AI Usage in Asthma, Allergic Rhinitis, Food Allergy Management

A recent study highlights how artificial intelligence (AI) improves the prediction, diagnosis, treatment, and management of allergic diseases, including asthma, atopic dermatitis, food allergy, allergic rhinitis, and urticaria.1

AI can analyze large volumes of textual, visual, and auditory data through techniques like Reinforcement Learning, Machine Learning, Deep Learning, and Natural Language Processing. Many studies support AI’s effectiveness in aiding clinicians across various stages of allergy care.2

Investigators, led by Hong Tan, from the department of pediatrics at Xijing Hospital, the Fourth Military Medical University, in China, conducted a study to summarize AI usage in allergic diseases and analyze the advantages and limitations of intelligent assistance methods.1

Asthma

To combat the difficulty of collecting cough sounds, investigators developed an AI model that can analyze cough and breath sounds via smartphone apps, wearable sensors, and smart masks with high accuracy. Research shows that AI can distinguish between normal and abnormal breathing sounds.

One study used a Time Delay Neural Network to analyze cough characteristics; another used a 4-channel system to collect the breath sounds from 4 different back locations to identify asthma. A combined channel classified asthma better than a single one.

AI can also extract asthma-related features from electronic health records; one model reviewed records of 13,498 patients over 11 years and identified 5 clinically important phenotypes.

In another study, a Deep Learning sensor placed at 9 lung auscultation sites identified wheeze and respiration counts with 95% accuracy, 96% sensitivity, and 93% specificity.

One study showed AI outperformed doctors in interpreting lung function tests —only 44.6% of 120 physicians judged results accurately, compared with AI’s perfect score.

AI also helps personalize treatment, optimize corticosteroid use, predict response to therapies, and assess relapse risk. Investigators noted that, despite these benefits, AI relies on single-time physiological data, skilled operators, high-quality data, and continuous monitoring.

Atopic Dermatitis

AI models can predict atopic dermatitis (AD) onset using maternal, genetic, and microbiome data. One study found maternal alcohol consumption before pregnancy and depressive symptoms during pregnancy were linked to an increased eczema and rhinitis risk.

Deep learning models can assess AD severity and skin features more efficiently than traditional scoring methods. Gender, age, exercise, and smoking status were associated with AD severity.

Machine learning predicts responses to dupilumab and helps identify new treatments, such as caffeoylmalic acid targeting TNF-α and IL-4.

Wrist activity trackers and AI chatbots (i.e., ChatGPT) support real-time symptom monitoring and provide reliable patient education. A study showed users of a mobile app had significantly reduced Patient Oriented Eczema Measure scores (P < .001).

Investigators noted that AI models may underperform in diverse populations, particularly for those with darker skin tones.

Food Allergies

AI predicts food allergy risk, oral food challenge outcomes, and identifies patients with food allergy using electronic health records and gut microecology. A study found the LSTM model outperformed others like the Hidden Markov Model, Multi-Layer Perceptron Neural Network, Support Vector Machine, Random Forest, and LASSO regression.

Investigators developed AllerTOP, a server predicting allergens based on amino acid sequences, with 94% sensitivity.

Research has shown that AI detects tiny traces of food allergens, such as wheat, using tools like Fourier transform infrared spectroscopy and a smartphone convolutional neural network with 99.1% accuracy. AI can also help create dietary plans.

Models have also identified common therapeutic targets for food allergy, including IL4R, IL5, JAK1, JAK2, JAK3, and NR3C1.

Allergic Rhinitis

AI can predict allergic rhinitis using pollution data (NO₂, PM10); the Random Forest model had an AUC of 0.84. Deep learning also forecasts daily allergic rhinitis cases from pollution trends.

Models also assist in allergic rhinitis diagnoses using clinical data and detect nasal polyps with 98.3% accuracy. AI can also predict suicide risk in teens with allergic rhinitis with 83.3% sensitivity.

Urticaria

AI assists urticaria severity assessments using tools such as Legit.Health-UAS-HiveNet for automated UAS7 scoring. Image-based models improve evaluation in dark skin tones.

Additionally, deep learning improves Skin Prick Test analysis via smartphone images. Ultimately, AI improves personalized diagnoses and cuts time and cost, but it also requires high-quality image devices, which may pose financial issues.

References

1. Tan H, Zhou X, Wu H, Wang M, Zhou H, Qin Y, Zhang Y, Li Q, Luo J, Su H, Sun X. Application and research progress of artificial intelligence in allergic diseases. Int J Med Sci. 2025 Apr 9;22(9):2088-2102. doi: 10.7150/ijms.105422. PMID: 40303497; PMCID: PMC12035833.

2. MacMath D, Chen M, Khoury P. Artificial Intelligence: Exploring the Future of Innovation in Allergy Immunology. Curr Allergy Asthma Rep. 2023 Jun;23(6):351-362. doi: 10.1007/s11882-023-01084-z. Epub 2023 May 9. PMID: 37160554; PMCID: PMC10169188.