Estimating tree-based dynamic treatment regimes using observational data with restricted treatment sequences

A dynamic treatment regime (DTR) is a sequence of decision rules that provide guidance on how to treat individuals based on their static and time-varying status. Existing observational data are often used to generate hypotheses about effective DTRs. A common challenge with observational data, however, is the need for analysts to consider “restrictions” on the treatment sequences. Such restrictions may be necessary for settings where (1) one or more treatment sequences that were offered to individuals when the data were collected are no longer considered viable in practice, (2) specific treatment sequences are no longer available, or (3) the scientific focus of the analysis concerns a specific type of treatment sequences (eg, “stepped-up” treatments). To address this challenge, we propose a restricted tree–based reinforcement learning (RT-RL) method that searches for an interpretable DTR with the maximum expected outcome, given a (set of) user-specified restriction(s), which specifies treatment options (at each stage) that ought not to be considered as part of the estimated tree-based DTR. In simulations, we evaluate the performance of RT-RL versus the standard approach of ignoring the partial data for individuals not following the (set of) restriction(s). The method is illustrated using an observational data set to estimate a two-stage stepped-up DTR for guiding the level of care placement for adolescents with substance use disorder.     

Integrating sample similarities into latent class analysis: a tree-structured shrinkage approach

This paper is concerned with using multivariate binary observations to estimate the probabilities of unobserved classes with scientific meanings. We focus on the setting where additional information about sample similarities is available and represented by a rooted weighted tree. Every leaf in the given tree contains multiple samples. Shorter distances over the tree between the leaves indicate a priori higher similarity in class probability vectors. We propose a novel data integrative extension to classical latent class models with tree-structured shrinkage. The proposed approach enables (1) borrowing of information across leaves, (2) estimating data-driven leaf groups with distinct vectors of class probabilities, and (3) individual-level probabilistic class assignment given the observed multivariate binary measurements. We derive and implement a scalable posterior inference algorithm in a variational Bayes framework. Extensive simulations show more accurate estimation of class probabilities than alternatives that suboptimally use the additional sample similarity information. A zoonotic infectious disease application is used to illustrate the proposed approach. The paper concludes by a brief discussion on model limitations and extensions.     

Combining parametric and nonparametric models to estimate treatment effects in observational studies

Performing causal inference in observational studies requires we assume confounding variables are correctly adjusted for. In settings with few discrete-valued confounders, standard models can be employed. However, as the number of confounders increases these models become less feasible as there are fewer observations available for each unique combination of confounding variables. In this paper, we propose a new model for estimating treatment effects in observational studies that incorporates both parametric and nonparametric outcome models. By conceptually splitting the data, we can combine these models while maintaining a conjugate framework, allowing us to avoid the use of Markov chain Monte Carlo (MCMC) methods. Approximations using the central limit theorem and random sampling allow our method to be scaled to high-dimensional confounders. Through simulation studies we show our method can be competitive with benchmark models while maintaining efficient computation, and illustrate the method on a large epidemiological health survey.     

Evidence of biotic resistance to exotic plant invasion in degraded Bornean forests

Intact tropical forests are generally considered to be resistant to invasions by exotic species, although the shrub Clidemia hirta (Melastomataceae) is highly invasive in tropical forests outside its native range. Release from natural enemies (e.g., herbivores and pathogens) contributes to C. hirta invasion success where native melastomes are absent, and here we examine the role of enemies when C. hirta co-occurs with native Melastomataceae species and associated herbivores and pathogens. We study 21 forest sites within agricultural landscapes in Sabah, Malaysian Borneo, recording herbivory rates in C. hirta and related native Melastoma spp. plants along two 100-m transects per site that varied in canopy cover. Overall, we found evidence of enemy release; C. hirta had significantly lower herbivory (median occurrence of herbivory per plant = 79% of leaves per plant; median intensity of herbivory per leaf = 6% of leaf area) than native melastomes (93% and 20%, respectively). Herbivory on C. hirta increased when closer to native Melastoma plants with high herbivory damage, and in more shaded locations, and was associated with fewer reproductive organs on C. hirta. This suggests host-sharing by specialist Melastomataceae herbivores is occurring and may explain why invasion success of C. hirta is lower on Borneo than at locations without related native species present. Thus, natural enemy populations may provide a “biological control service” to suppress invasions of exotic species (i.e., biotic resistance). However, lower herbivory pressures in more open canopy locations may make highly degraded forests within these landscapes more susceptible to invasion.     

Cavity-nesting birds are limited by nesting habitat in Neotropical agricultural landscapes

Most studies comparing biodiversity between natural and human-modified landscapes focus on patterns in species occurrence or abundance, but do not consider how different habitat types meet species' breeding requirements. Organisms that use or nest in tree cavities may be especially threatened by habitat conversion due to the loss of their nesting sites. Although cavity-nesting bird diversity is highest in the tropics, little is known about how tropical birds use cavities, how agriculture affects their reproductive biology, and how effective nest boxes could be as a conservation strategy in tropical agriculture. Here, we explored how habitat conversion from tropical forests to pasture affects the abundance, nesting habitat availability, and nest success of cavity-nesting birds in Northwest Ecuador. We conducted bird surveys and measured natural cavity availability and use in forest and agriculture. We also added artificial nest boxes to forest and agriculture to see whether cavity limitation in agriculture would elicit higher use of artificial nest boxes. We found evidence of cavity limitation in agriculture—there were many more natural cavities in forest than in agriculture, as well as more avian use of nest boxes placed in agriculture as compared to forest. Our results suggest that it is important to retain remnant trees in tropical agriculture to provide critical nesting habitat for birds. In addition, adding nest boxes to tropical agricultural systems could be a good conservation strategy for certain species, including insectivores that could provide pest-control services to farmers. Abstract in Spanish is available with online material.     

Forest resistance to drought in a humid tropical dipterocarp forest in the Western Ghats of India

Droughts are linked to tree die-offs in the biodiverse humid tropics. We assessed drought response of a Dipterocarp Forest and found a marginal decrease in tree survival, indicating drought resistance. Understory and emergent species were sensitive to drought. Urgent focus is needed to understand drought impacts and plant physiological responses in South Asian forests.     

D-amino acid levels in human physiological fluids

Plasma, urine, cerebrospinal fluid (CSF), and amniotic fluid were examined to determine whether free D-amino acids were present and if so at what levels. It was found that D-amino acids exist in all physiological fluids tested, but that their level varied, considerably. The lowest levels of D-amino acids were usually found in amniotic fluid or CSF (almost always <1% of the corresponding L-amino acid). The highest levels were found in urine (usually tenth percent to low percent levels). Pipecolic acid seemed to be different from the other amino acids tested in that it was excreted primarily as the D-enantiomer (often >90%). Correspondingly high levels of D-pipecolic acid were not found in plasma. Some of the trends found in this work seemed to be analogous to those found in a recent rodent study. © 1993 Wiley-Liss, Inc.     

Piping plover chick ecology following landscape-level disturbance

Population declines of disturbance-dependent species due to suppression of natural disturbances are realized across ecosystems. The piping plover (Charadrius melodus; plover), a disturbance-dependent and conservation-reliant shorebird that nests on sandy beaches and barrier islands on the Atlantic Coast, was listed under the United States Endangered Species Act in 1986. In 2012, Hurricane Sandy landed on Fire and Westhampton islands, barrier island nesting sites for plovers in New York, USA. Hurricane Sandy was a natural disturbance in this system, creating abundant nesting habitat. The number of chicks produced by a pair, or a population, is a direct measure of reproductive output, and gaining a better understanding of productivity and chick behavior following large-scale habitat creation may improve plover habitat management and potentially species persistence. We evaluated the effects of landscape features on habitat selection, behavior, and survival of plover broods using logistic regression, generalized linear mixed effects models, and survival models. Plover broods selected flatter sites with less dense vegetation than available at random. Chick foraging rates were highest in moist substrates and were lower in areas of higher nesting plover density. Chick survival was greater for broods that hatched earlier in the breeding season and increased as chicks aged. Generally, providing access to sites with flatter, moist substrates will likely result in higher quality brood rearing habitat on the landscape. Ultimately, vegetation removal and habitat management may be needed to reduce plover nesting density and ensure sufficient habitat, which may in turn improve plover chick survival. Moreover, sustaining natural landscape disturbances such as those resulting from storms, and not taking actions to prevent hurricane-created overwash, will allow these landscape features to persist.     

Decreased catalase activity is the underlying mechanism of oxidant susceptibility in glucose-6-phosphate dehydrogenase-deficient erythrocytes

Historically, it has been theorized that the oxidant sensitivity of glucose-6-phosphate dehydrogenase (G6PD)-deficient erythrocytes arises as a direct consequence of an inability to maintain cellular gluthione (GSH) levels. This study alternatively hypothesizes that decreased NADPH concentration leads to impaired to catalase activity which, in turn, underlies the observed oxidant susceptibility. To investigate this hypothesis, normal and G6PD-deficient erythrocytes and hemolysates were challenged with a H₂O₂-generating agent. The results of this study demonstrated that catalase activity was severely impaired upon H₂O₂ challenge in the G6PD-deficient cell whiel only decrease was observed in normal cells. Supplmentation of either normal or G6PD-deficient hemolysates with purified NADPH was found to significantly (P < 0.001) inhibit catalase inactivation upon oxidant challenge while addition of NADP⁺ had no effect. Analysis of these results demonstrated direct correlation between NADPH concentration and catalase activity (r = 0.881) and an inverse correlation between catalase activity and erythrocyte oxidant sensitivity (r = 0.906). In contrast, no correlation was found to exist between glutathione concentration (r = 0.170) and oxidant sensitivity. Analysis of NADPH/NADPt ration in acatalasemic mouse erythrocytes demonstrated that NADPH maintenance alone was not sufficient to explain oxidant resistance, and that catalase activity was required. This study supports the hypothesis that impaired catalase activity underlies the enhanced oxidant sensitivity of G6PD-deficient erythrocytes and elucidates the importance of NADPH in the maintenance of normal catalase activity.