Abstract
Since 2019, the coronavirus disease-19 (COVID-19) has been spreading rapidly worldwide, posing anunignorable threat to the global economy and human health. It is a disease caused by severe acute respiratorysyndrome coronavirus 2, a single-stranded RNA virus of the genus Betacoronavirus. This virus is highly infectiousand relies on its angiotensin-converting enzyme 2-receptor to enter cells. With the increase in the number ofconfirmed COVID-19 diagnoses, the difficulty of diagnosis due to the lack of global healthcare resources becomesincreasingly apparent. Deep learning-based computer-aided diagnosis models with high generalisability can effectivelyalleviate this pressure. Hyperparameter tuning is essential in training such models and significantly impacts theirfinalperformance and training speed. However, traditional hyperparameter tuning methods are usually time-consumingand unstable. To solve this issue, we introduce Particle Swarm Optimisation to build a PSO-guided Self-TuningConvolution Neural Network (PSTCNN), allowing the model to tune hyperparameters automatically. Therefore, theproposed approach can reduce human involvement. Also, the optimisation algorithm can select the combination ofhyperparameters in a targeted manner, thus stably achieving a solution closer to the global optimum. Experimentally,the PSTCNN can obtain quite excellent results, with a sensitivity of 93.65% +/- 1.86%, a specificity of 94.32% +/- 2.07%,a precision of 94.30% +/- 2.04%, an accuracy of 93.99% +/- 1.78%, an F1-score of 93.97% +/- 1.78%, Matthews CorrelationCoefficient of 87.99% +/- 3.56%, and Fowlkes-Mallows Index of 93.97% +/- 1.78%. Our experiments demonstrate thatcompared to traditional methods, hyperparameter tuning of the model using an optimisation algorithm is faster andmore effective