COVID-19 Detection from Cough Recording by means of Explainable Deep Learning

The new coronavirus disease (COVID-19), declared a pandemic on 11 March 2020 by the World Health Organization, has caused over 6 million victims worldwide. Because of the rapid spread of the virus, with the aim to perform screening we exploit deep learning model to quickly diagnose altered respiratory conditions. In this paper, we propose a method to recognize and classify cough audio files into three classes to distinguish patients with COVID-19 disease, symptomatic ones and healthy subjects, with the use of a convolutional neural network (CNN). Cough audios were recorded by using a smartphone and its built-in microphone. From cough recordings, we generate spectrogram images and we obtain an accuracy equal to 0.82 with a deep learning network developed by authors. Our method also provides heatmaps, which show the relevant input areas used by the model for the final forecast, and this aspect ensures the explainability of the method. © 2022 IEEE.

Combining simulation and machine learning for the management of healthcare systems

The growing research trends in the field of artificial intelligence have largely impacted the healthcare sector. Thanks to the high predictive power of machine learning approaches, new tools to support the clinical decision-making can be designed. However, since the demand for healthcare services is complex and highly changing, as it is affected by external unpredictable factors such as the CoViD-19, the reliability and robustness of such predictive tools is highly dependent on their capability of varying and adapting the forecasting in accordance with variations in environmental factors and health needs. In this work, we propose a combined simulation and machine learning approach to study the robustness and adaptability of predictive tools for healthcare management. Discrete event simulation is employed to simulate a generic healthcare service. The patients' length of stay (LOS) is monitored as a performance indicator of the care process. Three machine learning algorithms have been tested to predict the LOS in different simulated scenarios obtained by varying the level of demand for the healthcare service. The predictability of the tested algorithms has been studied in terms of mean errors. Preliminary results suggest that abrupt changes in the healthcare demand have a negative impact on the performance of the machine learning algorithms, which are not prone to adapt decisions to the surrounding environment. The design of novel intelligent health system, which aim to integrate artificial intelligence tools in the clinical decision-making process, should take into account these limitations. In this sense the use of simulation can be beneficial in the assessment of the new generation of decision support systems in healthcare. © 2022 IEEE.