The methods comprises: a) computing from audio data uttered by at least one first mathematical audio transformation function to obtain a first time series representation of said audio data b) computing at least one second mathematical audio transformation function, equal or different to the first mathematical function, over said obtained first time series representation to obtain a second time series representation of the audio data c) segmenting said first and second time series representations into one or more timestamps d) computing one or more different measures by using the one or more timestamps and the first and second time series representations and e) using the one or more different computed measures to train a machine learning algorithm which computes a score for assessing said disease.
