EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Machine learning integrating immune-related cells, immunoglobulins, and immuno-inflammatory markers identifies risk factors for pulmonary comorbidities in ankylosing spondylitis: stratified analyses by chronic obstructive pulmonary disease and pulmonary nodules

Shuang Ji, Jiayun Ping, Long Cheng, Tianqi Zhang, Zhangwei Huang, Lianzi Wang, Chao Cheng, Xu Zhang

Frontiers in Immunology · 19.08.2026 · 17

DOI: 10.3389/fimmu.2026.1855064 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Immunology, online erschienen am 19.08.2026. Verfasst von Shuang Ji und 7 weiteren Autor:innen. Beteiligte Einrichtung: Department of Respiratory and Critical Care Medicine, First Affiliated Hospital of Anhui Medical University, Hefei, China. Hauptthemen laut MEDLINE: Pulmonary Disease, Chronic Obstructive, Spondylitis, Ankylosing, Machine Learning, Solitary Pulmonary Nodule. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

Ankylosing spondylitis (AS) and pulmonary diseases (PD) comorbidities accelerates functional decline and increases disease burden, but risk stratification and identification of these comorbidities are lacking. This study developed machine learning (ML) models based on immune-related cells and immuno-inflammatory markers to identify risk factors for pulmonary comorbidities in AS.

Methods

Demographic characteristics, immuno-inflammatory indices (e.g., neutrophils, erythrocyte sedimentation rate [ESR], C-reactive protein [CRP], immunoglobulin G, neutrophil-to-lymphocyte Ratio [NLR], platelet-to-neutrophil Ratio [PNR], pan-immune-inflammation value [PIV]) were measured and calculated for each patient. Feature selection was performed via least absolute shrinkage and selection operator (LASSO). Four ML algorithms, including decision tree, random forest, XGBoost, and support vector Machine (SVM), were developed and evaluated to distinguish between AS and AS+PD. Subgroup analyses were conducted for pulmonary nodules (PN) and chronic obstructive pulmonary disease (COPD), and sensitivity analysis was performed using unimputed data.

Results

363 AS patients were enrolled, comprising 193 AS alone and 170 AS+PD patients (including 123 with PN and 47 with COPD). LASSO regression identified age, neutrophils, IgG, ESR, CRP, NLR, PNR, and PIV as key predictors for AS+PD. In the overall population, the SVM model demonstrated superior generalization and resistance to overfitting, whereas the random forest and XGBoost showed overfitting. In subgroup analyses, XGBoost emerged as the optimal classifier for AS+PN and AS+COPD, with good calibration and net clinical benefit. Sensitivity analyses confirmed the robustness and consistency of these findings.

Conclusion

AS patients with pulmonary comorbidities exhibit distinct immune-related cells and immuno-inflammatory markers (especially neutrophils, IgG, NLR, and PIV). Of the four ML models, SVM showed best overall generalization, whereas XGBoost performed excellently in PN and COPD subgroups. These findings support the integration of immuno-inflammatory markers with ML algorithms to improve identification of current status of pulmonary complications for pulmonary comorbidities in AS.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Pulmonary Disease, Chronic ObstructiveSpondylitis, AnkylosingMachine LearningSolitary Pulmonary NoduleHumansFemaleMaleBiomarkersRisk FactorsComorbidityMiddle AgedAdultRandom ForestBoosting Machine Learning AlgorithmsPredictive Learning ModelsClassification AlgorithmsSupport Vector Machine

Stichwörter der Autor:innen

ankylosing spondylitischronic obstructive pulmonary diseasecomorbiditymachine learningpulmonary diseasespulmonary nodules

Substanzen

Biomarkers

Autor:innen und Institutionen

  • Department of Respiratory and Critical Care Medicine, First Affiliated Hospital of Anhui Medical University, Hefei, China.
  • Department of Health Management Center, First Affiliated Hospital of Anhui Medical University, Hefei, China.
  • Department of Oncology, First Affiliated Hospital of Anhui Medical University, Hefei, China.
  • First Clinical College and First Affiliated Hospital of Anhui Medical University, Hefei, China.
  • Department of Clinical Laboratory, First Affiliated Hospital of Anhui Medical University, Hefei, China.
  • Department of Respiratory and Critical Care Medicine, Anhui Provincial Chest Hospital, Hefei, China.

Zitieren

APA

Ji, S., Ping, J., Cheng, L., Zhang, T., Huang, Z., Wang, L., Cheng, C., & Zhang, X. (2026). Machine learning integrating immune-related cells, immunoglobulins, and immuno-inflammatory markers identifies risk factors for pulmonary comorbidities in ankylosing spondylitis: stratified analyses by chronic obstructive pulmonary disease and pulmonary nodules. Frontiers in Immunology, 17. https://doi.org/10.3389/fimmu.2026.1855064

Vancouver

Ji S, Ping J, Cheng L, Zhang T, Huang Z, Wang L, et al. Machine learning integrating immune-related cells, immunoglobulins, and immuno-inflammatory markers identifies risk factors for pulmonary comorbidities in ankylosing spondylitis: stratified analyses by chronic obstructive pulmonary disease and pulmonary nodules. Frontiers in Immunology. 2026;17. doi:10.3389/fimmu.2026.1855064

BibTeX
@article{Ji2026,
  author = {Ji, Shuang and Ping, Jiayun and Cheng, Long and Zhang, Tianqi and Huang, Zhangwei and Wang, Lianzi and Cheng, Chao and Zhang, Xu},
  title = {Machine learning integrating immune-related cells, immunoglobulins, and immuno-inflammatory markers identifies risk factors for pulmonary comorbidities in ankylosing spondylitis: stratified analyses by chronic obstructive pulmonary disease and pulmonary nodules},
  journal = {Frontiers in Immunology},
  year = {2026},
  volume = {17},
  doi = {10.3389/fimmu.2026.1855064}
}
Kostenlos starten →