AI as an intervention: improving clinical outcomes relies on a causal approach to AI development and validation.

The primary practice of healthcare artificial intelligence (AI) starts with model development, often using state-of-the-art AI, retrospectively evaluated using metrics lifted from the AI literature like AUROC and DICE score. However, good performance on these metrics may not translate to improved clinical outcomes. Instead, we argue for a better development pipeline constructed by working backward from the end goal of positively impacting clinically relevant outcomes using AI, leading to considerations of causality in model development and validation, and subsequently a better development pipeline.

A time-calibrated salamander phylogeny including 765 species and 503 genes.

Recent time-calibrated amphibian phylogenies agree on the family-level relationships among extant salamanders but had disparate sampling regimes and inferred very different divergence times. For example, a recent phylogenomic study based on 220 nuclear loci had limited taxon sampling (41 species) and estimated relatively young divergence dates, whereas a more extensive supermatrix study based on 15 genes and 481 species estimated dates that were 22-45 million years older for major clades. Here, we combined phylogenomic and supermatrix approaches to estimate the largest salamander phylogeny to date based on molecular markers.

What Predicts Gene Flow During Speciation? The Relative Roles of Time, Space, Morphology and Climate.

The processes that restrict gene flow between populations are fundamental to speciation. Here, we develop a simple framework for studying whether divergence in morphology, climatic niche, time and space contribute to reduced gene flow among populations and species. We apply this framework to a model system involving a clade of spiny lizards (Sceloporus) occurring mostly in northeastern Mexico, which show striking variation in morphology and habitat among closely related species and populations.

Racial differences in laboratory testing as a potential mechanism for bias in AI: A matched cohort analysis in emergency department visits.

AI models are often trained using available laboratory test results. Racial differences in laboratory testing may bias AI models for clinical decision support, amplifying existing inequities. This study aims to measure the extent of racial differences in laboratory testing in adult emergency department (ED) visits.