Abstract
The early diagnosis of esophageal cancer (EC) is extremely challenging due to a lack of effective diagnostic methods. The study presented herein aims to assess whether serum volatile organic compounds (VOCs) could be utilised as emerging diagnostic biomarkers for EC. Gas chromatography-ion mobility spectrometry (GC-IMS) was used to detect VOCs in the serum samples of 55 patients with EC, with samples from 84 healthy controls (HCs) patients analysed as a comparison. All machine learning analyses were based on data from serum VOCs obtained by GC-IMS. A total of 33 substance peak heights were detected in all patient serum samples. The ROC analysis revealed that four machine learning models were effective in facilitating the diagnosis of EC. In addition, the random forests model for 5 VOCs had an AUC of 0.951, with sensitivities and specificities of 94.1 and 96.0%, respectively.