Project description:BackgroundAn increasing number of studies have identified spatial differences in colorectal cancer survival. However, little is known about the spatially varying effects of predictors in survival prediction modeling studies of colorectal cancer that have focused on estimating the absolute survival risk for patients from a wide range of populations. This study aimed to demonstrate the spatially varying effects of predictors of survival for nonmetastatic colorectal cancer patients.MethodsPatients diagnosed with nonmetastatic colorectal cancer from 2004 to 2013 who were followed up through the end of 2013 were extracted from the Surveillance Epidemiology End Results registry (Patients: 128061). The log-rank test and the restricted mean survival time were used to evaluate survival outcome differences among spatial clusters corresponding to a widely used clinical predictor: stage determined by AJCC 7th edition staging system. The heterogeneity test, which is used in meta-analyses, revealed the spatially varying effects of single predictors. Then, considering the above predictors in a standard survival prediction model based on spatially clustered data, the spatially varying coefficients of these models revealed that some covariate effects may not be constant across the geographic regions of the study. Then, two types of survival prediction models (a statistical model and a machine learning model) were built; these models considered the predictors and enabled survival prediction for patients from a wide range of geographic regions.ResultsBased on univariate and multivariate analysis, some prognostic factors, such as "TNM stage", "tumor size" and "age at diagnosis," have significant spatially varying effects among different regions. When considering these spatially varying effects, machine learning models have fewer assumption constraints (such as proportional hazard assumptions) and better predictive performance compared with statistical models. Upon comparing the concordance indexes of these two models, the machine learning model was found to be more accurate (0.898[0.895,0.902]) than the statistical model (0.732 [0.726, 0.738]).ConclusionsBased on this study, it's recommended that the spatially varying effect of predictors should be considered when building survival prediction models involving large-scale and multicenter research data. Machine learning models that are not limited by the requirement of a statistical hypothesis are promising alternative models.

Project description:Populations of the same species can experience different responses to the environment throughout their distributional range as a result of spatial and temporal heterogeneity in habitat conditions. This highlights the importance of understanding the processes governing species distribution at local scales. However, research on species distribution often averages environmental covariates across large geographic areas, missing variability in population-environment interactions within geographically distinct regions. We used spatially explicit models to identify interactions between species and environmental, including chlorophyll a (Chla) and sea surface temperature (SST), and trophic (prey density) conditions, along with processes governing the distribution of two cephalopods with contrasting life-histories (octopus and squid) across the western Mediterranean Sea. This approach is relevant for cephalopods, since their population dynamics are especially sensitive to variations in habitat conditions and rarely stable in abundance and location. The regional distributions of the two cephalopod species matched two different trophic pathways present in the western Mediterranean Sea, associated with the Gulf of Lion upwelling and the Ebro river discharges respectively. The effects of the studied environmental and trophic conditions were spatially variant in both species, with usually stronger effects along their distributional boundaries. We identify areas where prey availability limited the abundance of cephalopod populations as well as contrasting effects of temperature in the warmest regions. Despite distributional patterns matching productive areas, a general negative effect of Chla on cephalopod densities suggests that competition pressure is common in the study area. Additionally, results highlight the importance of trophic interactions, beyond other common environmental factors, in shaping the distribution of cephalopod populations. Our study presents a valuable approach for understanding the spatially variant ecology of cephalopod populations, which is important for fisheries and ecosystem management.

Project description:BackgroundPrediction models for time-to-event outcomes are commonly used in biomedical research to obtain subject-specific probabilities that aid in making important clinical care decisions. There are several regression and machine learning methods for building these models that have been designed or modified to account for the censoring that occurs in time-to-event data. Discrete-time survival models, which have often been overlooked in the literature, provide an alternative approach for predictive modeling in the presence of censoring with limited loss in predictive accuracy. These models can take advantage of the range of nonparametric machine learning classification algorithms and their available software to predict survival outcomes.MethodsDiscrete-time survival models are applied to a person-period data set to predict the hazard of experiencing the failure event in pre-specified time intervals. This framework allows for any binary classification method to be applied to predict these conditional survival probabilities. Using time-dependent performance metrics that account for censoring, we compare the predictions from parametric and machine learning classification approaches applied within the discrete time-to-event framework to those from continuous-time survival prediction models. We outline the process for training and validating discrete-time prediction models, and demonstrate its application using the open-source R statistical programming environment.ResultsUsing publicly available data sets, we show that some discrete-time prediction models achieve better prediction performance than the continuous-time Cox proportional hazards model. Random survival forests, a machine learning algorithm adapted to survival data, also had improved performance compared to the Cox model, but was sometimes outperformed by the discrete-time approaches. In comparing the binary classification methods in the discrete time-to-event framework, the relative performance of the different methods varied depending on the data set.ConclusionsWe present a guide for developing survival prediction models using discrete-time methods and assessing their predictive performance with the aim of encouraging their use in medical research settings. These methods can be applied to data sets that have continuous time-to-event outcomes and multiple clinical predictors. They can also be extended to accommodate new binary classification algorithms as they become available. We provide R code for fitting discrete-time survival prediction models in a github repository.

Project description:BackgroundThe discovery of CRISPR-based gene editing and its application to homing-based gene drive systems has been greeted with excitement, for its potential to control mosquito-borne diseases on a wide scale, and concern, for the invasiveness and potential irreversibility of a release. Gene drive systems that display threshold-dependent behavior could potentially be used during the trial phase of this technology, or when localized control is otherwise desired, as simple models predict them to spread into partially isolated populations in a confineable manner, and to be reversible through releases of wild-type organisms. Here, we model hypothetical releases of two recently engineered threshold-dependent gene drive systems-reciprocal chromosomal translocations and a form of toxin-antidote-based underdominance known as UDMEL-to explore their ability to be confined and remediated.ResultsWe simulate releases of Aedes aegypti, the mosquito vector of dengue, Zika, and other arboviruses, in Yorkeys Knob, a suburb of Cairns, Australia, where previous biological control interventions have been undertaken on this species. We monitor spread to the neighboring suburb of Trinity Park to assess confinement. Results suggest that translocations could be introduced on a suburban scale, and remediated through releases of non-disease-transmitting male mosquitoes with release sizes on the scale of what has been previously implemented. UDMEL requires fewer releases to introduce, but more releases to remediate, including of females capable of disease transmission. Both systems are expected to be confineable to the release site; however, spillover of translocations into neighboring populations is less likely.ConclusionsOur analysis supports the use of translocations as a threshold-dependent drive system capable of spreading disease-refractory genes into Ae. aegypti populations in a confineable and reversible manner. It also highlights increased release requirements when incorporating life history and population structure into models. As the technology nears implementation, further ecological work will be essential to enhance model predictions in preparation for field trials.

Project description:Few studies have evaluated the optimal sampling design for tracking small mammal population trends, especially for rare or difficult to detect species. Spatially explicit capture-recapture (SECR) models present an advancement over non-spatial models by accounting for individual movement when estimating density. The salt marsh harvest mouse (SMHM; Reithrodontomys raviventris) is a federal and California state listed endangered species endemic to the San Francisco Bay-Delta estuary, California, USA; where a population in a subembayment has been continually monitored over an 18-year period using mark-recapture methods. We analyzed capture data within a SECR modeling framework that allowed us to account for differences in detection and movement between sexes. We compared the full dataset to subsampling scenarios to evaluate how the grid size (area) of the trap design, trap density (spacing), and number of consecutive trapping occasions (duration) influenced density estimates. To validate the subsampling methods, we ran Monte Carlo simulations based on the true parameter estimates for each specific year. We found that reducing the area of the trapping design by more than 36% resulted in the inability of the SECR model to replicate density estimates within the SE of the original density estimates. However, when trapping occasions were reduced from 4 to 3-nights the density estimates were indistinguishable from the full dataset. Furthermore, reducing trap density by 50% also resulted in density estimates comparable to the full dataset and was a substantially better model than reducing the trap area by 50%. Overall, our results indicated that moderate reductions in the number of trapping occasions or trap density could yield similar density estimates when using a SECR approach. This approach allows the optimization of field trapping efforts and designs by reducing field efforts while maintaining the same population estimate compared to the full dataset. Using a SECR approach may help other wildlife programs identify sampling efficiencies without sacrificing data integrity for long term monitoring of population densities.

Project description:BackgroundA method that assesses bacterial spatial dissemination was explored. It measures microbial genotypes (defined by electrophoretic patterns or EP), host, location (farm), interfarm Euclidean distance, and time. Its proof of concept (construct and internal validity) was evaluated using a dataset that included 113 Staphylococcus aureus EPs from 1126 bovine milk isolates collected on 23 farms between 1988 and 2005.ResultsConstruct validity was assessed by comparing results based on the interfarm Euclidean distance (a spatially explicit measure) and those produced by the (non-spatial) interfarm number of isolates reporting the same EP. The distance associated with EP spread correlated with the interfarm number of isolates/EP (r = .59, P < 0.02). Internal validity was estimated by comparing results obtained with different versions of the same indices. Concordance was observed between: (a) EP distance (estimated microbial dispersal over space) and EP speed (distance/year, r = .72, P < 0.01), and (b) the interfarm number of isolates/EP (when measured on the basis of non-repeated cow testing) and the same measure as expressed by repeated testing of the same animals (r = .87, P < 0.01). Three EPs (2.6% of all EPs) appeared to be super-spreaders: they were found in 26.75% of all isolates. Various indices differentiated local from spatially disseminated infections and, within the local type, infections suspected to be farm-related were distinguished from cow-related ones.ConclusionFindings supported both construct and internal validity. Because 3 EPs explained 12 times more isolates than expected and at least twice as many isolates as other EPs did, false negative results associated with the remaining EPs (those erroneously identified as lacking spatial dispersal when, in fact, they disseminated spatially), if they occurred, seemed to have negligible effects. Spatial analysis of laboratory data may support disease surveillance systems by generating hypotheses on microbial dispersal ability.

Dataset Information

Spatially-explicit survival modeling with discrete grouping of cancer predictors.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets