University of Twente Student Theses
Diagnosing spondyloarthritis by means of an exhaled breath analysis
Vrielink, Manouk (2024) Diagnosing spondyloarthritis by means of an exhaled breath analysis.
|
PDF
1MB |
| Abstract: | Introduction: Early diagnosis and treatment of spondyloarthritis (SpA) is crucial for preventing structural damage and stiffness often encountered in advanced stages of the disease. Currently, diagnosis relies on identifying typical SpA features along with laboratory findings and imaging techniques. Nonetheless, diagnostic challenges exist due to the absence of diagnostic criteria and the heterogeneous presentation of SpA. Exhaled breath analysis of volatile organic compounds (VOCs) through an electronic nose offers a promising opportunity for early, non-invasive diagnosis. This technique has already demonstrated effectiveness in identifying various conditions. This study aimed to investigate the ability of the electronic nose to differentiate between healthy controls and SpA patients by means of an exhaled breath analysis. Methods: Data were collected between December 2021 and April 2024. All participants performed a breath test with the aeoNose and provided relevant demographic and medical information. Random forest machine learning with 10-fold cross-validation was employed to analyze the breath samples and create an optimal discriminating algorithm. The performance of this model was compared to that of a multivariate logistic regression model, which incorporated readily available clinical variables in addition to the aeoNose predictive values. Results: Breath samples of 59 SpA patients (mean±SD age 50.3±15.8 years, 61% male) were compared with 180 healthy controls (mean±SD age 53.3±17.6 years, 42.8% male). The model that only included the aeoNose classification value resulted in an area under the receiver operator curve (AUC-ROC) of 0.95 (95% CI 0.92-0.99), a sensitivity of 95%, and a negative predictive value (NPV) of 98%. The second model, including solely the clinical variables age and gender, resulted in an AUC-ROC of 0.63 (95% CI 0.54-0.71), a sensitivity of 95%, and an NPV of 85%. Combining the clinical variables with the aeoNose classification value in a multivariate logistic regression model only slightly improved overall performance with an AUC-ROC of 0.96 (95% CI 0.93-0.99), while sensitivity remained 95% and NPV 98%. Conclusion: The aeoNose shows promise in discriminating between SpA patients and controls with high diagnostic accuracy, indicating its potential use as a diagnostic tool. However, cross-validation of the algorithm in independent samples is necessary. Adding readily available clinical variables in a multivariate logistic regression model only slightly enhances performance. |
| Item Type: | Essay (Master) |
| Clients: | Medisch Spectrum Twente, Enschede, the Netherlands |
| Faculty: | TNW: Science and Technology |
| Subject: | 01 general works, 42 biology, 44 medicine |
| Programme: | Health Sciences MSc (66851) |
| Link to this item: | https://purl.utwente.nl/essays/100588 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page