An Incremental EM-based Learning Approach for On-Line Prediction of Hospital Resource Utilization

Objective: Inpatient length of stay (LOS) is an important measure of hospital activity, health care resource consumption, and patient acuity. This research work aims at developing an incremental expectation maximization (EM) based learning approach on mixture of experts (ME) system for on-line prediction of LOS. The use of a batchmode learning process in most existing artificial neural networks to predict LOS is unrealistic, as the data become available over time and their pattern change dynamically. In contrast, an on-line process is capable of providing an output whenever a new datum becomes available. This on-the-spot information is therefore more useful and practical for making decisions, especially when one deals with a tremendous amount of data. Methods and material: The proposed approach is illustrated using a real example of gastroenteritis LOS data. The data set was extracted from a retrospective cohort study on all infants born in 1995-1997 and their subsequent admissions for gastroenteritis. The total number of admissions in this data set was n = 692. Linked hospitalization records of the cohort were retrieved retrospectively to derive the outcome measure, patient demographics, and associated co-morbidities information. A comparative study of the incremental learning and the batch-mode learning algorithms is considered. The performances of the learning algorithms are compared based on the mean absolute difference (MAD) between the predictions and the actual LOS, and the proportion of predictions with MAD < 1 day (Prop(MAD < 1)). The significance of the comparison is assessed through a regression analysis. Results: The incremental learning algorithm provides better on-line prediction of LOS when the system has gained sufficient training from more examples (MAD = 1.77 days and Prop(MAD < 1) = 54.3%), compared to that using the batch-mode learning. The regression analysis indicates a significant decrease of MAD (p-value = 0.063) and a significant (p-value = 0.044) increase of Prop(MAD

A funny thing happened on the way to the forum: Electronic discussion and foreign language learning

Amongst the opportunities for cross-cultural contact created by the burgeoning use of the Internet are those provided by electronic discussion lists. This study looks at what happens when language students venture out of the classroom (virtual or otherwise) to participate in on-line discussion groups with native speakers. Responses to messages and commentary by moderators and other participants on the (in) appropriateness of contributions allow us to determine what constitutes successful participation and to make suggestions regarding effective teaching strategies for this medium. A case study examines the threads started by four anglophone students of French when they post messages to a forum on the Web site of the French newspaper Le Monde. Investigation of these examples points to the ways in which electronic discussion inflects and is inflected by cultural and generic expectations. We suggest that successful participation on Internet fora depends on awareness of such cultural and generic mores and an ability to work within and/or with them. Teachers therefore need to find ways in which students can be sensitized to such issues so that their participation in such electronic discussion is no longer seen as linguistic training, but as engagement with a cultural practice.

Evaluation by respirometry of the degradability of retort water using a shale ash and overburden packed column

Oil shale processing produces an aqueous wastewater stream known as retort water. The fate of the organic content of retort water from the Stuart oil shale project (Gladstone, Queensland) is examined in a proposed packed bed treatment system consisting of a 1:1 mixture of residual shale from the retorting process and mining overburden. The retort water had a neutral pH and an average unfiltered TOC of 2,900 mg l(-1). The inorganic composition of the retort water was dominated by NH4+. Only 40% of the total organic carbon (TOC) in the retort water was identifiable, and this was dominated by carboxylic acids. In addition to monitoring influent and effluent TOC concentrations, CO2 evolution was monitored on line by continuous measurements of headspace concentrations and air flow rates. The column was run for 64 days before it blocked and was dismantled for analysis. Over 98% of the TOC was removed from the retort water. Respirometry measurements were confounded by CO2 production from inorganic sources. Based on predictions with the chemical equilibrium package PHREEQE, approximately 15% of the total CO2 production arose from the reaction of NH4+ with carbonates. The balance of the CO2 production accounted for at least 80% of the carbon removed from the retort water. Direct measurements of solid organic carbon showed that approximately 20% of the influent carbon was held-up in the top 20cm of the column. Less than 20% of this held-up carbon was present as either biomass or as adsorbed species. Therefore, the column was ultimately blocked by either extracellular polymeric substances or by a sludge that had precipitated out of the retort water.