pediatric allergy, including (1) an introduction to essential AI concepts and techniques, (2) a blueprint for structuring analysis pipelines (from selection of variables to interpretation of results), and (3) an overview of common pitfalls and remedies. Furthermore, the state-of-the art in the implementation of AI in pediatric allergy research, as well as implications and future perspectives are discussed. Conclusion: AI-based solutions will undoubtedly transform pediatric allergy research, as showcased by promising findings and innovative technical solutions, but to fully harness the potential, methodologically robust implementation of more advanced techniques on richer data will be needed. What is Known:• Pediatric allergies are heterogeneous and common, inflicting substantial morbidity and societal costs.• The field of artificial intelligence is undergoing rapid development, with increasing implementation in various fields of medicine and research. What is New:• Promising applications of AI in pediatric allergy have been reported, but implementation largely lags behind other fields, particularly in regard to use of advanced algorithms and non-tabular data. Furthermore, lacking reporting on computational approaches hampers evidence synthesis and critical appraisal.• Multi-center collaborations with multi-omics and rich unstructured data as well as utilization of deep learning algorithms are lacking and will likely provide the most impactful discoveries.

show abstract