Tree-based modeling of complex interactions of phosphorus loadings and environmental factors.

Phosphorus (P) enrichment has been observed in the historic oligotrophic Greater Everglades in Florida mainly due to P influx from upstream, agriculturally dominated, low relief drainage basins of the Everglades Agricultural Area (EAA). Our specific objectives were to: (1) investigate relationships between various environmental factors and P loads in 10 farm basins within the EAA, (2) identify those environmental factors that impart major effects on P loads using three different tree-based modeling approaches, and (3) evaluate predictive models to assess P loads. We assembled thirteen environmental variable sets for all 10 sub-basins characterizing water level management, cropping practices, soils, hydrology, and farm-specific properties. Drainage flow and P concentrations were measured at each sub-basin outlet from 1992-2002 and aggregated to derive monthly P loads. We used three different tree-based models including single regression trees (ST), committee trees in Bagging (CTb) and ARCing (CTa) modes and ten-fold cross-validation to test prediction performances. The monthly P loads (MPL) during the monitoring period showed a maximum of 2528 kg (mean: 103 kg) and maximum monthly unit area P loads (UAL) of 4.88 kg P ha(-1) (mean: 0.16 kg P ha(-1)). Our results suggest that hydrologic/water management properties are the major controlling variables to predict MPL and UAL in the EAA. Tree-based modeling was successful in identifying relationships between P loads and environmental predictor variables on 10 farms in the EAA indicated by high R(2) (>0.80) and low prediction errors. Committee trees in ARCing mode generated the best performing models to predict P loads and P loads per unit area. Tree-based models had the ability to analyze complex, non-linear relationships between P loads and multiple variables describing hydrologic/water management, cropping practices, soil and farm-specific properties within the EAA.

Malaria drug and vaccine trials in Africa: obstacles and opportunities.

There are several new treatments and vaccine technologies in clinical development for childhood malaria that have arrived in the clinical phase of evaluation during the past 5-10 years. This is a long-awaited change as until this time there had been little in the pipeline. As these products progress, evaluating them in the populations for whom they are being developed is becoming increasingly challenging. Many more capable trial sites are required and thousands of children and their parents need to be willing to take part in all the clinical trials that will be necessary if even a handful of these products make it through to obtaining a marketing approval license. Then, beyond licensure, these products will need to be assessed in more 'real-life' phase IV trials to establish whether they can truly impact the high level of mortality that malaria brings to the under-five population in Africa. Here we explore the issues that face both the trial sites and the product developers and present how this opportunity should be utilised to develop experienced African clinical researchers and facilities alongside getting these products through into public health use.

Approaching the community about screening children for a multicentre malaria vaccine trial.

Community sensitisation, as a component of community engagement, plays an important role in strengthening the ethics of community-based trials in developing countries and is fundamental to trial success. However, few researchers have shared their community sensitisation strategies and experiences. We report on our perspective as researchers on the sensitisation activities undertaken for a phase II malaria vaccine trial in Kilifi District (Kenya) and Korogwe District (Tanzania), with the aim of informing and guiding the operational planning of future trials. We report wide variability in recruitment rates within both sites; a variability that occurred against a backdrop of similarity in overall approaches to sensitisation across the two sites but significant differences in community exposure to biomedical research. We present a range of potential factors contributing to these differences in recruitment rates, which we believe are worth considering in future community sensitisation plans. We conclude by arguing for carefully designed social science research around the implementation and impact of community sensitisation activities.

The illusion of certainty and the certainty of illusion: a caution when reading scientific articles.

Critical thinking is necessary to read the scientific literature. However, in addition to questions about the science, often one must also question the meaning of the text. This article provides an example of the analyses needed to understand a single sentence. In so doing, it raises several interesting issues of meaning, measurement, statistical analyses, and the form in which results are presented and interpreted.

Katsu: traditional Japanese resuscitation methods.

Twelve traditional Japanese resuscitation methods, called Katsu, are described, and their history and medical aspects are compared with those of traditional Western methods of manual resuscitation. Five katsu, developed parallel to but independently of traditional Western methods, are inferior to CPR and are of historical interest only. Seven katsu not evaluated by Western researchers could be valuable if proven effective because they can be applied where CPR cannot.

Efficacy of laser interferometry in predicting visual result of YAG laser posterior capsulotomy.

One hundred fifteen patients with cloudy posterior capsules that required capsulotomies were evaluated by laser interferometry. The predicted acuity was within one line of postoperative acuity in 44 patients (38.3%) and within two lines in 73 patients (63.5%). The correlation coefficient between the predicted and actual post-capsulotomy acuities was 0.38. However, the predicted retinal visual acuity was compared to postoperative Snellen visual acuity and found to differ by an average of three lines. A large number of false negative results reduced the clinical usefulness of the test.