Browse Articles

Stereodivergent Construction of 3,3′-Disubstituted Oxindoles via One-Pot Sequential Allylation/Alkylation and Its Application to the Total Synthesis of Trigolute B and D

Computational study of skyrmion stability and transport on W/CoFeB

The Local-Scale Origin of Ferroic Properties in BiVO₄

Evaluating the contribution of parallel processing of color and shape in a conjunction search task

Molecular Photoelectrodes with Enhanced Photogenerated Charge Transport for Efficient Solar Hydrogen Evolution

A mixed methods study protocol to develop an educational program based on salutogenesis theory to improve the postpartum quality of life among nulliparous women

Diastereomeric Configuration Drives an On-Surface Specific Rearrangement into Low Bandgap Non-Benzenoid Graphene Nanoribbons

Cold storage surpasses the impact of biological age and donor characteristics on red blood cell morphology classified by deep machine learning

Irreversible Deactivation Pathways in Ni(II)-Catalyzed Nonalternating Ethylene–Carbon Monoxide Copolymerization

Influence of soil-rock composite stratum on mechanical response and failure modes of underground utility tunnel

Bro̷nsted Acid-Catalyzed Reduction of Furans

Rethinking model prototyping through the MedMNIST+ dataset collection

Abstract The integration of deep learning based systems in clinical practice is often impeded by challenges rooted in limited and heterogeneous medical datasets. In addition, the field has increasingly prioritized marginal performance gains on a few, narrowly scoped benchmarks over clinical applicability, slowing down meaningful algorithmic progress. This trend often results in excessive fine-tuning of existing methods on selected datasets rather than fostering clinically relevant innovations. In response, this work introduces a comprehensive benchmark for the MedMNIST+ dataset collection, designed to diversify the evaluation landscape across several imaging modalities, anatomical regions, classification tasks and sample sizes. We systematically reassess commonly used Convolutional Neural Networks (CNNs) and Vision Transformer (ViT) architectures across distinct medical datasets, training methodologies, and input resolutions to validate and refine existing assumptions about model effectiveness and development. Our findings suggest that computationally efficient training schemes and modern foundation models offer viable alternatives to costly end-to-end training. Additionally, we observe that higher image resolutions do not consistently improve performance beyond a certain threshold. This highlights the potential benefits of using lower resolutions, particularly in prototyping stages, to reduce computational demands without sacrificing accuracy. Notably, our analysis reaffirms the competitiveness of CNNs compared to ViTs, emphasizing the importance of comprehending the intrinsic capabilities of different architectures. Finally, by establishing a standardized evaluation framework, we aim to enhance transparency, reproducibility, and comparability within the MedMNIST+ dataset collection as well as future research. Code is available at (https://github.com/sdoerrich97/rethinking-model-prototyping-MedMNISTPlus).

Intermolecular N–N Coupling of a Dinitrosyl Iron Complex Induced by Hydrogen Bond Donors in the Secondary Coordination Sphere

Spontaneous evolution of density peaking factor in TEM turbulence-dominated H-mode plasma on the EAST Tokamak

Differential Packing of Cs₂Mo₆Br₁₄ Cluster-Based Halide in Variable Diameter Carbon Nanotubes with Elimination and Polymerization to 1D [Mo₂Br₆]_<i>x</i> Ising Model Structures by Steric Confinement

Family-based genetics identifies association of CUBN, IL1RL1 and PRKN variants with leprosy in Bangladesh

Multimodal Precise Control Over Multiselective Carbonylation of 1,3-Enynes

Uncovering hidden insights in the chair rise performance of older adults using Dynamic Time Warping and K-means clustering

Abstract The five time chair rise test (5CRT) is commonly used in geriatric medicine and research to assess functional capacity and lower extremity strength to detect early age-related changes in older adults. Traditional stopwatch-based analyses may mask temporal variations in 5CRT transitions due to averaging. Temporal variations and dynamic characteristics are better assessed by motion variability analysis. This work employs k-means clustering using Dynamic Time Warping (DTW) as a metric for 5CRT to examine compensation mechanisms of older adults. The observational study included 172 healthy, community-dwelling adults aged 70+, yielding 860 chair rises recorded on a force plate and clustered using k-means. Descriptive statistics summarized performance distribution across clusters. Optimal clustering revealed two movement patterns, differing significantly (p $$&lt;0.01$$ ) in 5CRT duration and forces during the stabilization phase. These patterns did not correlate directly with shorter or longer 5CRT durations, indicating overlap and highlighting the limitations of traditional stopwatch methods. This study demonstrates the potential of DTW and k-means clustering in geriatric medicine and research, enabling analysis of 5CRT performance independent of temporal variations, identifying potential health issues undetectable by conventional methods. The k-means model can be further trained to automate analysis, enhancing insights from 5CRT.

A Route toward the On-Surface Synthesis of Organic Ferromagnetic Quantum Spin Chains

Deep learning for hepatocellular carcinoma recurrence before and after liver transplantation: a multicenter cohort study