Search for: All records

Abstract Physics-informed machine learning bridges the gap between the high fidelity of mechanistic models and the adaptive insights of artificial intelligence. In chemical reaction network modeling, this synergy proves valuable, addressing the high computational costs of detailed mechanistic models while leveraging the predictive power of machine learning. This study applies this fusion to the biomedical challenge of A$$\beta$$fibril aggregation, a key factor in Alzheimer’s disease. Central to the research is the introduction of an automatic reaction order model reduction framework, designed to optimize reduced-order kinetic models. This framework represents a shift in model construction, automatically determining the appropriate level of detail for reaction network modeling. The proposed approach significantly improves simulation efficiency and accuracy, particularly in systems like A$$\beta$$aggregation, where precise modeling of nucleation and growth kinetics can reveal potential therapeutic targets. Additionally, the automatic model reduction technique has the potential to generalize to other network models. The methodology offers a scalable and adaptable tool for applications beyond biomedical research. Its ability to dynamically adjust model complexity based on system-specific needs ensures that models remain both computationally feasible and scientifically relevant, accommodating new data and evolving understandings of complex phenomena. 

The Caribbean through-flow (CTF) is a vital component of Earth’s climate system, facilitating and impacted by heat and salt fluxes from major circulation systems like the North Atlantic Subtropical Gyre (NASTG) and Atlantic Meridional Overturning Circulation (AMOC). Here, we show significant changes have occurred in upper ocean water mass properties of the CTF since 1960, including subsurface warming of ~ 0.2 °C decade−1, surface freshening of ~ 0.13 g kg−1 decade−1, and subsurface salinification of ~ 0.05 g kg−1 decade−1. In the upper 0–200 m, temperature and stability increases are nearly 3 and 20 times larger than globally averaged trends, respectively, with implications for tropical cyclones, sea level rise, and marine ecosystems. We show these upper ocean changes are likely impacting water mass formation in the NASTG, thereby indirectly influencing the AMOC. These findings highlight the CTF as a bottleneck for climatically important water masses and emphasize the need for sustained subsurface observations here. 

The Caribbean Through-Flow (CTF) is a critical chokepoint for North and South Atlantic waters that form the North Atlantic western boundary current system and the upper ocean limb of the Atlantic Meridional Overturning Circulation. While the circulation and energetics of the CTF have been well studied, its water mass transformations remain poorly constrained. Using over 7700 Argo float profiles from 2014 to 2024, we document a prominent westward modification in water mass structure across the Caribbean Sea. From the eastern to western Caribbean, we observe systematic increases in ocean heat content, a deepening of isopycnals, and a freshening and deepening of the subsurface salinity maximum. These changes result in a net mid-depth (~50–500 m) density reduction of 0.40 ± 0.27 kg m-3. We hypothesize that regional variations in mesoscale eddy activity, complex bathymetry, and meridional wind stress curl gradients drive this transformation. The resulting water mass structure has critical implications for regional climate, weather, ecosystems, and sea level rise, as it modifies the density and stratification of source waters entering the Gulf of Mexico and North Atlantic western boundary current system. Our findings highlight the importance of internal Caribbean processes in shaping upper-ocean heat and salt transport in the Atlantic and underscore the need for sustained in situ observations in the region and targeted modeling analyses of the underlying modification processes. 

Sepsis is a dysregulated host response to infection with high mortality and morbidity. Early detection and intervention have been shown to improve patient outcomes, but existing computational models relying on structured electronic health record data often miss contextual information from unstructured clinical notes. This study introduces COMPOSER-LLM, an open-source large language model (LLM) integrated with the COMPOSER model to enhance early sepsis prediction. For high-uncertainty predictions, the LLM extracts additional context to assess sepsis-mimics, improving accuracy. Evaluated on 2500 patient encounters, COMPOSER-LLM achieved a sensitivity of 72.1%, positive predictive value of 52.9%, F-1 score of 61.0%, and 0.0087 false alarms per patient hour, outperforming the standalone COMPOSER model. Prospective validation yielded similar results. Manual chart review found 62% of false positives had bacterial infections, demonstrating potential clinical utility. Our findings suggest that integrating LLMs with traditional models can enhance predictive performance by leveraging unstructured data, representing a significant advance in healthcare analytics.