To employ a reduced-order cardiovascular model as a digital twin for personalised medicine, it is essential to understand how uncertainties in the model's input parameters affect its outputs. The aim is to identify a set of input parameters that can serve as clinical biomarkers, providing insight into a patient's physiological state. Given the challenge of finding useful clinical data, careful consideration must be given to the experimental design used to acquire patient-specific input parameters. Model sloppiness-where numerous parameter combinations have minimal impact on model predictions, whilst only a few parameters significantly influence outcomes-is a critical concept in this context. In this paper, we conduct the first quantification of a cardiovascular system's sloppiness to elucidate the structure of the input parameter space. By utilising Sobol indices and examining various synthetic cardiovascular measures with increasing invasiveness, we uncover how the personalisation process and the cardiovascular system's sloppiness are contingent upon the chosen experimental design. Our findings reveal that continuous clinical measures induce system sloppiness and increase the number of personalisable biomarkers, whereas discrete clinical measurements produce a non-sloppy system with a reduced number of biomarkers. This study underscores the necessity for careful consideration of available clinical data as differing measurement sets can significantly impact model personalisation.
Computational modelling of biological processes poses multiple challenges in each stage of the modelling exercise. Some significant challenges include identifiability, precisely estimating parameters from limited data, informative experiments and anisotropic sensitivity in the parameter space. One of these challenges' crucial but inconspicuous sources is the possible presence of large regions in the parameter space over which model predictions are nearly identical. This property, known as sloppiness, has been reasonably well-addressed in the past decade, studying its possible impacts and remedies. However, certain critical unanswered questions concerning sloppiness, particularly related to its quantification and practical implications in various stages of system identification, still prevail. In this work, we systematically examine sloppiness at a fundamental level and formalise two new theoretical definitions of sloppiness. Using the proposed definitions, we establish a mathematical relationship between the parameter estimates' precision and sloppiness in linear predictors. Further, we develop a novel computational method and a visual tool to assess the goodness of a model around a point in parameter space by identifying local structural identifiability and sloppiness and finding the most sensitive and least sensitive parameters for non-infinitesimal perturbations. We demonstrate the working of our method in benchmark systems biology models of various complexities. The pharmacokinetic HIV infection model analysis identified a new set of biologically relevant parameters that can be used to control the free virus in an active HIV infection.
Recent experiments have shown that training trajectories of multiple deep neural networks with different architectures, optimization algorithms, hyperparameter settings, and regularization methods evolve on a remarkably low-dimensional "hyperribbon-like" manifold in the space of probability distributions. Inspired by the similarities in the training trajectories of deep networks and linear networks, we analytically characterize this phenomenon for the latter. We show, using tools in dynamical systems theory, that the geometry of this low-dimensional manifold is controlled by (i) the decay rate of the eigenvalues of the input correlation matrix of the training data, (ii) the relative scale of the ground-truth output to the weights at the beginning of training, and (iii) the number of steps of gradient descent. By analytically computing and bounding the contributions of these quantities, we characterize phase boundaries of the region where hyperribbons are to be expected. We also extend our analysis to kernel machines and linear models that are trained with stochastic gradient descent.
We amend and extend the Chiarella model of financial markets to deal with arbitrary drift in long-term value in a consistent way. This allows us to improve upon existing calibration schemes, opening the possibility of calibrating individual monthly time series instead of classes of time series. The technique is employed on spot prices of four asset classes from ca. 1800 onward (stock indices, bonds, commodities, currencies). The so-called fundamental value is a direct output of the calibration, which allows us to (a) quantify the amount of excess volatility in these markets, which we find to be large (e.g. a factor [Formula: see text] for stock indices) and consistent with previous estimates; and (b) determine the distribution of mispricings (i.e. the log-difference between market price and value), which we find in many cases to be bimodal. Both findings are strongly at odds with the Efficient Market Hypothesis. We also study in detail the 'sloppiness' of the calibration, that is, the directions in parameter space that are weakly constrained by data. The main conclusions of our study are remarkably consistent across different asset classes, and reinforce the hypothesis that the medium-term fate of financial markets is determined by a tug-of-war between trend followers and fundamentalists.
RAFs initiate the cascade leading to activation of the extracellular signal-regulated kinases (ERKs). In a substantial fraction of cancer cells, RAFs are the least abundant pathway proteins between receptor tyrosine kinases and ERKs. In some cases, active RAF kinases are present at the plasma membrane at just hundreds of copies per cell, but the consequences of such limited RAF abundance are unclear. By developing continuum and stochastic computational models of the epidermal growth factor receptor (EGFR)-ERK pathway, we showed that low RAF abundance creates signaling bottlenecks between receptor tyrosine kinases and ERK with a potential for stochastic RAF dynamics that can propagate especially to low-abundance downstream pathway proteins. RAF bottlenecks were also predicted to impede ERK activation by oncogenic RAS mutants. Advanced parameter sensitivity and sloppiness analyses identified RAS activation and RAS-RAF interactions as strong determinants of signaling in low-RAF settings and revealed an efficient model fitting approach. This work provides quantitative insight into a common, but unexplored, regime for EGFR-ERK signaling and a systematic approach to develop and characterize dynamic models of receptor-mediated signaling.
The individual functions of most iron-containing species in Saccharomyces cerevisiae are fairly-well understood, but less is known regarding how they function collectively as a unified system. Here, an ODE-based kinetic cell model was developed to reveal system's-level behavior of iron metabolism. The dimensionally-accurate in silico cell was divided into 5 compartments. It contained 80 components that engaged in 169 reactions. The cell grew on nutrients IRON, CARBON and OXYGEN. All major iron-related processes were represented including the biosynthesis and metallation of iron-containing proteins, trafficking of labile iron pools, homeostatic regulation, respiration, the TCA cycle, iron-sulfur-cluster and heme biosynthesis, the synthesis of DNA, phospholipids, amino acids, and nucleotide triphosphates, and reactions involving oxygen and reactive-oxygen-species. Iron and carbon were conserved in reaction stoichiometries. The time-dependent model was solved using the Basic Pathways approach, despite limited kinetic information. Once regulated appropriately, the system could withstand perturbations in component concentrations by returning to its original steady-state. It responded to changes in nutrient iron and oxygen concentrations and to changes in rate-constants, yielding altered sets of steady-state component concentrations. The latter type of perturbation is tantamount to altering the expression level of a gene. This ability offers the potential to explain phenotypic changes of genetic mutations on the mechanistic molecular level. The model included all established iron-related cellular processes (albeit in combined forms), and a highly interrelated reaction network reflecting a mutually autocatalytic system. Steady-state iron concentrations in the cell, organelles, and components were reasonably near to those observed/estimated experimentally.
In the VP ellipsis sentence "Bill liked himself, and John did too," the second conjunct "John did too" can be interpreted in two ways: either as "John liked Bill too" (strict interpretation) or "John liked himself too" (sloppy interpretation). Previous research has yielded inconclusive findings regarding which interpretation is preferred during real-time sentence processing. To investigate potential causes of these mixed findings, we conducted three reading experiments (Experiments 1-3) testing whether the inconclusive results could be attributed to (i) insufficient statistical power, (ii) limitations in experimental materials, or (iii) the influence of processing depth. We also examined whether form mismatches (e.g., "herself/himself") affect the resolution of VP ellipsis ambiguities, as in "Mary liked herself, and John did [ellipsis like himself] too". Experiment 1, with a large sample size (360 participants), showed no clear preference for either interpretation. Experiment 2, using revised materials, revealed a preference for the strict interpretation. Experiment 3 included comprehension questions targeting the interpretation of the elided material to assess the influence of processing depth. Results indicated that participants who answered correctly preferred the strict interpretation, while those who answered incorrectly preferred the sloppy interpretation. Across all three experiments, form mismatches produced no clear effect. We propose an account of the mechanisms underlying the real-time resolution of VP ellipsis ambiguities and discuss how processing depth influences preferences for the strict or sloppy interpretation.
To study the effect of a behavioral intervention, it should be compared to a control or an existing treatment in an intervention study. There exist many guidelines in the literature about the design and analysis of intervention studies, including recommendations for a priori sample size determination. The vast majority of these guidelines are based on the framework of null hypothesis significance testing, where a p-value is compared to a user-selected type I error rate to determine whether an effect is significant or not. This approach has received severe criticism over the past decades as it has resulted in publication bias, sloppy science, and fraud. The Bayesian approach to hypothesis testing has been developed to overcome some of these drawbacks. The Bayes factor quantifies the relative support in the data for one hypothesis over another hypothesis. The hypotheses do not necessarily have to include a null hypothesis and can be formulated based on observations, findings in the literature, or an expert's opinion. Posterior Model Probabilities, which are a function of the Bayes Factor, can be used to compare a set of hypotheses to one another and select the one most supported by the data. In this paper, we summarize the shortcomings of null hypothesis significance testing, introduce the Bayes factor and Posterior Model Probabilities, explain how they are calculated, and how they are interpreted. We also focus on a priori sample size determination in the Bayesian hypothesis testing framework. We introduce a criterion for sample size determination and a procedure to find the required sample size. We illustrate our methodology using a cluster randomized trial on the effectiveness of an online training in improving primary care doctors' competency in brief tobacco interventions. All analyses are done in R, and we provide the dataset and R syntax for straightforward replication.
Explaining individual differences in cognitive abilities requires both identifying brain parameters that vary across individuals and understanding how brain networks are recruited for specific tasks. Typically, task performance relies on the integration and segregation of functional subnetworks, often captured by parameters like regional excitability and connectivity. Yet, the high dimensionality of these parameters hinders pinpointing their functional relevance. Here, we apply stiff-sloppy analysis to human brain data, revealing that certain subtle parameter combinations ("stiff dimensions") powerfully influence neural activity during task processing, whereas others ("sloppy dimensions") vary more extensively but exert minimal impact. Using a pairwise maximum entropy model of task fMRI, we show that even small deviations in stiff dimensions-derived through Fisher Information Matrix analysis-govern the dynamic interplay of segregation and integration between the default mode network (DMN) and a working memory network (WMN). Crucially, separating a 0-back task (vigilant attention) from a 2-back task (working memory updating) uncovers partially distinct stiff dimensions predicting performance in each condition, along with a global DMN-WMN segregation shared across both tasks. Altogether, stiff-sloppy analysis challenges the conventional focus on large parameter variability by highlighting these subtle yet functionally decisive parameter combinations.
Metagenomic data has significantly advanced microbiome research by employing ecological models, particularly in personalized medicine. The generalized Lotka-Volterra (gLV) model is commonly used to understand microbial interactions and predict ecosystem dynamics. However, gLV models often fail to capture complex interactions, especially when data are limited or noisy. This study critically assesses the effectiveness of gLV and similar models using Bayesian inference and a model reduction method based on information theory. We found that ecological data often leads to non-interpretability and overfitting due to limited information, noisy data and parameter sloppiness. Our results highlight the need for simpler models that align with the available data and propose a distribution-based approach to better capture ecosystem diversity, stability and competition. These findings challenge current bottom-up ecological modelling practices and aim to shift the focus towards a statistical mechanics view of ecology based on distributions of parameters.
αβ T cell receptor (TCR) recognition of peptide-MHC complexes lies at the core of adaptive immunity, balancing specificity and cross-reactivity to facilitate effective antigen discrimination. Early structural studies established basic frameworks helpful for understanding and contextualizing TCR recognition and features such as peptide specificity and MHC restriction. However, the growing TCR structural database and studies launched from structural work continue to reveal exceptions to common assumptions and simplifications derived from earlier work. Here we explore our evolving understanding of TCR recognition, illustrating how structural and biophysical investigations regularly uncover complex phenomena that push against paradigms and expand our understanding of how TCRs bind to and discriminate between peptide/MHC complexes. We discuss the implications of these findings for basic, translational, and predictive immunology, including the challenges in accounting for the inherent adaptability, flexibility, and occasional biophysical sloppiness that characterize TCR recognition.
In her book Why trust Science?, Naomi Oreskes examines the question of what it means to say that "science corrects itself", highlighting the importance of the social process of science and specifically the importance of scientists challenging each other in the pursuit of truth. In a recent preprint, a colleague and I did exactly that, reviewing a corpus of work by Australian neuroethologist Mandyam Srinivasan and identifying numerous problems across ten of his papers, including several instances of identical data being reported for different experiments. In a recent editorial, Eric Warrant dismisses our critiques of Srinivasan's work as "sloppiness all of us are capable of", and instead focuses on attacking us, sometimes conflating criticisms of others of Srinivasan's work with ours. Here I review his claims and argue for the importance of truth in the advancement of science.
Malignant bowel obstruction (MBO) presents with multiple symptoms. The 4-step BOUNCED diet educates patients to self-manage oral intake according to symptoms. It includes clear fluids, thin liquids, purée and soft, sloppy foods, which are low in fibre. This mixed methods single-arm feasibility study aimed to establish if the diet could reduce MBO symptoms in patients with inoperable colorectal and gynaecological malignancies. The secondary objectives were to investigate if it was easily followed, improved quality of life (QOL) and reduced hospital admissions. Patients able to tolerate an oral diet with one or more symptoms (pain, bloating, early satiety, nausea and vomiting) were eligible. Following informed consent, an oncology dietitian took a diet history and determined which step of the diet they needed to follow using a detailed patient information leaflet. Patients remained on the trial for 28 days. Symptom and QOL data were collected on Days 1 and 28 using the Memorial Symptom Assessment Scale and the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ-C30) questionnaires. Thirty patients (24 female, 6 male) aged 18-85 years consented from March 2021 to November 2022. Twenty-six participants completed the trial and 25 found the diet very easy or easy to understand. There was a significant reduction in pain from 96% on Day 1 to 63% on Day 28 (p = 0.004). The mean increase of EORTC global health status/QOL was 23.5 points, 95% CI (12.4-32.5) (p ≤ 0.001). There was a significant difference between number of admissions (p = 0.018) and bed days (p = 0.004) in the 28 days prior to consent compared to the trial period. A modified consistency low-fibre diet is easily followed, may reduce symptoms of MBO, admissions to hospital and improve QOL.
Leaf litter decomposition is a vital ecosystem process in which macroinvertebrate-shredders produce substantial amounts of fine particulate organic matter (FPOM) via sloppy feeding and defecation, creating a substratum and substrate for microbial assemblages. However, microbial communities colonizing the shredder-produced FPOM are understudied compared to those in streams and on original leaves. Here, we investigated the bacterial community composition on shredder-produced FPOM in a laboratory experiment. We fed alder, beech, and maple leaves conditioned under oxic or anoxic conditions to Sericostoma (Insecta: Trichoptera) larvae. We collected shredded leaf particles and faecal pellets as shredder-produced FPOM at different times and examined their microbial communities using 16S rRNA amplicon sequencing. We hypothesized that shredder-produced FPOM types harbor diverse, distinct, and specialized microbial taxa in response to leaf species and conditioning. We found significantly higher alpha diversity on shredded leaves compared to faecal pellets. Microbial communities on faecal pellets differed from initial leaf communities and with anoxic and oxic conditioning. Bacterial communities developing on leaves were dominated by common leaf decomposers including Flavobacterium and Pseudomonas whereas faecal pellets harbored gut bacterial taxa including Acinetobacter and Carnobacterium. These results underline the importance of conditioning and shredder activity in shaping FPOM-attached bacterial communities, increasing bacterial diversity in stream ecosystems.
Present study was conducted in the Northern Ethiopia with the aim to identify and map potential groundwater sites using the state art and science of geo-spatial technology techniques. This method used to prepare the spatial factors derived from various sources including satellite images, existing thematic maps and to develop the model. The data employed include climatic and biophysical data like rainfall, geology, soil and land use land cover. Computation of the parameters weight impact was calculated and accordingly, geology, lineament density and geomorphology were found the most determinant factors influencing groundwater potential (GWP) occurrence; whereas, land use land cover, soil and rainfall were perceived the least significant elements. The final GWP map was developed through the integration of the selected eight variables into a system of weight overlay modeling with the ArcGIS interface environment. The result GWP map was generated in to five suitability classes as very good (14%), good (16%), moderate (24%), and the rest (47%) is under poor and very poor classes respectively. Very good potential sites were geographically lied in the plateau, flat to gentle sloppy, while areas under good classes were situated on moderately sloppy of Alaje formation under undulated surface of the study area. Similarly, the moderate potential areas were found in undulated surface of Alaje and Ashengie formations. In contrast, the poor and very poor GWP areas was lied beneath Aiba and small amount in Alaje formation in the steeply slope. Generally, the spatial distribution of the groundwater occurrence and movement in the area is mainly regulated by geology, geomorphology and lineament density. Finally, the performance of the groundwater potential of the model was cross validated by area under curve of (AUC) of receiver operating characteristic (ROC) and was found 85% accurate. Furthermore studies are demanded considering high quality data supported with intensive field measurement for better accuracy and outcome.
This paper introduces the concept of 'hyper-ambition' in academia as a contributing factor to what has been termed a 'replication crisis' across some sciences. The replication crisis is an umbrella term that covers a range of 'questionable research practices', from sloppy reporting to fraud. There are already many proposals to address questionable research practices, some of which focus on the values, norms, and motivations of researchers and institutes, and suggest measures to promote research integrity. Yet it is not easy to promote integrity in hyper-competitive academic environments that value high levels of ambition. I argue that in such contexts, it is as likely that a kind of hyper-ambition is fostered that (inadvertently or otherwise) prioritises individual success above all, including to the detriment of scientific quality. In addition, efforts to promote values like integrity falter because they rely on sufficient uniformity in motivations or tendencies. Codes and guidance promoting integrity are, however, likely to influence those for whom such values are not optional, while others simply find ways around them. To demonstrate this I offer a thought experiment in which we consider the imaginary working situations of two ordinary academics. I conclude that tackling questionable research practices in the light of the replication crisis requires robust 'top down' measures that expect and accommodate a broader range of academic values, motivations, and tendencies, while challenging those that help to promote hyper-ambition.
Logophoric pronouns in West African languages occur in attitude environments and are anaphorically linked to an attitude holder in a superordinate clause. This has motivated theorists to treat logophoric pronouns semantically as an obligatorily bound variable (bound from the edge of an embedded complement clause). Culy (Linguistics 32:1055-1094, 1994) and Bimpeh and Sode (in Proceedings of TripleA, vol. 6 pp. 1-16, 2021), however, point out that logophoric pronouns in Ewe do not behave like obligatorily bound variables, allowing both sloppy (bound) and strict (non-bound) readings in focus contexts involving 'only' and ellipsis. We strengthen this line of criticism by providing novel cross-linguistic data that indicate that logophoric pronouns in Ewe, Igbo and Yoruba support strict readings. We offer an alternative formal account to existing approaches that builds on Bimpeh et al. (in Proceedings of the 40th WCCFL, pp. 1-10, 2024) and can capture both strict and sloppy interpretations, while preserving the requirement that a logophoric pronoun be anaphoric to an attitude holder. The main novelty involves decomposition of logophoric pronouns into two syntactic components at LF-a variable that can in principle be free and refer strictly, and a semantic presuppositional feature log that can be ignored in ellipsis and focus sites, following similar ideas in the literature on pronominal features (Sauerland in Proceedings of SALT 23, pp. 156-173, 2013). Our analysis implies that in terms of their syntactic and semantic make up, logophors are essentially no different from other pronouns, consisting of a referential index plus semantic features. The online version contains supplementary material available at 10.1007/s11050-025-09242-x.
Marine plankton communities consist of numerous species, and their composition and physiological states are closely linked to ecosystem functions. Understanding biogeochemical cycles requires measuring taxon-specific mortality due to viral lysis, sloppy feeding, and other mechanical stresses as the dissolved organic matter released contributes to rapid nutrient recycling and long-term carbon sequestration following microbial transformation. To examine the lytic cell death of marine microeukaryotes, we applied a quantitative and comprehensive analysis of the dissolved constituents of seawater using the Mortality by Ribosomal Sequencing (MoRS) method. Our experimental pipeline successfully recovered 83% of cell-free rRNA. A higher number of protist phylotypes was significantly lysed in the mesopelagic zone than in the surface ecosystems, indicating that the mesopelagic zone is a potential hotspot for eukaryotic cell lysis. Many protist lineages, including phytoplankton such as haptophytes, were less susceptible to cell lysis in the epipelagic layer yet were actively lysed in the mesopelagic zone. Notably, over 86% of the significantly lysed species in the mesopelagic layer had a habitat preference for the epipelagic layer. These findings suggest that sinking from the surface and lysis in the mesopelagic may represent prevalent dynamics for various eukaryotes.
Exploring the degree to which phenotypic variation, influenced by intrinsic nonlinear biological mechanisms, can be accurately captured using statistical methods is essential for advancing our comprehension of complex biological systems and predicting their functionality. Here, we examine this issue by combining a computational model of gene regulation networks with a linear additive prediction model, akin to polygenic scores utilized in genetic analyses. Inspired by the variational framework of quantitative genetics, we create a population of individual networks possessing identical topology yet showcasing diversity in regulatory strengths. By discerning which regulatory connections determine the prediction of phenotypes, we contextualize our findings within the framework of core and peripheral causal determinants, as proposed by the omnigenic model of complex traits. We establish connections between our results and concepts such as global sensitivity and local stability in dynamical systems, alongside the notion of sloppy parameters in biological models. Furthermore, we explore the implications of our investigation for the broader discourse surrounding the role of epistatic interactions in the prediction of complex phenotypes.
Many biological systems perform close to their physical limits, but promoting this optimality to a general principle seems to require implausibly fine tuning of parameters. Using examples from a wide range of systems, we show that this intuition is wrong. Near an optimum, functional performance depends on parameters in a "sloppy" way, with some combinations of parameters being only weakly constrained. Absent any other constraints, this predicts that we should observe widely varying parameters, and we make this precise: the entropy in parameter space can be extensive even if performance on average is very close to optimal. This removes a major objection to optimization as a general principle, and rationalizes the observed variability.