Elia El-Darzi

A simulation modelling approach to evaluating length of stay, occupancy, emptiness and bed blocking in a hospital geriatric department.

The flow of patients through geriatric hospitals has been previously described in terms of acute (short‐stay), rehabilitation (medium‐stay), and long‐stay states where the bed occupancy at a census point is modelled by a mixed exponential model using BOMPS (Bed Occupancy Modelling and Planning System). In this a patient is initially admitted to acute care. The majority of the patients are discharged within a few days into their own homes or through death. The rest are converted into medium‐stay patients where they could stay for a few months and thereafter either leave the system or move on to a long‐stay compartment where they could stay until they die. The model forecasts the average length of stay as well as the average number of patients in each state. The average length of stay in the acute compartment is artificially high if some would‐be long‐term patients are kept waiting in the short‐stay compartment until beds become available in long‐stay (residential and nursing homes). In this paper we consider the problem as a queueing system to assess the effect of blockage on the flow of patients in geriatric departments. What‐if analysis is used to allow a greater understanding of bed requirements and effective utilisation of resources.

A simulation study of the winter bed crisis

The winter bed crisis is a cyclical phenomenon which appears in British hospitals every year, two or three weeks after Christmas. The crisis is usually attributed to factors such as the bad weather, influenza, older people, geriatricians, lack of cash or nurse shortages. However, a possible alternative explanation could be that beds within the hospital are blocked because of lack of social services for discharge of hospital patients during the Christmas period. Adopting this explanation of why the bed crisis occurs, the problem was considered as a queuing system and discrete event simulation was employed to evaluate the model numerically. The model shows that stopping discharges of rehabilitating patients for 21 days accompanied by a cessation of planned patients for 14 days precipitate a bed crisis when the planned admissions recommence. The extensive “what-if” capabilities of such models could be proved to be crucial to the designing and implementation of possible solutions to the problem.

A Breast Cancer Diagnosis System: A Combined Approach Using Rough Sets and Probabilistic Neural Networks

In this paper, we present a medical decision support system based on a hybrid approach utilizing rough sets and a probabilistic neural network. We utilized the ability of rough sets to perform dimensionality reduction to eliminate redundant attributes from a biomedical dataset. We then utilized a probabilistic neural network to perform supervised classification. Our results indicate that rough sets were able to reduce the number of attributes in the dataset by 67% without sacrificing classification accuracy. Our classification accuracy results yielded results on the order of 93%

Graph theoretic relaxations of set covering and set partitioning problems

In this paper we review the graph theoretic relaxations of the set covering problem (SCP) and the set partitioning problem (SSP). These are: a network relaxation which can be solved by the greedy method, a matching relaxation and a graph covering relaxation. Other relaxations such as assignment, shortest route and minimum spanning tree are also presented.

Healthcare Data Mining: Prediction Inpatient Length of Stay

Data mining approaches have been widely applied in the field of healthcare. At the same time it is recognized that most healthcare datasets are full of missing values. In this paper we apply decision trees, Naive Bayesian classifiers and feature selection methods to a geriatric hospital dataset in order to predict inpatient length of stay, especially for the long stay patients

Support vector machine learning with an evolutionary engine

The paper presents a novel evolutionary technique constructed as an alternative of the standard support vector machines architecture. The approach adopts the learning strategy of the latter but aims to simplify and generalize its training, by offering a transparent substitute to the initial black-box. Contrary to the canonical technique, the evolutionary approach can at all times explicitly acquire the coefficients of the decision function, without any further constraints. Moreover, in order to converge, the evolutionary method does not require the positive (semi-)definition properties for kernels within nonlinear learning. Several potential structures, enhancements and additions are proposed, tested and confirmed using available benchmarking test problems. Computational results show the validity of the new approach in terms of runtime, prediction accuracy and flexibility.

An evolutionary computational approach to probabilistic neural network with application to hepatic cancer diagnosis

The performance of a probabilistic neural network is strongly influenced by the smoothing parameter. This paper introduces an evolutionary approach based on genetic algorithm to optimise the search of the smoothing parameter in a modified probabilistic neural network. A Java implementation is introduced and the computational results showed the viability of this hybrid approach to determine the optimum diagnosis for hepatic diseases.

Analysis of stopping criteria for the EM algorithm in the context of patient grouping according to length of stay

The expectation maximisation (EM) algorithm is an iterative maximum likelihood procedure often used for estimating the parameters of a mixture model. Theoretically, increases in the likelihood function are guaranteed as the algorithm iteratively improves upon previously derived parameter estimates. The algorithm is considered to converge when all parameter estimates become stable and no further improvements can be made to the likelihood value. However, to reduce computational time, it is often common practice for the algorithm to be stopped before complete convergence using heuristic approaches. In this paper, we consider various stopping criteria and evaluate their effect on fitting Gaussian mixture models (GMMs) to patient length of stay (LOS) data. Although the GMM can be successfully fitted to positively skewed data such as LOS, the fitting procedure often requires many iterations of the EM algorithm. To our knowledge, no previous study has evaluated the effect of different stopping criteria on fitting GMMs to skewed distributions. Hence, the aim of this paper is to evaluate the effect of various stopping criteria in order to select and justify their use within a patient spell classification methodology. Results illustrate that criteria based on the difference in the likelihood value and on the GMM parameters may not always be a good indicator for stopping the algorithm. In fact we show that the values of the difference in the variance parameters should be used instead, as these parameters are the last to stabilise. In addition, we also specify threshold values for the other stopping criteria.

Statistical Comparison of a Probabilistic Neural Network Approach in Hepatic Cancer Diagnosis

Probabilistic neural network (PNN) may provide an alternative to establish predictive algorithms for the cancer early diagnosis. We trained a PNN, using three different techniques for searching the smoothing parameter, with a database of 299 patients. This paper deals with the comparison of the prediction capabilities of different PNN approaches used to assist the diagnosis process for hepatic diseases

Intelligent Analysis of Acute Bed Overflow in a Tertiary Hospital in Singapore

Hospital beds are a scarce resource and always in need. The beds are often organized by clinical specialties for better patient care. When the Accident & Emergency Department (A&E) admits a patient, there may not be an available bed that matches the requested specialty. The patient may be thus asked to wait at the A&E till a matching bed is available, or assigned a bed from a different specialty, which results in bed overflow. While this allows the patient to have faster access to an inpatient bed and treatment, it creates other problems. For instance, nursing care may be suboptimal and the doctors will need to spend more time to locate the overflow patients. The decision to allocate an overflow bed, or to let the patient wait a bit longer, can be a complicated one. While there can be a policy to guide the bed allocation decision, in reality it depends on clinical calls, current supply and waiting list, projected supply (i.e. planned discharges) and demand. The extent of bed overflow can therefore vary greatly, both in time dimension and across specialties. In this study, we extracted hospital data and used statistical and data mining approaches to identify the patterns behind bed overflow. With this insight, the hospital administration can be better equipped to devise strategies to reduce bed overflow and therefore improve patient care. Computational results show the viability of these intelligent data analysis techniques for understanding and managing the bed overflow problem