Generating a multimodal artificial intelligence model to differentiate benign and malignant follicular neoplasms of the thyroid: A proof-of-concept study

BACKGROUND

Machine learning has been increasingly used to develop algorithms that can improve medical diagnostics and prognostication and has shown promise in improving the classification of thyroid ultrasound images. This proof-of-concept study aims to develop a multimodal machine-learning model to classify follicular carcinoma from adenoma.

METHODS

This is a retrospective study of patients with follicular adenoma or carcinoma at a single institution between 2010 and 2022. Demographics, imaging, and perioperative variables were collected. The region of interest was annotated on ultrasound and used to perform radiomics analysis. Imaging features and clinical variables were then used to create a random forest classifier to predict malignancy. Leave-one-out cross-validation was conducted to evaluate classifier performance using the area under the receiver operating characteristic curve.

RESULTS

Patients with follicular adenomas (n = 7) and carcinomas (n = 11) with complete imaging and perioperative data were included. A total of 910 features were extracted from each image. The t-distributed stochastic neighbor embedding method reduced the dimension to 2 primary represented components. The random forest classifier achieved an area under the receiver operating characteristic curve of 0.76 (clinical only), 0.29 (image only), and 0.79 (multimodal data).

CONCLUSION

Our multimodal machine learning model demonstrates promising results in classifying follicular carcinoma from adenoma. This approach can potentially be applied in future studies to generate models for preoperative differentiation of follicular thyroid neoplasms.

Classifying contaminated cell cultures using time series features

We examine the use of time series data, derived from Electric Cell-substrate Impedance Sensing (ECIS), to differentiate between standard mammalian cell cultures and those infected with a mycoplasma organism. With the goal of easy visualization and interpretation, we perform low-dimensional feature-based classification, extracting application-relevant features from the ECIS time courses. We can achieve very high classification accuracy using only two features, which depend on the cell line under examination. Initial results also show the existence of experimental variation between plates and suggest types of features that may prove more robust to such variation. Our paper is the first to perform a broad examination of ECIS time course features in the context of detecting contamination; to combine different types of features to achieve classification accuracy while preserving interpretability; and to describe and suggest possibilities for ameliorating plate-to-plate variation.

Addressing the Embeddability Problem in Transition Rate Estimation

Markov State Models (MSM) and related techniques have gained significant traction as a tool for analyzing and guiding molecular dynamics (MD) simulations due to their ability to extract structural, thermodynamic, and kinetic information on proteins using computationally feasible MD simulations. The MSM analysis often relies on spectral decomposition of empirically generated transition matrices. This work discusses an alternative approach for extracting the thermodynamic and kinetic information from the so-called rate/generator matrix rather than the transition matrix. Although the rate matrix itself is built from the empirical transition matrix, it provides an alternative approach for estimating both thermodynamic and kinetic quantities, particularly in diffusive processes. A fundamental issue with this approach is known as the embeddability problem. The key contribution of this work is the introduction of a novel method to address the embeddability problem as well as the collection and utilization of existing algorithms previously used in the literature. The algorithms are tested on data from a one-dimensional toy model to show the workings of these methods and discuss the robustness of each method in dependence of lag time and trajectory length.

Group linear non-Gaussian component analysis with applications to neuroimaging

Independent component analysis (ICA) is an unsupervised learning method popular in functional magnetic resonance imaging (fMRI). Group ICA has been used to search for biomarkers in neurological disorders including autism spectrum disorder and dementia. However, current methods use a principal component analysis (PCA) step that may remove low-variance features. Linear non-Gaussian component analysis (LNGCA) enables simultaneous dimension reduction and feature estimation including low-variance features in single-subject fMRI. A group LNGCA model is proposed to extract group components shared by more than one subject. Unlike group ICA methods, this novel approach also estimates individual (subject-specific) components orthogonal to the group components. To determine the total number of components in each subject, a parametric resampling test is proposed that samples spatially correlated Gaussian noise to match the spatial dependence observed in data. In simulations, estimated group components achieve higher accuracy compared to group ICA. The method is applied to a resting-state fMRI study on autism spectrum disorder in 342 children (252 typically developing, 90 with autism), where the group signals include resting-state networks. The discovered group components appear to exhibit different levels of temporal engagement in autism versus typically developing children, as revealed using group LNGCA. This novel approach to matrix decomposition is a promising direction for feature detection in neuroimaging.

Exploring Blob Detection to Determine Atomic Column Positions and Intensities in Time-Resolved TEM Images with Ultra-Low Signal-to-Noise

Spatially resolved in situ transmission electron microscopy (TEM), equipped with direct electron detection systems, is a suitable technique to record information about the atom-scale dynamics with millisecond temporal resolution from materials. However, characterizing dynamics or fluxional behavior requires processing short time exposure images which usually have severely degraded signal-to-noise ratios. The poor signal-to-noise associated with high temporal resolution makes it challenging to determine the position and intensity of atomic columns in materials undergoing structural dynamics. To address this challenge, we propose a noise-robust, processing approach based on blob detection, which has been previously established for identifying objects in images in the community of computer vision. In particular, a blob detection algorithm has been tailored to deal with noisy TEM image series from nanoparticle systems. In the presence of high noise content, our blob detection approach is demonstrated to outperform the results of other algorithms, enabling the determination of atomic column position and its intensity with a higher degree of precision.

Sparse Identification and Estimation of Large-Scale Vector AutoRegressive Moving Averages

The Vector AutoRegressive Moving Average (VARMA) model is fundamental to the theory of multivariate time series; however, identifiability issues have led practitioners to abandon it in favor of the simpler but more restrictive Vector AutoRegressive (VAR) model. We narrow this gap with a new optimization-based approach to VARMA identification built upon the principle of parsimony. Among all equivalent data-generating models, we use convex optimization to seek the parameterization that is simplest in a certain sense. A user-specified strongly convex penalty is used to measure model simplicity, and that same penalty is then used to define an estimator that can be efficiently computed. We establish consistency of our estimators in a double-asymptotic regime. Our non-asymptotic error bound analysis accommodates both model specification and parameter estimation steps, a feature that is crucial for studying large-scale VARMA algorithms. Our analysis also provides new results on penalized estimation of infinite-order VAR, and elastic net regression under a singular covariance structure of regressors, which may be of independent interest. We illustrate the advantage of our method over VAR alternatives on three real data examples.

Critical Risk Indicators (CRIs) for the electric power grid: a survey and discussion of interconnected effects

The electric power grid is a critical societal resource connecting multiple infrastructural domains such as agriculture, transportation, and manufacturing. The electrical grid as an infrastructure is shaped by human activity and public policy in terms of demand and supply requirements. Further, the grid is subject to changes and stresses due to diverse factors including solar weather, climate, hydrology, and ecology. The emerging interconnected and complex network dependencies make such interactions increasingly dynamic, posing novel risks, and presenting new challenges to manage the coupled human–natural system. This paper provides a survey of models and methods that seek to explore the significant interconnected impact of the electric power grid and interdependent domains. We also provide relevant critical risk indicators (CRIs) across diverse domains that may be used to assess risks to electric grid reliability, including climate, ecology, hydrology, finance, space weather, and agriculture. We discuss the convergence of indicators from individual domains to explore possible systemic risk, i.e., holistic risk arising from cross-domain interconnections. Further, we propose a compositional approach to risk assessment that incorporates diverse domain expertise and information, data science, and computer science to identify domain-specific CRIs and their union in systemic risk indicators. Our study provides an important first step towards data-driven analysis and predictive modeling of risks in interconnected human–natural systems.

PyXtal_FF: a python library for automated force field generation

We present PyXtal_FF—a package based on Python programming language—for developing machine learning potentials (MLPs). The aim of PyXtal_FF is to promote the application of atomistic simulations through providing several choices of atom-centered descriptors and machine learning regressions in one platform. Based on the given choice of descriptors (including the atom-centered symmetry functions, embedded atom density, SO4 bispectrum, and smooth SO3 power spectrum), PyXtal_FF can train MLPs with either generalized linear regression or neural network models, by simultaneously minimizing the errors of energy/forces/stress tensors in comparison with the data from ab-initio simulations. The trained MLP model from PyXtal_FF is interfaced with the Atomic Simulation Environment (ASE) package, which allows different types of light-weight simulations such as geometry optimization, molecular dynamics simulation, and physical properties prediction. Finally, we will illustrate the performance of PyXtal_FF by applying it to investigate several material systems, including the bulk SiO2, high entropy alloy NbMoTaW, and elemental Pt for general purposes. Full documentation of PyXtal_FF is available at https://pyxtal-ff.readthedocs.io.

Social distancing merely stabilized COVID-19 in the US

Social distancing measures have been imposed across the United States in order to stem the spread of COVID‐19. We quantify the reduction in the doubling rate, by state, that is associated with this intervention. Using the earlier of K‐12 school closures and restaurant closures, by state, to define the start of the intervention, and considering daily confirmed cases through April 23, 2020, we find that social distancing is associated with a statistically‐significant (p < 0.01) reduction in the doubling rate for all states except for Nebraska, North Dakota, and South Dakota, when controlling for false discovery, with the doubling rate averaged across the states falling from 0.302 (0.285, 0.320) days⁻¹ to 0.010 (−0.007, 0.028) days⁻¹. However, we do not find that social distancing has made the spread subcritical. Instead, social distancing has merely stabilized the spread of the disease. We provide an illustration of our findings for each state, including estimates of the effective reproduction number, R, both with and without social distancing. We also discuss the policy implications of our findings.

Cell Line Classification Using Electric Cell-Substrate Impedance Sensing (ECIS)

We present new methods for cell line classification using multivariate time series bioimpedance data obtained from electric cell-substrate impedance sensing (ECIS) technology. The ECIS technology, which monitors the attachment and spreading of mammalian cells in real time through the collection of electrical impedance data, has historically been used to study one cell line at a time. However, we show that if applied to data from multiple cell lines, ECIS can be used to classify unknown or potentially mislabeled cells, factors which have previously been associated with the reproducibility crisis in the biological literature. We assess a range of approaches to this new problem, testing different classification methods and deriving a dictionary of 29 features to characterize ECIS data. Most notably, our analysis enriches the current field by making use of simultaneous multi-frequency ECIS data, where previous studies have focused on only one frequency; using classification methods to distinguish multiple cell lines, rather than simple statistical tests that compare only two cell lines; and assessing a range of features derived from ECIS data based on their classification performance. In classification tests on fifteen mammalian cell lines, we obtain very high out-of-sample predictive accuracy. These preliminary findings provide a baseline for future large-scale studies in this field.

Maternal use of thyroid hormone replacement therapy before, during, and after pregnancy: agreement between self-report and prescription records and group-based trajectory modeling of prescription patterns

Purpose

A reliable definition of exposure and knowledge about long-term medication patterns is important for drug safety studies during pregnancy. Few studies have investigated these measures for thyroid hormone replacement therapy (THRT). The purpose of this study was to 1) calculate the agreement between self-report and dispensed prescriptions of THRT and 2) classify women with similar adherence patterns to THRT into disjoint longitudinal trajectories.

Methods

Our analysis used data from the Norwegian Mother and Child Cohort Study (MoBa), a prospective population-based cohort study. MoBa was linked to prescription records from the Norwegian Prescription Database (NorPD). We estimated Cohen’s kappa coefficients (k) and approximate 95% CIs for agreement between self-report and prescription records for the 6-month period prior to pregnancy and for each pregnancy trimester. Using group-based trajectory models (GBTMs), we estimated adherence trajectories among women who self-reported and had a THRT prescription.

Results

There were 56,148 women in MoBa, who had both a record in NorPD and available prescription history up to 1 year prior to pregnancy. Of these, 1,171 (2.1%) self-reported and received a prescription for THRT. Agreement was “perfect” in the 6-month period prior to pregnancy (k=0.86; CI 0.85–0.88), in the first (k=0.83; CI 0.82–0.85) and in the second trimesters (k=0.89; CI 0.87–0.90), while this was moderate (k=0.57; CI 0.54–0.59) in the third trimester. Among the subset of the 1,171 women, we identified four disjoint GBTM adherence groups: Constant-High (50.2%), Constant-Medium (32.9%), Increasing-Medium (11.0%), and Decreasing-Low (5.8%).

Conclusion

Agreement between self-report and prescription records was high for THRT in the early pregnancy period. Based on our GBTM results, about one in two women with hypothyroidism had adequate adherence to prescribed THRT throughout pregnancy. Given the potential consequences, evidence of low adherence in 5.8% of pregnant women with hypothyroidism is of concern.

An evaluation of independent component analyses with an application to resting-state fMRI

We examine differences between independent component analyses (ICAs) arising from different assumptions, measures of dependence, and starting points of the algorithms. ICA is a popular method with diverse applications including artifact removal in electrophysiology data, feature extraction in microarray data, and identifying brain networks in functional magnetic resonance imaging (fMRI). ICA can be viewed as a generalization of principal component analysis (PCA) that takes into account higher-order cross-correlations. Whereas the PCA solution is unique, there are many ICA methods-whose solutions may differ. Infomax, FastICA, and JADE are commonly applied to fMRI studies, with FastICA being arguably the most popular. Hastie and Tibshirani (2003) demonstrated that ProDenICA outperformed FastICA in simulations with two components. We introduce the application of ProDenICA to simulations with more components and to fMRI data. ProDenICA was more accurate in simulations, and we identified differences between biologically meaningful ICs from ProDenICA versus other methods in the fMRI analysis. ICA methods require nonconvex optimization, yet current practices do not recognize the importance of, nor adequately address sensitivity to, initial values. We found that local optima led to dramatically different estimates in both simulations and group ICA of fMRI, and we provide evidence that the global optimum from ProDenICA is the best estimate. We applied a modification of the Hungarian (Kuhn-Munkres) algorithm to match ICs from multiple estimates, thereby gaining novel insights into how brain networks vary in their sensitivity to initial values and ICA method.

Asymmetric homologation of boronic esters bearing azido and silyloxy substituents

In the asymmetric homologation of boronic esters with a (dihalomethyl)lithium, substituents that can bind metal cations tend to interfere. Accordingly, we undertook the introduction of weakly basic oxygen and nitrogen substituents into boronic esters in order to maximize the efficiency of multistep syntheses utilizing this chemistry. Silyloxy boronic esters cannot be made efficiently by direct substitution, but a (hydroxymethyl)boronic ester has been silylated in the usual manner. Conversion of alpha-halo boronic esters to alpha-azido boronic esters has been carried out with sodium azide and a tetrabutylammonium salt as phase-transfer catalyst in a two-phase system with water and either nitromethane or ethyl acetate. These are safer solvents than the previously used dichloromethane, which can form an explosive byproduct with azide ion. Boronic esters containing silyloxy or alkoxy and azido substituents have been shown to react efficiently with (dihalomethyl)lithiums, resulting in efficient asymmetric insertion of the halomethyl group into the carbon-boron bond.

Inhibition of HIV-1 protease by a boron-modified polypeptide

Six boronated tetrapeptides with the carboxy moiety of phenylalanine replaced by dihydroxyboron were synthesized, and their activities against human immunodeficiency virus 1 (HIV-1) protease subsequently investigated. The sequences of these peptides were derived from HIV-1 protease substrates, which included the C-terminal part of the scissile bond (Phe-Pro) within the gag-pol polyprotein. Enzymatic studies showed that these compounds were competitive inhibitors of HIV-1 protease with K(i) values ranging from 5 to 18 microM when experiments were performed at high enzyme concentrations (above 5 x 10(-8) M); however, at low protease concentrations inhibition was due in part to an increase of the association constants of the protease subunits. Ac-Thr-Leu-Asn-PheB inhibited HIV-1 protease with a K(i) of 5 microM, whereas the non-boronated parental compound was inactive at concentrations up to 400 microM, which indicates the significance of boronation in enzyme inhibition. The boronated tetrapeptides were inhibitory to an HIV-1 protease variant that is resistant to several HIV-1 protease inhibitors. Finally, fluorescence analysis showed that the interactions between the boronated peptide Ac-Thr-Leu-Asn-PheB and HIV-1 protease resulted in a rapid decrease of fluorescence emission at 360 nm, which suggests the formation of a compound/enzyme complex. Boronated peptides may provide useful reagents for studying protease biochemistry and yield valuable information toward the development of protease dimerization inhibitors.

A highly enantioselective and diastereoselective synthesis of cyclobutanes via boronic esters

[formula: see text] Deprotonation of enantiopure (R,R)-1,2-dicyclohexyl-1,2-ethanediol 1-chloro-4-cyanobutylboronates 5 with LDA followed by treatment with anhydrous magnesium bromide yields (R)-(trans-2-cyanocyclobutyl)boronic esters 7 in high diastereomeric and enantiomeric purity. No cyclobutane formation has been observed in the absence of at least a catalytic amount of magnesium halide.

A highly enantioselective and diastereoselective synthesis of cyclobutanes via boronic esters

[formula: see text] Deprotonation of enantiopure (R,R)-1,2-dicyclohexyl-1,2-ethanediol 1-chloro-4-cyanobutylboronates 5 with LDA followed by treatment with anhydrous magnesium bromide yields (R)-(trans-2-cyanocyclobutyl)boronic esters 7 in high diastereomeric and enantiomeric purity. No cyclobutane formation has been observed in the absence of at least a catalytic amount of magnesium halide.

(1-Aminoethyl)boronic acid: a novel inhibitor for Bacillus stearothermophilus alanine racemase and Salmonella typhimurium D-alanine:D-alanine ligase (ADP-forming)

(1-Aminoethyl)boronic acid (Ala-B), an analogue of alanine in which a boronic acid group replaces the carboxyl group, has been synthesized and found to inhibit the first two enzymes, alanine racemase (from Bacillus stearothermophilus, EC 5.1.1.1) and D-alanine:D-alanine ligase (ADP-forming) (from Salmonella typhimurium, EC 6.3.2.4), of the D-alanine branch of bacterial peptidoglycan biosynthesis. In both cases, time-dependent, slow binding inhibition is observed due to the generation of long-lived, slowly dissociating complexes. Ala-B inhibits alanine racemase with a Ki of 20 mM and a kappa inact of 0.15-0.35 min-1. Time-dependent loss of activity is paralleled by conversion of the 420-nm chromophore of initial bound PLP aldimine to a 324-nm absorbing species. On dilution of Ala-B, racemase activity is regained with a t1/2 of ca. 1 h. The D-Ala-D-Ala ligase also shows progressive inhibition by Ala-B provided ATP (but not AMP-PNP or AMP-PCP) is present. The presence of D-alanine along with ATP also leads to Ala-B-induced inactivation. Kinetic analysis suggests Ala-B can compete with D-alanine at either of the two D-alanine binding sites, and on inactivation with Ala-B, labeled D-alanine, and labeled ATP, the inactive enzyme has stoichiometric amounts of D-alanine, ADP, Pi, and Ala-B bound. The half-life of inactive enzyme complexes varied from approximately 2 h (without D-alanine) to 4.5 days (with D-alanine). No D-Ala-D-Ala-B dipeptide was detected.(ABSTRACT TRUNCATED AT 250 WORDS)

Benzamidomethaneboronic acid: Synthesis and inhibition of chymotrypsin

Benzamidomethaneboronic acid (2) has been synthesized unambiguously from the reaction of dibutyl iodomethaneboronate and N-lithiohexamethyldisilazane to form dibutyl [bis(trimethylsilyl)amino]methaneboronate (4), which was desilylated, benzoylated, and hydrolyzed to 2. It has been shown that 2 is a strong competitive inhibitor of alpha-chymotrypsin (Ki = 8.1 X 10(-6) M, pH 7.5). The reaction product from dibutyl iodomethaneboronate and sodiobenzamide, previously shown to be a potent inhibitor of chymotrypsin, was shown by this work to be O-linked isomer, benzimidoxy-methaneboronic acid (3). The pH-Ki profile over the pH range 6.5-9.5 was consistent with the formation of an enzyme-inhibitor complex which resembled the metastable tetrahedral reaction intermediates occurring during acylation and deacylation of chymotrypsin-catalyzed hydrolysis.