Bio papers — 2026-10-18

The Morning Preprint — this day as one edition: the day's papers, the five things worth knowing, and the episodes behind them.

Current validation practices are undermining the development of surgical artificial intelligence because without proper testing, these tools cannot be safely adopted in operating rooms. A catalogue of validation pitfalls derived from expert consensus shows that existing methods often ignore the temporal and hierarchical structure of intraoperative videos, leading to misleading results AKMPC.

Ignoring these data structures can obscure critical failure modes and change how algorithms are ranked when analyzing surgical video datasets. This problem is made worse by issues in metric selection and reporting, where experiments on surgical video datasets show that failing to account for temporal or hierarchical data can understate uncertainty AKMPC.

Experts also identified issues in how results are aggregated and reported, which is a category of pitfall found across surgical AI papers through the Delphi process AKMPC.

Other related work touches on data interpretation; for instance, studies on allele-frequency trajectories show that ignoring trajectories leading to loss can produce severely erroneous statistics when analyzing strong selection regimes Waxman. This situation is similar to how misinterpreting video structure can skew AI rankings in surgery.

Integrating different data types presents a challenge, such as linking cardiac imaging phenotypes with molecular states, which requires careful handling of missing data and batch effects Imaging-anchored multiomics in cardiovascular disease. This complexity reflects the need for rigorous validation frameworks to ensure clinical translation.

Finally, there is work on modeling complex dynamics; neural network approaches can effectively handle large chemical reaction networks and rare events by using enhanced sampling strategies Tracking large chemical reaction networks and rare events by neural networks. This demonstrates the need for sophisticated modeling when dealing with high-dimensional systems, whether in biology or surgery.

Today's papers

Current validation practice undermines surgical AI development Surgical data science is rapidly advancing, yet clinical adoption of artificial intelligence in surgery remains limited, with inadequate validation as an important contributing factor. [paper] [episode]
How deterministic trajectories and their fluctuations underlie allele-frequency statistics in the strong-selection regime We determine statistics of allele-frequency trajectories across replicate populations. [paper] [episode]
Imaging-anchored multiomics in cardiovascular disease integrating cardiac imaging, bulk, single-cell, and spatial transcriptomics This review examines joint representations that link cardiac imaging phenotypes to transcriptomic and spatially resolved molecular states. [paper] [episode]
Integrated Open-Source Framework for Simulation of Transcatheter Pulmonary Valves in Native Right Ventricular Outflow Tracts We demonstrate an integrated open-source workflow for simulation of TPVR in image-derived models to inform device selection. [paper] [episode]
Tracking large chemical reaction networks and rare events by neural networks We present an advanced neural-network approach that effectively tackles large reaction networks and rare-event dynamics within a unified framework. [paper] [episode]
Synergy and Complementarity: The Generative Basis of Chemical Organizations Identifying structural features ensuring the persistence of a reaction network is fundamental for understanding the evolution and complexification of biological entities. [paper] [episode]
Stochastic partial differential equation model for environmental DNA dynamics in river environments We propose a first-step mathematical and computational framework for the eDNA dynamics of migratory fish based on a novel stochastic partial differential equation model with a delayed source input. [paper] [episode]
State Space Models Naturally Produce Time Cell and Oscillatory Behaviors and Scale to Abstract Cognitive Functions We propose that a mathematical framework based on State Space Models offers mathematical sufficiency and computational plausibility to generate response profiles that resemble time cells, ramping activity, and oscillatory responses. [paper] [episode]

The papers

Current validation practice undermines surgical AI development — Comprehensive Synthesis of Research on Validation Pitfalls in Surgical AI This research addresses a critical bottleneck in the advancement of Surgical Data Science (SDS): the inadequate clinical adoption of Artificial Intelligence (AI) in surgery, primarily due to deficient valid [episode]
How deterministic trajectories and their fluctuations underlie allele-frequency statistics in the strong-selection regime — The gist: This work presents an approximation scheme for allele-frequency statistics in strong selection, ensuring that both fixation and loss trajectories contribute to the statistics by conditioning on their eventual outcome. [episode]
Imaging-anchored multiomics in cardiovascular disease: integrating cardiac imaging, bulk, single-cell, and spatial transcriptomics — The gist: Imaging-anchored multiomics learns joint representations linking cardiac imaging phenotypes to bulk, sc/snRNA-seq, and spatial transcriptomics. [episode]
Integrated Open-Source Framework for Simulation of Transcatheter Pulmonary Valves in Native Right Ventricular Outflow Tracts — The gist: This study demonstrates a novel open-source workflow for simulating transcatheter pulmonary valve replacement in image-derived models to inform device selection. [episode]
Tracking large chemical reaction networks and rare events by neural networks — The gist Tracking large chemical reaction networks and rare events by neural networks demonstrates an advanced framework, NNCME-2, that simultaneously enhances computational efficiency and characterizes rare events in high-dimensional stochastic systems like chemical reaction net [episode]
Synergy and Complementarity: The Generative Basis of Chemical Organizations — The gist Identifying structural features ensuring the persistence of a reaction network is fundamental for understanding the evolution and complexification of biological entities, from origins of life to metabolic engineering Chemical Organization Theory (COT) and Persistent Modu [episode]
Stochastic partial differential equation model for environmental DNA dynamics in river environments — The gist: This study proposes a first-step mathematical and computational framework for eDNA dynamics in river environments using a novel stochastic partial differential equation model with a delayed source input to govern spatiotemporal eDNA concentration during fish migration. [episode]
State Space Models Naturally Produce Time Cell and Oscillatory Behaviors and Scale to Abstract Cognitive Functions — The gist The proposed model suggests that State Space Models can naturally produce time cells and oscillatory behaviors, which are essential for cognitive functions, by leveraging pre-configured structural basis functions and rotational dynamics. [episode]

Important terms

Validation Pitfalls: Common mistakes when testing surgical AI tools, often stemming from ignoring how intraoperative videos have a specific time sequence or structure.
Temporal and Hierarchical Structure: The important way that surgical videos are organized in time and layers. Ignoring this structure can hide serious problems and change how algorithms are judged.
Metric Selection and Reporting: How we choose the right measurements to evaluate AI performance. Failing to consider time or hierarchy in these metrics can make uncertainty look much smaller than it really is.
Data Interpretation Trajectories: In complex data, like allele frequencies, ignoring the path of change can lead to very wrong statistics when analyzing strong selection processes.
Batch Effects and Missing Data: Challenges in combining different types of data, such as imaging and molecular states. This requires careful handling of missing information to ensure accurate results.