Search

We present an ensemble transfer learning method to predict suicide from Veterans Affairs (VA) electronic medical records (EMR). A diverse set of base models was trained to predict a binary outcome constructed from reported suicide, suicide attempt, and overdose diagnoses with varying choices of study design and prediction methodology. Each model used twenty cross-sectional and 190 longitudinal variables observed in eight time intervals covering 7.5 years prior to the time of prediction. Ensembles of seven base models were created and fine-tuned with ten variables expected to change with study design and outcome definition in order to predict suicide and combined outcome in a prospective cohort. The ensemble models achieved c-statistics of 0.73 on 2-year suicide risk and 0.83 on the combined outcome when predicting on a prospective cohort of [Formula: see text] 4.2 M veterans. The ensembles rely on nonlinear base models trained using a matched retrospective nested case-control (Rcc) study cohort and show good calibration across a diversity of subgroups, including risk strata, age, sex, race, and level of healthcare utilization. In addition, a linear Rcc base model provided a rich set of biological predictors, including indicators of suicide, substance use disorder, mental health diagnoses and treatments, hypoxia and vascular damage, and demographics.

Abstract We present an ensemble transfer learning method to predict suicide from Veterans Affairs (VA) electronic medical records (EMR). A diverse set of base models was trained to predict a binary outcome constructed from reported suicide, suicide attempt, and overdose diagnoses with varying choices of study design and prediction methodology. Each model used twenty cross-sectional and 190 longitudinal variables observed in eight time intervals covering 7.5 years prior to the time of prediction. Ensembles of seven base models were created and fine-tuned with ten variables expected to change with study design and outcome definition in order to predict suicide and combined outcome in a prospective cohort. The ensemble models achieved c-statistics of 0.73 on 2-year suicide risk and 0.83 on the combined outcome when predicting on a prospective cohort of $$\sim$$ ∼ 4.2 M veterans. The ensembles rely on nonlinear base models trained using a matched retrospective nested case-control (Rcc) study cohort and show good calibration across a diversity of subgroups, including risk strata, age, sex, race, and level of healthcare utilization. In addition, a linear Rcc base model provided a rich set of biological predictors, including indicators of suicide, substance use disorder, mental health diagnoses and treatments, hypoxia and vascular damage, and demographics.

Abstract Psychological-distress increased at the onset of the COVID-19 pandemic in Argentina. Longitudinal studies in developing countries are scarce. Particularly, Argentina had one of the longest lockdowns. Differences in preventive measures against the virus spread between countries may differentially affect the mental health of the populations. Here we aimed to characterize distinct psychological-distress and related-symptoms trajectories associated with the pandemic and explore risk/protective factors. In this longitudinal study, data from 832 Argentineans were collected every 3–5 months, between April 2020–August 2021. Mean psychological-distress levels and related-symptoms tended to increase over time. However, latent-class analysis identified four distinct psychological-distress trajectories. Most individuals had consistently good mental health (Resilient). Two classes showed psychological-distress worsening during the initial phase of the pandemic and recovered at different time points (Fast Recovery; Slow Recovery). The remaining class maintained a mild -level of psychological-distress and began to deteriorate in March 2021 (Deteriorating) continuously. Individuals who are younger, female, have pre-existing psychiatric diagnoses, or have high neuroticism or lower resilience were more likely to experiencing fluctuations in psychological-distress. The mental health trajectory during the pandemic had a complex dynamic. Although most participants remained resilient, a vulnerable group was detected, which deteriorated over time and should be considered by health-services.

Every two years groups worldwide participate in the Critical Assessment of Protein Structure Prediction (CASP) experiment to blindly test the strengths and weaknesses of their computational methods. CASP has significantly advanced the field but many hurdles still remain, which may require new ideas and collaborations. In 2012 a web-based effort called WeFold, was initiated to promote collaboration within the CASP community and attract researchers from other fields to contribute new ideas to CASP. Members of the WeFold coopetition (cooperation and competition) participated in CASP as individual teams, but also shared components of their methods to create hybrid pipelines and actively contributed to this effort. We assert that the scale and diversity of integrative prediction pipelines could not have been achieved by any individual lab or even by any collaboration among a few partners. The models contributed by the participating groups and generated by the pipelines are publicly available at the WeFold website providing a wealth of data that remains to be tapped. Here, we analyze the results of the 2014 and 2016 pipelines showing improvements according to the CASP assessment as well as areas that require further adjustments and research.

Abstract To reduce inferior vena cava filter (IVCF) related complications, retrieval is recommended whenever possible. Nevertheless, IVCF retrieval rates remain lower than expected, likely due to insufficient follow-up after placement. We evaluated the value of a structured program designed to follow patients by the interventional radiology team up to 5 months after IVCF placement. We prospectively enrolled 366 consecutive patients (mean age 64 ± 17 years; 201 men and 165 women) who benefited from IVCF between March 2015 and February 2020. The program consisted of advising the patient and clinicians to consider IVCF retrieval as soon as possible (standard workflow) and systematically planning an additional follow-up visit at 5-month. Clinical and technical eligibility, as well as technical success for retrieval (TSR) were evaluated. At 5-months, 38 (10.4%) patients were lost to follow-up, and 47 (12.8%) had died. Among survivors, the overall retrieval rate was 58%. The retrieval rates were 83% and 97% for the clinically eligible and technically eligible patients for retrieval, respectively. The 5-month visit enabled 89 additional retrievals (47.8%) compared to the standard workflow. No significant difference was seen in TSR before and after 5 months (p = 0.95). Improved patient tracking with a dedicated IVCF program results in an effective process to identify suitable patients for retrieval and drastically improves retrieval rates in eligible patients. Involving interventionalists in the process improved IVCF patient management.

Abstract In Super-G alpine ski racing mean speed is nearly as high as in Downhill. Hence, the energy dissipated in typical impact accidents is similar. However, unlike Downhill, on Super-G courses no training runs are performed. Accordingly, speed control through course design is a challenging but important task to ensure safety in Super-G. In four male World Cup alpine Super-G races, terrain shape, course setting and the mechanics of a high-level athlete skiing the course were measured with differential global navigation satellite systems (dGNSS). The effects of course setting on skier mechanics were analysed using a linear mixed effects model. To reduce speed by 0.5 m/s throughout a turn, the gate offset needs to be increased by + 51%. This change simultaneously leads to a decrease in minimal turn radius (− 19%), an increase in impulse (+ 27%) and an increase in maximal ground reaction force (+ 6%). In contrast, the same reduction in speed can also be achieved by a − 13% change in vertical gate distance, which also leads to a small reduction in minimal turn radius (− 4%) impulse (− 2%), and no change in maximal ground reaction force; i.e. fewer adverse side effects in terms of safety. It appears that shortening the vertical gate distance is a better and safer way to reduce speed in Super-G than increasing the gate offset.

Abstract Given the intrinsic complexity of cognitive and affective processes affecting how people reason about taxes and their decisions to be compliant with such social duty, we aimed at exploring those latent processes by combining the analysis of their central and peripheral physiological correlates. We asked participants to make realistic economic decisions concerning tax-payment and manipulated the social vs. individual decisional frame. In addition, we took into account the potential role of tax-compliance trait. Thirty self-employed professionals took part in the study and completed a public good game while their autonomic (skin conductance – SC – and heart rate – HR) and neural brain (electroencephalography – EEG) activities were recorded. The analysis of physiological responses during the feedback phase – where participants could be presented or not with a fiscal audit – highlighted: (i) increased tonic SC levels and theta activity in the social condition than in the individual one; (ii) increased HR values when a fiscal audit did not take place, especially in participants who presented an enforced tax-compliance trait. Present findings support the idea that classic economic theories of tax behaviour developed under the assumption that taxpayers act as rational and individualist agents do not provide a comprehensive account for the decision-making process. They add to available evidence highlighting the contribution of psychological and social-affective variables to individuals’ decision-making processes to pay or evade taxes and to their appraisal of the consequences of such choice, as suggested by the ‘slippery slope’ framework.

Abstract Every two years groups worldwide participate in the Critical Assessment of Protein Structure Prediction (CASP) experiment to blindly test the strengths and weaknesses of their computational methods. CASP has significantly advanced the field but many hurdles still remain, which may require new ideas and collaborations. In 2012 a web-based effort called WeFold, was initiated to promote collaboration within the CASP community and attract researchers from other fields to contribute new ideas to CASP. Members of the WeFold coopetition (cooperation and competition) participated in CASP as individual teams, but also shared components of their methods to create hybrid pipelines and actively contributed to this effort. We assert that the scale and diversity of integrative prediction pipelines could not have been achieved by any individual lab or even by any collaboration among a few partners. The models contributed by the participating groups and generated by the pipelines are publicly available at the WeFold website providing a wealth of data that remains to be tapped. Here, we analyze the results of the 2014 and 2016 pipelines showing improvements according to the CASP assessment as well as areas that require further adjustments and research.