# Dermatological Manifestations of Smoking-Induced Oxidative Stress and Inflammation: A Multifaceted Analysis of Cutaneous Aging and Disease Progression

## TL;DR

Smoking is linked to various skin issues like psoriasis and aging, but machine learning models failed to accurately predict these outcomes.

## Contribution

The study highlights smoking's dermatological impact and evaluates machine learning's limitations in predicting skin conditions.

## Key findings

- Smoking is significantly associated with psoriasis, atopic dermatitis, aging, and skin cancer.
- Machine learning models showed poor predictive performance for psoriasis.
- CRP, IL-6, pack-years, and age were identified as key contributors to dermatological outcomes.

## Abstract

Background

Smoking is a leading preventable cause of morbidity and mortality, with significant yet underrecognized dermatological consequences. Tobacco-induced oxidative stress and inflammation contribute to cutaneous aging, chronic inflammatory dermatoses, and carcinogenesis.

Objective

This study aimed to investigate dermatological manifestations associated with smoking, focusing on clinical outcomes, laboratory biomarkers, and predictive modeling using machine learning.

Methods

A retrospective dataset of 350 patients was analyzed. Demographic, clinical, and laboratory variables were assessed using chi-square tests, analysis of variance (ANOVA), t-tests, and correlation analyses. Machine learning models (Random Forest, Logistic Regression, (XGBoost Developers, Distributed Machine Learning Community, Washington, DC), and LightGBM (Microsoft Corporation, Redmond, WA)) were applied to predict psoriasis presence; however, all demonstrated poor discriminatory performance (area under the curve (AUC) < 0.50). Despite limited accuracy, feature importance analysis highlighted CRP, IL-6, pack-years, and age as relevant contributors, underscoring the methodological challenges of prediction using a synthetic dataset.

Results

The cohort included 51.1% males and 48.9% females, with a mean age of 51.4 years. Current smokers comprised 46.9% of patients and had significantly higher pack-year exposure than former smokers (t = 2.53, p = 0.013). Smoking status was significantly associated with psoriasis prevalence (χ² = 21.38, p < 0.001), atopic dermatitis (χ² = 14.27, p < 0.001), advanced cutaneous aging (χ² = 12.61, p < 0.001), and skin cancer risk (χ² = 18.64, p < 0.001). No significant associations were observed with acne severity, treatment type, or comorbidities. Machine learning models showed poor predictive ability (AUC < 0.50), though feature importance identified CRP, IL-6, pack-years, and age as major contributors.

Conclusion

Smoking is significantly associated with multiple dermatological outcomes, emphasizing the need to integrate smoking history into dermatological evaluation and counseling.