Table 3.
Summary of various algorithms applied to lung cancer diagnosis using multiple VOCs.
| Algorithms | Analytical Platform | Patients with Cancer No. | Analyzed VOC No. | Sensitivity % | Specificity % | AUC | Reference/(Year) |
|---|---|---|---|---|---|---|---|
| Stepwise Discriminant Analysis | GC-MS | 67 | 9 | 85.1 | 80.5 | NR | [35]/(2003) |
| Logistic Regression | GC-MS | 193 | 16 | 84.6 | 80.0 | 0.88 | [50]/(2007) |
| Weighted Digital Sum Discriminator | GC-MS | 193 | 30 | 84.5 | 81 | 0.9 | [32]/(2008) |
| Support Vector Machine | GS-MS | 107 | 5 | 95 | 89 | NR * | [51]/(2016) |
| Artificial Neural Networks | GC-MS | 108 | 88 | 86.36 | 86.36 | 0.86 | [52]/(2019) |
| K-nearest Neighbor | GC-MS | 325 | NR | NR | NR | 0.63 † | [53]/(2020) |
| Extreme Gradient Boosting | SIFT-MS | 148 | 116 | 82 | 94 | 0.95 | This WorkConsidering only participants’ VOCs |
| 96 | 88 | 0.98 | Considering both participants’ VOCs and environmental VOCs |
Abbreviations: AUC, area under the curve; GC-MS, gas chromatography-mass spectrometry; NR, not reported; SIFT-MS, selected ion flow tube mass spectrometry; * Accuracy: 89%, † Classify adenocarcinoma and squamous cell carcinoma patients.