Table 8 State-of-the-Art comparison.
From: LDSC: enhancing lung disease diagnosis using a simple 1D-CNN
Method | Year | Model Type | ICBHI Acc (%) | KAUH Acc (%) | Advantages | Disadvantages |
|---|---|---|---|---|---|---|
CNN-LSTM15 | 2022 | Hybrid CNN-LSTM | 82.4–99.6 | 99.4–99.8 | High accuracy with augmentation | > 5 layers, spectrogram, high compute |
EasyNet22 | 2024 | Deep CNN | 97.5 | 98.2 | Strong feature learning | 6 + layers, spectrogram, slow inference |
CRNN2 | 2021 | CNN + RNN | ~ 73 (score) | – | Robust to noise | Complex, overfitting risk, spectrogram |
LDSC (Ours) | 2025 | 1D-CNN (2 layers) | 98 | 99 | Minimal, fast, real-time, no spectrogram | Needs larger diverse data |
