Table 7 Performance criteria with SG feature
From: Performance evaluation of lung sounds classification using deep learning under variable parameters
Parameter combination | Performance criteria (%) | Stage | ||||
|---|---|---|---|---|---|---|
Sensitivity (CI) | Specificity (CI) | Accuracy (CI) | Precision (CI) | F1_score (CI) | ||
Lwin = 64 OP = 25% | 62.3 (60.1–65.1) | 64.5 (61.8–66.7) | 62.9 (60.6–65.4) | 82.4 (79.1–89.4) | 71.0 (68.3–75.4) | Training |
56.6 (54.5–58.8) | 50.5 (48.7–53.1) | 55.0 (52.9–57.5) | 76.3 (73.6–80.9) | 65.0 (62.6–68.1) | Test | |
Lwin = 64 OP = 50% | 68.9 (65.9–70.8) | 70.5 (68.8–72.6) | 69.3 (66.9–71.2) | 87.5 (80.1–90.0) | 77.1 (72.3–79.3) | Training |
58.7 (56.1–60.9) | 55.5 (53.2–58.1) | 57.8 (55.3–60.2) | 77.1 (75.9–81.3) | 66.7 (64.5–69.7) | Test | |
Lwin = 64 OP = 75% | 76.5 (74.1–78.1) | 75.5 (73.5–79.6) | 76.2 (73.9–78.6) | 87.9 (84.8–88.4) | 81.8 (79.1–83.0) | Training |
59.2 (56.9–62.2) | 56.0 (53.2–59.2) | 58.3 (55.8–61.3) | 77.5 (74.2–78.1) | 67.1 (64.4–69.2) | Test | |
Lwin = 128 OP = 75% | 88.3 (86.1–91.1) | 89.5 (87.1–92.3) | 88.7 (86.5–91.2) | 94.4 (90.9–99.2) | 91.2 (88.4–95.0) | Training |
72.6 (69.8–73.9) | 68.4 (66.2–71.4) | 71.5 (68.6–73.3) | 86.6 (80.5–89.1) | 79.0 (74.8–80.8) | Test | |
Lwin = 256 OP = 75% | 88.8 (86.2–91.2) | 87.5 (85.3–89.6) | 88.4 (85.9–90.9) | 94.1 (92.1–97.4) | 91.4 (89.1–94.2) | Training |
79.7 (77.3–82.7) | 70.5 (68.1–72.5) | 76.6 (74.1–79.4) | 84.2 (82.0–86.3) | 81.9 (79.6–84.5) | Test | |