Breath of Health: spectroscopy-based breath test for the detection of SARS-CoV-2

Figures & data

Figure 1. (A) Thermo Fisher Fourier-transform infra-red spectroscopy (FTIR) device showing an illustration of result spectra for an exhaled breath sample (B) Tedlar bag for breath sampling (C) The Breath of Health Merkava 1.1 – Exhaled breath analysis system.

Table 1. Subgroup analyses of BOH breath test.

Group	N	BOH positive	PCR positive	True positive	True negative	False positive	False negative	Sensitivity (95% CI)	Specificity (95% CI)	Positive predictive value (95% CI)	Negative predictive value (95% CI)	Accuracy (95% CI)
Total cohort	531	182	171	136	314	46	35	79.5% (72.7–85.3%)	87.2% (83.3–90.5%)	74.7% (69.1–79.6%)	90.0% (87.0–92.4%)	84.8% (81.4–87.7%)
Patients with respiratory symptoms	180/523	61	75	52	96	9	23	69.3% (57.6–79.5%)	91.4% (84.4–96.0%)	85.3% (73.8–93.0%)	80.7% (72.4–87.3%)	82.2% (75.8–87.5%)
Patients without respiratory symptoms	343/523	117	93	81	214	36	12	87.1% (78.6–93.2%)	85.6% (80.6–89.7%)	69.2% (60.0–77.4%)	94.7% (90.9–97.2%)	86.0% (81.9–89.5%)
Patients with Diabetes mellitus	106/502	33	33	25	68	8	8	75.8% (57.7–88.9%)	89.5% (80.3–95.3%)	75.8% (57.7–88.9%)	89.5% (80.3–95.3%)	85.3% (77.3–91.4%)
Patients without Diabetes mellitus	396/502	132	118	96	242	36	22	81.4% (73.1–87.9%)	87.1% (82.5–90.7%)	72.7% (64.2–80.1%)	91.7% (87.7–94.7%)	85.4% (81.5–88.7%)
Patients with Asthma	39/504	9	10	6	26	3	4	60.0% (26.2–87.8%)	89.7% (72.7–97.8%)	66.7% (29.9–92.5%)	86.7% (69.3–96.2%)	82.1% (66.5–92.4%)
Patients without Asthma	465/504	155	161	115	284	40	26	81.6% (74.2–87.6%)	87.7% (83.6–91.0%)	74.2% (66.6–80.9%)	91.6% (88.0–94.5%)	85.8% (82.3–88.9%)
Smokers	111/504	26	21	17	81	9	4	81.0% (58.1–94.6%)	90.0% (81.9–95.3%)	65.4% (44.3–82.8%)	95.3% (88.4–98.7%)	88.3% (80.8–93.6%)
Non smokers	393/504	138	130	104	229	34	26	80.0% (72.1–86.5%)	87.1% (82.4–90.9%)	75.4% (67.3–82.3%)	89.8% (85.4–93.2%)	84.7% (80.8–88.1%)

Figure 2. Receiver operating characteristic (ROC) curve for the diagnostic accuracy of BOH test and positive predictive value (PPV) and negative predictive value (NPV) analysis. [A] The ROC curve illustrates the diagnostic performance of BOH test in terms of sensitivity, specificity, and the area under the curve (AUC). The analysis demonstrates that BOH test achieves a sensitivity of 79.5% and specificity of 87.2%, with an AUC of 0.834. [B] The blue curve represents the PPV with confidence intervals (CI, dotted lines), and the green line shows the NPV with CI (dotted lines) of the BOH test according to disease prevalence. The calculated PPV for the BOH test is 74%, with a CI of 69-79%. the NPV for the BOH test is 90%, with a CI of 87-92%.

Figure 3. Subgroup analysis of Breath of Health (BOH) test compared to the polymerase chain reaction (PCR, the gold standard) stratified according to the presence or absence of respiratory symptoms. Each column represents a sub set of study participants stratified according to BOH test results (left positive results, right negative results) and the presence (the first and third columns to the left) or absence (second and fourth columns to the left) of symptoms. PCR results are presented as white (negative) or black (positive) colour. The colour composition of each column represents the percentage of a certain PCR result in that specific sub-set of study population. For example, 94.7% of samples obtained from participants without respiratory symptoms and negative BOH result had a negative PCR result while the percentage of participants with negative BOH result but had respiratory symptoms, was significantly lower (80.7%). * P-value <0.05, ***P-value <0.0001.