Search for: All records

Abstract Accurate prediction of postoperative complications can inform shared decisions regarding prognosis, preoperative risk-reduction, and postoperative resource use. We hypothesized that multi-task deep learning models would outperform conventional machine learning models in predicting postoperative complications, and that integrating high-resolution intraoperative physiological time series would result in more granular and personalized health representations that would improve prognostication compared to preoperative predictions. In a longitudinal cohort study of 56,242 patients undergoing 67,481 inpatient surgical procedures at a university medical center, we compared deep learning models with random forests and XGBoost for predicting nine common postoperative complications using preoperative, intraoperative, and perioperative patient data. Our study indicated several significant results across experimental settings that suggest the utility of deep learning for capturing more precise representations of patient health for augmented surgical decision support. Multi-task learning improved efficiency by reducing computational resources without compromising predictive performance. Integrated gradients interpretability mechanisms identified potentially modifiable risk factors for each complication. Monte Carlo dropout methods provided a quantitative measure of prediction uncertainty that has the potential to enhance clinical trust. Multi-task learning, interpretability mechanisms, and uncertainty metrics demonstrated potential to facilitate effective clinical implementation. 

We present a comprehensive study of the inhomogeneous mixed-valence compound, EuPd₃S₄, by electrical transport, X-ray diffraction, time-domain¹⁵¹Eu synchrotron Mössbauer spectroscopy, and X-ray absorption spectroscopy measurements under high pressure. Electrical transport measurements show that the antiferromagnetic ordering temperature,T_N, increases rapidly from 2.8 K at ambient pressure to 23.5 K at ~19 GPa and plateaus between ~19 and ~29 GPa after which no anomaly associated withT_Nis detected. A pressure-induced first-order structural transition from cubic to tetragonal is observed, with a rather broad coexistence region (~20 GPa to ~30 GPa) that corresponds to theT_Nplateau. Mössbauer spectroscopy measurements show a clear valence transition from approximately 50:50 Eu²⁺:Eu³⁺to fully Eu³⁺at ~28 GPa, consistent with the vanishing of the magnetic order at the same pressure. X-ray absorption data show a transition to a fully trivalent state at a similar pressure. Our results show that pressure first greatly enhancesT_N, most likely via enhanced hybridization between the Eu 4fstates and the conduction band, and then, second, causes a structural phase transition that coincides with the conversion of the europium to a fully trivalent state.

 

Partially observable Markov decision processes (POMDPs) provide a flexible representation for real-world decision and control problems. However, POMDPs are notoriously difficult to solve, especially when the state and observation spaces are continuous or hybrid, which is often the case for physical systems. While recent online sampling-based POMDP algorithms that plan with observation likelihood weighting have shown practical effectiveness, a general theory characterizing the approximation error of the particle filtering techniques that these algorithms use has not previously been proposed. Our main contribution is bounding the error between any POMDP and its corresponding finite sample particle belief MDP (PB-MDP) approximation. This fundamental bridge between PB-MDPs and POMDPs allows us to adapt any sampling-based MDP algorithm to a POMDP by solving the corresponding particle belief MDP, thereby extending the convergence guarantees of the MDP algorithm to the POMDP. Practically, this is implemented by using the particle filter belief transition model as the generative model for the MDP solver. While this requires access to the observation density model from the POMDP, it only increases the transition sampling complexity of the MDP solver by a factor of O(C), where C is the number of particles. Thus, when combined with sparse sampling MDP algorithms, this approach can yield algorithms for POMDPs that have no direct theoretical dependence on the size of the state and observation spaces. In addition to our theoretical contribution, we perform five numerical experiments on benchmark POMDPs to demonstrate that a simple MDP algorithm adapted using PB-MDP approximation, Sparse-PFT, achieves performance competitive with other leading continuous observation POMDP solvers.

 

Abstract Aggregative multicellularity relies on cooperation among formerly independent cells to form a multicellular body. Previous work with Dictyostelium discoideum showed that experimental evolution under low relatedness profoundly decreased cooperation, as indicated by the loss of fruiting body formation in many clones and an increase of cheaters that contribute proportionally more to spores than to the dead stalk. Using whole-genome sequencing and variant analysis of these lines, we identified 38 single nucleotide polymorphisms in 29 genes. Each gene had 1 variant except for grlG (encoding a G protein-coupled receptor), which had 10 unique SNPs and 5 structural variants. Variants in the 5′ half of grlG—the region encoding the signal peptide and the extracellular binding domain—were significantly associated with the loss of fruiting body formation; the association was not significant in the 3′ half of the gene. These results suggest that the loss of grlG was adaptive under low relatedness and that at least the 5′ half of the gene is important for cooperation and multicellular development. This is surprising given some previous evidence that grlG encodes a folate receptor involved in predation, which occurs only during the single-celled stage. However, non-fruiting mutants showed little increase in a parallel evolution experiment where the multicellular stage was prevented from happening. This shows that non-fruiting mutants are not generally selected by any predation advantage but rather by something—likely cheating—during the multicellular stage. 

Prochlorococcusis an abundant photosynthetic bacterium in the open ocean, where nitrogen (N) often limits phytoplankton growth. In the low-light-adapted LLI clade ofProchlorococcus, nearly all cells can assimilate nitrite (NO₂⁻), with a subset capable of assimilating nitrate (NO₃⁻). LLI cells are maximally abundant near the primary NO₂⁻maximum layer, an oceanographic feature that may, in part, be due to incomplete assimilatory NO₃⁻reduction and subsequent NO₂⁻release by phytoplankton. We hypothesized that someProchlorococcusexhibit incomplete assimilatory NO₃⁻reduction and examined NO₂⁻accumulation in cultures of threeProchlorococcusstrains (MIT0915, MIT0917, and SB) and twoSynechococcusstrains (WH8102 and WH7803). Only MIT0917 and SB accumulated external NO₂⁻during growth on NO₃⁻. Approximately 20–30% of the NO₃⁻transported into the cell by MIT0917 was released as NO₂⁻, with the rest assimilated into biomass. We further observed that co-cultures using NO₃⁻as the sole N source could be established for MIT0917 andProchlorococcusstrain MIT1214 that can assimilate NO₂⁻but not NO₃⁻. In these co-cultures, the NO₂⁻released by MIT0917 is efficiently consumed by its partner strain, MIT1214. Our findings highlight the potential for emergent metabolic partnerships that are mediated by the production and consumption of N cycle intermediates withinProchlorococcuspopulations.

Earth’s biogeochemical cycles are substantially driven by microorganisms and their interactions. Given that N often limits marine photosynthesis, we investigated the potential for N cross-feeding within populations ofProchlorococcus, the numerically dominant photosynthetic cell in the subtropical open ocean. In laboratory cultures, someProchlorococcuscells release extracellular NO₂⁻during growth on NO₃⁻. In the wild,Prochlorococcuspopulations are composed of multiple functional types, including those that cannot use NO₃⁻but can still assimilate NO₂⁻. We show that metabolic dependencies arise whenProchlorococcusstrains with complementary NO₂⁻production and consumption phenotypes are grown together on NO₃⁻. These findings demonstrate the potential for emergent metabolic partnerships, possibly modulating ocean nutrient gradients, that are mediated by cross-feeding of N cycle intermediates.

 

The social amoeba Dictyostelium discoideum engages in a complex relationship with bacterial endosymbionts in the genus Paraburkholderia, which can benefit their host by imbuing it with the ability to carry prey bacteria throughout its life cycle. The relationship between D. discoideum and Paraburkholderia has been shown to take place across many strains and a large geographical area, but little is known about Paraburkholderia’s potential interaction with other dictyostelid species. We explore the ability of three Paraburkholderia species to stably infect and induce bacterial carriage in other dictyostelid hosts. We found that all three Paraburkholderia species successfully infected and induced carriage in seven species of Dictyostelium hosts. While the overall behaviour was qualitatively similar to that previously observed in infections of D. discoideum, differences in the outcomes of different host/symbiont combinations suggest a degree of specialization between partners. Paraburkholderia was unable to maintain a stable association with the more distantly related host Polysphondylium violaceum. Our results suggest that the mechanisms and evolutionary history of Paraburkholderia’s symbiotic relationships may be general within Dictyostelium hosts, but not so general that it can associate with hosts of other genera. Our work further develops an emerging model system for the study of symbiosis in microbes.