Association of Recent Violence Encounters With Suicidal Ideation Among Adolescents With Depression.

Importance: Suicide prevention is an important component of depression management. Knowledge about depressed adolescents with increased risk for suicide can inform suicide prevention efforts. Objective: To describe the risk of documented suicidal ideation within a year following a diagnosis of depression and to examine how the risk of documented suicidal ideation differed by recent violence encounter status among adolescents with new depression diagnoses. Design, Setting, and Participants: Retrospective cohort study in clinical settings including outpatient facilities, emergency departments, and hospitals. Using IBM's Explorys database containing electronic health records from 26 US health care networks, this study observed a cohort of adolescents with new depression diagnoses from 2017 to 2018 for up to 1 year. Data were analyzed from July 2020 to July 2021. Exposures: Recent violence encounter was defined by a diagnosis of child maltreatment (physical, sexual, or psychological abuse or neglect) or physical assault within 1 year before depression diagnosis. Main Outcomes and Measures: The main outcome was diagnosis of suicidal ideation within 1 year following depression diagnosis. Multivariable adjusted risk ratios of suicidal ideation were calculated for overall recent violence encounters and for individual forms of violence. Results: Among a total of 24 047 adolescents with depression, 16 106 (67.0%) were female and 13 437 (55.9%) were White. A total of 378 had experienceda violence (hereafter, encounter group) and 23 669 had not (hereafter, nonencounter group). Following the diagnosis of depression, 104 adolescents with any past-year violence encounter (27.5%) documented suicidal ideation within 1 year. In contrast, 3185 adolescents in the nonencounter group (13.5%) experienced thoughts of suicide following the diagnosis of depression. In multivariable analyses, those with any violence encounter had 1.7 times (95% CI 1.4-2.0) higher risk of documented suicidal ideation compared with those in the nonencounter group (P < .001). Among different forms of violence, sexual abuse (risk ratio, 2.1; 95% CI, 1.6-2.8) and physical assault (risk ratio, 1.7; 95% CI, 1.3-2.2) were associated with significantly increased risk of suicidal ideation. Conclusions and Relevance: Among adolescents with depression, persons who experienced past-year violence encounters showed a higher rate of suicidal ideation than those who had not. These findings highlight the importance of identifying and accounting for past violence encounters when treating adolescents with depression to reduce risk of suicide. Public health approaches to prevent violence may help to avert morbidity associated with depression and suicidal ideation.

Blending Knowledge in Deep Recurrent Networks for Adverse Event Prediction at Hospital Discharge.

Deep learning architectures have an extremely high-capacity for modeling complex data in a wide variety of domains. However, these architectures have been limited in their ability to support complex prediction problems using insurance claims data, such as readmission at 30 days, mainly due to data sparsity issue. Consequently, classical machine learning methods, especially those that embed domain knowledge in handcrafted features, are often on par with, and sometimes outperform, deep learning approaches. In this paper, we illustrate how the potential of deep learning can be achieved by blending domain knowledge within deep learning architectures to predict adverse events at hospital discharge, including readmissions. More specifically, we introduce a learning architecture that fuses a representation of patient data computed by a self-attention based recurrent neural network, with clinically relevant features. We conduct extensive experiments on a large claims dataset and show that the blended method outperforms the standard machine learning approaches.

Disturbance structures canopy and understory productivity along an environmental gradient.

Disturbances often disproportionately impact different vegetation layers in forests and other vertically stratified ecosystems, shaping community structure and ecosystem function. However, disturbance-driven changes may be mediated by environmental conditions that affect habitat quality and species interactions. In a decade-long field experiment, we tested how kelp forest net primary productivity (NPP) responds to repeated canopy loss along a gradient in grazing and substrate suitability. We discovered that habitat quality can mediate the effects of intensified disturbance on canopy and understory NPP. Experimental annual and quarterly disturbances suppressed total macroalgal NPP, but effects were strongest in high-quality habitats that supported dense kelp canopies that were removed by disturbance. Understory macroalgae partly compensated for canopy NPP losses and this effect magnified with increasing habitat quality. Disturbance-driven increases in understory NPP were still rising after 5-10 years of disturbance, demonstrating the value of long-term experimentation for understanding ecosystem responses to changing disturbance regimes.

Improved estimates of net primary production, growth, and standing crop of Macrocystis pyrifera in Southern California.

The giant kelp Macrocystis pyrifera forms subtidal forests on shallow reefs in temperate regions of the world. It is one of the fastest-growing multicellular autotrophs on Earth and its high productivity supports diverse marine food webs. In 2008, we published a method for estimating biomass and net primary production (NPP) of giant kelp along with five years of data, to provide a more integrated measure of NPP than those yielded by previous methods. Our method combines monthly field measurements of standing crop and loss rates with a model of kelp biomass dynamics to estimate instantaneous mass-specific growth rates and NPP for each season of each year. We have since improved our approach to account for several previously unresolved sources of biomass loss. These improvements have led to a near doubling of our prior estimates of growth and NPP. At our site with the most persistent stand of giant kelp, NPP averages ~5.2 kg dry mass·m-2 ·yr-1 and results from the rapid growth (~3.5% per d) of a relatively small standing biomass (~0.4 kg dry mass/m2 on average) that turns over ~12 times annually. Here we provide revised estimates of seasonal biomass, growth, and NPP for the five years covered by our previous publication (2002-2006), along with more than a decade of additional data (2007-2017). We also present updated relationships for predicting giant kelp biomass and NPP from much more easily obtained measurements of frond density. These data can be used to understand the mechanisms that drive variation in giant kelp NPP at a wide range of temporal scales. No copyright or proprietary restrictions are associated with the use of this data set other than citation of this Data Paper.

Extreme warming challenges sentinel status of kelp forests as indicators of climate change.

The desire to use sentinel species as early warning indicators of impending climate change effects on entire ecosystems is attractive, but we need to verify that such approaches have sound biological foundations. A recent large-scale warming event in the North Pacific Ocean of unprecedented magnitude and duration allowed us to evaluate the sentinel status of giant kelp, a coastal foundation species that thrives in cold, nutrient-rich waters and is considered sensitive to warming. Here, we show that giant kelp and the majority of species that associate with it did not presage ecosystem effects of extreme warming off southern California despite giant kelp's expected vulnerability. Our results challenge the general perception that kelp-dominated systems are highly vulnerable to extreme warming events and expose the more general risk of relying on supposed sentinel species that are assumed to be very sensitive to climate change.

Patterns and controls of the dynamics of net primary production by understory macroalgal assemblages in giant kelp forests.

Macroalgae are important primary producers in many subtidal habitats, yet little information exists on the temporal and spatial dynamics of net primary production (NPP) by entire subtidal assemblages. This knowledge gap reflects the logistical challenges in measuring NPP of diverse macroalgal assemblages in shallow marine habitats. Here, we couple a simple primary production model with nondestructive estimates of taxon-specific biomass on subtidal reefs off Santa Barbara, California to produce a 4-year time series of net primary production by intact assemblages of understory macroalgae in giant kelp forests off Santa Barbara, California, USA. Daily bottom irradiance varied significantly throughout the year, and algal assemblages were on average exposed to saturating irradiance for only 1.3-4.5 h per day, depending on the time of year. Despite these variable light-limiting conditions, biomass rather than irradiance explained the vast majority of variation observed in daily NPP at all times of the year. Measurements of peak biomass in spring and summer proved to be good predictors of NPP for the entire year, explaining as much as 76% of the observed variation. In contrast, bottom irradiance was a poor predictor of NPP, explaining <10% of the variation in NPP when analyzed seasonally and ~2% when evaluated annually. Our finding that annual NPP by macroalgal assemblages can be predicted from a single, nondestructive measurement of biomass should prove useful for developing time series data that are necessary to evaluate natural and anthropogenic changes in NPP by one of the world's most productive ecosystems.

Addition of species abundance and performance predicts community primary production of macroalgae.

Understory plant assemblages are important sources of primary production in both terrestrial and marine environments, and they may exhibit different dynamics than their overstory counterparts. For example, production within dense upper canopies is typically light-limited by shading, whereas such canopy architecture effects are likely unimportant in low-light environments, such as those inhabited by sparser understory assemblages. In these assemblages, light saturation of understory production may be common as species become limited by their photosynthetic capacity, which is adapted to low-light levels. Here we show that a simple model relating species-specific light use relationships measured in the laboratory to biomass and light levels measured in nature accurately predicts community gross primary production (GPP) in a marine understory algal community. We validate the model by comparing GPP measured in situ in enclosed chambers with model estimates for the same incubations. Model estimates of GPP explained 70% of the variation in the measured estimates. The results show that GPP was accurately estimated by simple addition of the photosynthetic capacity of each species in the community based on their biomass and the available light. The difference between modeled and measured GPP did not show any relationship with community biomass or diversity, and the results suggest that diversity does not significantly affect productivity in this system. This type of model should be applicable in other environments where canopy architecture does not play a significant role in limiting photosynthesis.