Search for: All records

In this work, we introduce a machine learning (ML) pipeline that predicts the time of death (TOD) of a subject from gene expression (GE) profiles, addressing a critical gap in genomic research where TOD data are scarce. Our contributions are fourfold: (1) a data-driven, clinically domain-guided pipeline that learns temporal GE patterns for TOD prediction; (2) a two-stage dimensionality reduction approach combining (I) AutoEncoders and (II) ISOMAP for BA11 & PCA for BA47, preserving the temporal sequence of the data while incorporating domain knowledge and obviating exhaustive searches for optimal circadian gene sequences; (3) systematic training and hyperparameter tuning of 16 ML regressors—including five single, eight ensemble, and three deep learning models—to identify the most effective ML model (i.e., ExtraTrees Regressor for BA11 and AdaBoost Regressor for BA47); and (4) a comprehensive evaluation on 146 subjects, examining 235 circadian gene’s expression patterns per subject across five performance metrics. Our method surpasses both non-temporal-encoding-based and temporal-encoding-based models, achieving hourly scaled MAEs of 0.839 (BA11) and 1.227 (BA47), with corresponding MSE and RMSE values of 1.013/1.006 and 2.153/1.467, respectively. Consequently, TOD predictions fall within a one hour error margin for BA11 and a two hour margin for BA47. 

Abstract Sea ice plays multiple important roles in regulating the global climate. Rapid sea ice loss in the Arctic has been documented over recent decades, yet our understanding of long‐term sea ice variability and its feedbacks remains limited by a lack of quantitative sea ice reconstructions. The sea ice diatom‐derived biomarker has been combined with sterols produced by open‐water phytoplankton in the index as a sea ice proxy to achieve semi‐quantitative reconstructions. Here, we analyze a compilation of over 600 published core‐top measurements of paired with brassicasterol and/or dinosterol across (sub‐)Arctic oceans to calculate a newln() index that correlates nonlinearly with sea ice concentration. Leveraging sediment trap and sea ice observational studies, we develop a spatially varying Bayesian calibration (BaySIC) for ln() to account for its non‐stationary relationship with sea ice concentration and other environmental drivers (e.g., sea surface salinity). The model is fully invertible, allowing probabilistic forward modeling of the ln() index as well as inverse modeling of past sea ice concentration with bi‐directional uncertainty quantification.BaySICfacilitates direct proxy‐model comparisons and palaeoclimate data assimilation, providing the polar proxy constraints currently missing in climate model simulations and enabling, for the first time, fully quantitative Arctic sea ice reconstructions. 

Baffin Bay exports Arctic Water to the North Atlantic while receiving northward flowing Atlantic Water. Warm Atlantic Water has impacted the retreat of tidewater glaciers draining the Greenland Ice Sheet. Periods of enhanced Atlantic Water transport into Baffin Bay have been observed, but the oceanic processes are still not fully explained. At the end of 2010 the net transport at Davis Strait, the southern gateway to Baffin Bay, reversed from southward to northward for a month, leading to significant northward oceanic heat transport into Baffin Bay. This was associated with an extreme high in the Greenland Blocking Index and a stormtrack path that shifted away from Baffin Bay. Thus fewer cyclones in the Irminger Sea resulted in less frequent northerly winds along the western coast of Greenland, allowing anomalous northward penetration of warm waters, reversing the volume and heat transport at Davis Strait.