Search Results (3)

Machine learning (ML) excels over regression by automatically capturing complex, non-linear relationships and interactions, enabling more flexible and accurate predictions without strict assumptions. This study focuses on developing ML-based predictive models for key post-PCI outcomes: 30-day mortality, in-hospital major bleeding, and one-year mortality. Data from 104,665 consecutive PCI cases in the Victorian Cardiac Outcomes Registry (VCOR), collected between 2013 and 2022, will be analyzed. Candidate variables, informed by prior systematic reviews and dataset availability, will undergo multiple imputations for missing values. The Boruta method will be applied to identify influential predictors. Risk-adjusted models will be developed using sophisticated ML algorithms, with performance compared across standard metrics for validation. The dataset will be split, optimized via 10-fold cross-validation, and class imbalance addressed using Adaptive Synthetic resampling technique. SHapley Additive exPlanations will interpret the most influential predictors. The variables from the best model will be converted into simplified numeric scores. External validation will be performed using the Tasmanian dataset or equivalent datasets. This study is expected to identify the most influential variables associated with 30-day all-cause mortality, in-hospital major bleeding, and long-term mortality post-PCI. These variables will form the basis for developing robust risk-scoring models to support clinical decision-making and outcome prediction. Full article

Energy Expenditure (EE) (kcal/day), a key element to guide obesity treatment, is measured from CO₂ production, VCO₂ (mL/min), and/or O₂ consumption, VO₂ (mL/min). Current technologies are limited due to the requirement of wearable facial accessories. A novel system, the Smart Pad, which measures EE via VCO₂ from a room’s ambient CO₂ concentration transients was evaluated. Resting EE (REE) and exercise VCO₂ measurements were recorded using Smart Pad and a reference instrument to study measurement duration’s influence on accuracy. The Smart Pad displayed 90% accuracy (±1 SD) for 14–19 min of REE measurement and for 4.8–7.0 min of exercise, using known room’s air exchange rate. Additionally, the Smart Pad was validated measuring subjects with a wide range of body mass indexes (BMI = 18.8 to 31.4 kg/m²), successfully validating the system accuracy across REE’s measures of ~1200 to ~3000 kcal/day. Furthermore, high correlation between subjects’ VCO₂ and λ for CO₂ accumulation was observed (p < 0.00001, R = 0.785) in a 14.0 m³ sized room. This finding led to development of a new model for REE measurement from ambient CO₂ without λ calibration using a reference instrument. The model correlated in nearly 100% agreement with reference instrument measures (y = 1.06x, R = 0.937) using an independent dataset (N = 56). Full article

The concept of searching and localizing vehicles from live traffic videos based on descriptive textual input has yet to be explored in the scholarly literature. Endowing Intelligent Transportation Systems (ITS) with such a capability could help solve crimes on roadways. One major impediment to the advancement of fine-grain vehicle recognition models is the lack of video testbench datasets with annotated ground truth data. Additionally, to the best of our knowledge, no metrics currently exist for evaluating the robustness and performance efficiency of a vehicle recognition model on live videos and even less so for vehicle search and localization models. In this paper, we address these challenges by proposing V-Localize, a novel artificial intelligence framework for vehicle search and continuous localization captured from live traffic videos based on input textual descriptions. An efficient hashgraph algorithm is introduced to compute valid target information from textual input. This work further introduces two novel datasets to advance AI research in these challenging areas. These datasets include (a) the most diverse and large-scale Vehicle Color Recognition (VCoR) dataset with 15 color classes—twice as many as the number of color classes in the largest existing such dataset—to facilitate finer-grain recognition with color information; and (b) a Vehicle Recognition in Video (VRiV) dataset, a first of its kind video testbench dataset for evaluating the performance of vehicle recognition models in live videos rather than still image data. The VRiV dataset will open new avenues for AI researchers to investigate innovative approaches that were previously intractable due to the lack of annotated traffic vehicle recognition video testbench dataset. Finally, to address the gap in the field, five novel metrics are introduced in this paper for adequately accessing the performance of vehicle recognition models in live videos. Ultimately, the proposed metrics could also prove intuitively effective at quantitative model evaluation in other video recognition applications. T One major advantage of the proposed vehicle search and continuous localization framework is that it could be integrated in ITS software solution to aid law enforcement, especially in critical cases such as of amber alerts or hit-and-run incidents. Full article