Harnessing Machine Learning and Electronic Health Record Data to Improve Asthma Management
Clinical Snapshot
PICO Framework
| P — Population | Patients with asthma across clinical settings where electronic health record (EHR) data and digital health tools are deployed |
| I — Intervention | Machine learning (ML) models and large language models (LLMs) applied to EHR data for asthma management — including prediction of medication response, exacerbation risk, inhaler adherence monitoring, and patient education chatbots |
| C — Comparator | Standard clinical care without ML/LLM-assisted decision support; no formal comparator explicitly defined across included literature |
| O — Outcomes | Clinically relevant applications including: prediction of response to inhaled corticosteroids and biologics, exacerbation risk prediction accuracy, inhaler adherence improvement via digital inhaler systems, and patient education effectiveness via LLM-based tools |
Bottom Line
This narrative review from Brigham and Women's Hospital provides a clinically oriented overview of machine learning and large language model applications in asthma management, with a focus on EHR-derived data. It identifies four promising domains: predicting response to inhaled corticosteroids and biologics, improving inhaler adherence via digital systems, stratifying exacerbation risk, and supporting patient education through chatbot interfaces. However, the review is methodologically limited — it lacks a systematic search strategy, pre-specified inclusion criteria, formal risk of bias assessment, and quantitative synthesis. Accuracy claims of 'moderate-to-high' for exacerbation prediction are not supported by pooled metrics. The authors appropriately caution that external validation, equitable implementation, and clinician oversight are prerequisites for clinical translation. For Australian clinicians, this review signals an emerging landscape rather than practice-ready guidance. No ML-assisted asthma management tools are currently TGA-approved or RACGP-endorsed. Clinicians should monitor this space critically, particularly regarding biologic patient selection and digital inhaler adherence support, while remaining alert to algorithmic bias risks in diverse patient populations.
Key Findings
P Value: Not reported at the review level
Effect Size: Not quantitatively synthesised; exacerbation risk prediction described as achieving 'moderate-to-high accuracy' across reviewed models; no pooled effect size reported
Primary Outcome: Identification and narrative synthesis of ML and LLM applications in asthma management using EHR data, across four domains: medication response prediction (ICS and biologics), digital inhaler adherence monitoring, exacerbation risk prediction, and LLM-based patient education
Nnt Or Sensitivity: No NNT, sensitivity, specificity, AUC, or other diagnostic/prognostic performance metrics are pooled or reported at the review level; individual study metrics are referenced narratively without synthesis
Confidence Interval: Not reported; no meta-analytic synthesis performed
Clinical Application
Clinical implementation of ML-assisted asthma management remains largely in the research and pilot phase. Barriers include the need for external validation across diverse populations, integration with existing EHR workflows, clinician training, regulatory approval of clinical decision support tools, and equitable access. Digital inhaler systems are the most commercially mature application. LLM-based patient education tools are accessible but require careful governance to mitigate hallucination risks and health literacy mismatches. In Australia, ML-assisted asthma management tools are not currently endorsed by the RACGP, the National Asthma Council Australia, or the Thoracic Society of Australia and New Zealand (TSANZ). The Australian Digital Health Agency's national digital health strategy supports EHR integration and clinical decision support development, which provides a policy framework for future ML tool adoption. PBS-listed biologics for severe asthma (mepolizumab, benralizumab, dupilumab, tezepelumab) represent a domain where ML-assisted patient selection could have significant cost-effectiveness implications, but no TGA-approved ML decision support tools for biologic prescribing currently exist. Digital inhaler systems (e.g., Propeller Health, Adherium) have been evaluated in Australian pilot studies but are not universally available or PBS-subsidised. Health equity considerations are particularly salient in Australia given the disproportionate asthma burden in Aboriginal and Torres Strait Islander communities, where algorithmic bias risks must be proactively addressed before any ML tool deployment. Adult and paediatric patients with asthma managed in settings with established EHR infrastructure and access to digital health technologies; most directly applicable to tertiary and academic medical centre contexts where ML tool development and validation are feasible
Abstract
PURPOSE OF REVIEW: The review examines the application of machine learning (ML) and large language models (LLMs) to asthma management. We sought to identify clinically relevant applications, and particularly those that harness electronic health record data. We review methodological challenges and future directions for translating these tools into meaningful improvements in asthma care. RECENT FINDINGS: ML applied to electronic health record data has been utilized across several domains of asthma management: predicting medication response to inhaled corticosteroids and biologics, improving inhaler adherence through digital inhaler systems, and predicting exacerbation risk with moderate-to-high accuracy. Tools for patient education include clinician-guided chatbots, as well as publicly available LLMs. Limitations include accuracy, hallucinations, and patient health literacy. ML and LLMs offer promising pathways towards personalized, data-driven asthma management. However, harnessing this potential will require rigorous external validation, transparent model design, equitable implementation, and adaptive clinician oversight. Collaboration among clinicians, data scientists, and policymakers will be essential to implement these tools for patient-centered asthma care.
References
- 1.Olayiwola, O., & Foer, D. (2026). Harnessing machine learning and electronic health record data to improve asthma management. Current Allergy and Asthma Reports. https://doi.org/10.1016/j.jaip.2023.05.042
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service