Biometeorological Eczema Prediction: A Synthetic Environmental Stressors into a Pipeline with Machine Learning

Abstract

The erratic course of Atopic Dermatitis (AD) is a major dilemma to preventive dermatology especially with regard to external fluctuations, which provoke exogenous flare-ups. Whereas the current diagnostic instruments are aimed at the severity at a single point in time, there is no combination of systems that can be used to model the longitudinal progression. In this work, an end-to-end machine learning model is suggested to forecast the changes in eczema severity (Mild, Moderate, Severe) through a synthesis of clinical patient data and the biometeorological variable. The paper provides a methodological framework on how synthetic \"digital twins\"[11] can be used to test predictive architectures prior to clinical application with web framework.

Introduction

This study presents a machine learning-based system for predicting eczema severity (mild, moderate, or severe) using a simulated dataset that includes demographic, clinical, environmental, and lifestyle factors.

The model is designed to support personalized eczema management by identifying how different factors contribute to disease progression. Traditional statistical methods are considered insufficient because they cannot effectively capture the complex, nonlinear relationships involved in eczema development.

The methodology involves generating a synthetic dataset, training a machine learning pipeline, and deploying the model through a web-based application for real-time predictions. The system is structured as an end-to-end workflow covering data processing, training, evaluation, and deployment.

Results show good performance, with an accuracy of 85.5%, along with strong precision, recall, and F1-scores. The model performs better than traditional approaches, especially in identifying severe cases due to balancing techniques.

However, a key limitation is the use of simulated (synthetic) data, which may not fully represent real-world patient variability and rare clinical cases. Future work focuses on validating the system using real clinical data while ensuring compliance with privacy regulations like GDPR and HIPAA.

Overall, the study demonstrates that machine learning combined with web platforms can improve eczema severity prediction and support clinical decision-making.

Conclusion

The proposed AI-driven model is highly likely to predict the development of eczema using a multidimensional patient data. It was deployed with the help of the Flask based web application, which explained its implementation as a strong clinical decision-support model. The model is based on the multifactorial nature of eczema and allows medical workers to detect those patients against which there is an increased risk of developing an acute disease and take the preventive measures timely and design personal treatment programs. Although the current study utilizes the synthetical data, interface of the web application is more intuitive, and less complicated framework that allows clinicians to input patient data and receive real-time predictions may help narrow the gap between extremely cumbersome machine learning models and applicability of AI in clinical settings to help improve patient outcomes even more.

References

[1] Weidinger, S., Beck, L. A., Bieber, T., Kabashima, K., & Irvine, A. D., “Atopic dermatitis,” Nature Reviews Disease Primers, vol. 4, Article no. 1, pp. 1–20, 2018. [2] Silverberg, J. I., “Comorbidities and the impact of atopic dermatitis,” Annals of Allergy, Asthma & Immunology, vol. 120, no. 2, pp. 144–151, 2018. [3] Pedregosa, F. et al., “Scikit-learn: Machine learning in Python,” Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011. [4] Chen, T. & Guestrin, C., “XGBoost: A scalable tree boosting system,” Proceedings of the 22nd ACM SIGKDD Conference, pp. 785–794, 2016. [5] Obermeyer, Z. & Emanuel, E.J., “Predicting the future — Big data, machine learning, and clinical medicine,” New England Journal of Medicine, vol. 375, no. 13, pp. 1216–1219, 2016. [6] Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P., “SMOTE: Synthetic Minority Over-sampling Technique,” Journal of Artificial Intelligence Research, vol. 16, pp. 321–357, 2002. [7] Beam, A.L. & Kohane, I.S., “Big data and machine learning in health care,” JAMA, vol. 319, no. 13, pp. 1317–1318, 2018. [8] Goldstein, B.A. et al., “Opportunities and challenges in developing risk prediction models with electronic health records data,” Journal of Biomedical Informatics, vol. 76, pp. 168–177, 2017 [9] Esteva, A. et al., “A guide to deep learning in healthcare,” Nature Medicine, vol. 25, pp. 24–29, 2019 [10] Rajkomar, A., Dean, J., & Kohane, I., “Machine learning in medicine,” New England Journal of Medicine, vol. 380, no. 14, pp. 1347–1358, 2019 [11] Engebretsen, K. A., Johansen, J. D., Kezic, S., Linneberg, A., & Thyssen, J. P. (2016). The effect of environmental humidity and temperature on skin barrier function and dermatitis. Journal of the European Academy of Dermatology and Venereology, 30(2), 223-249.Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specification, IEEE Std. 802.11, 1997.

Copyright

Copyright © 2026 Suparna Das, Sourabh Eklare. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.