As an aid to predicting future hospital admissions, we compare use of the Multinomial Logit and the Utility Maximising Nested Logit models to describe how patients choose their hospitals. The models are fitted to real data from Derbyshire, United Kingdom, which lists the postcodes of more than 200,000 admissions to six different local hospitals. Both elective and emergency admissions are analysed for this mixed urban/rural area. For characteristics that may affect a patient's choice of hospital, we consider the distance of the patient from the hospital, the number of beds at the hospital and the number of car parking spaces available at the hospital, as well as several statistics publicly available on National Health Service (NHS) websites: an average waiting time, the patient survey score for ward cleanliness, the patient safety score and the inpatient survey score for overall care. The Multinomial Logit model is successfully fitted to the data. Results obtained with the Utility Maximising Nested Logit model show that nesting according to city or town may be invalid for these data; in other words, the choice of hospital does not appear to be preceded by choice of city. In all of the analysis carried out, distance appears to be one of the main influences on a patient's choice of hospital rather than statistics available on the Internet.
Partial discharge measurements taken online are severely corrupted by noise due to external disturbances. In this paper a powerful noise reduction technique, based on a wavelet packet denoising algorithm, is employed to isolate the signals from the noise. This methodology enables the denoising of partial discharges that are heavily corrupted by noise without assuming any a priori knowledge about the partial discharge features. A brief description of the wavelet packet theory as an extension of the multi-resolution analysis is given. Results of the application of this algorithm to simulated data of low signal-to-noise ratio are presented, demonstrating substantial improvement in signal recovery with minimum shape distortion. Finally, the capability of this technique is highlighted by applying it to experimental field data taken from three-phase 11 kV cables.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.