Assessing fairness in machine learning models: A study of racial bias using matched counterparts in mortality prediction for patients with chronic diseases.

OBJECTIVE: Existing approaches to fairness evaluation often overlook systematic differences in the social determinants of health, like demographics and socioeconomics, among comparison groups, potentially leading to inaccurate or even contradictory conclusions. This study aims to evaluate racial disparities in predicting mortality among patients with chronic diseases using a fairness detection method that considers systematic differences. METHODS: We created five datasets from Mass General Brigham's electronic health records (EHR), each focusing on a different chronic condition: congestive heart failure (CHF), chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), chronic liver disease (CLD), and dementia. For each dataset, we developed separate machine learning models to predict 1-year mortality and examined racial disparities by comparing prediction performances between Black and White individuals. We compared racial fairness evaluation between the overall Black and White individuals versus their counterparts who were Black and matched White individuals identified by propensity score matching, where the systematic differences were mitigated. RESULTS: We identified significant differences between Black and White individuals in age, gender, marital status, education level, smoking status, health insurance type, body mass index, and Charlson comorbidity index (p-value < 0.001). When examining matched Black and White subpopulations identified through propensity score matching, significant differences between particular covariates existed. We observed weaker significance levels in the CHF cohort for insurance type (p = 0.043), in the CKD cohort for insurance type (p = 0.005) and education level (p = 0.016), and in the dementia cohort for body mass index (p = 0.041); with no significant differences for other covariates. When examining mortality prediction models across the five study cohorts, we conducted a comparison of fairness evaluations before and after mitigating systematic differences. We revealed significant differences in the CHF cohort with p-values of 0.021 and 0.001 in terms of F1 measure and Sensitivity for the AdaBoost model, and p-values of 0.014 and 0.003 in terms of F1 measure and Sensitivity for the MLP model, respectively. DISCUSSION AND CONCLUSION: This study contributes to research on fairness assessment by focusing on the examination of systematic disparities and underscores the potential for revealing racial bias in machine learning models used in clinical settings.

Predicting the stereoselectivity of chemical reactions by composite machine learning method.

Stereoselective reactions have played a vital role in the emergence of life, evolution, human biology, and medicine. However, for a long time, most industrial and academic efforts followed a trial-and-error approach for asymmetric synthesis in stereoselective reactions. In addition, most previous studies have been qualitatively focused on the influence of steric and electronic effects on stereoselective reactions. Therefore, quantitatively understanding the stereoselectivity of a given chemical reaction is extremely difficult. As proof of principle, this paper develops a novel composite machine learning method for quantitatively predicting the enantioselectivity representing the degree to which one enantiomer is preferentially produced from the reactions. Specifically, machine learning methods that are widely used in data analytics, including Random Forest, Support Vector Regression, and LASSO, are utilized. In addition, the Bayesian optimization and permutation importance tests are provided for an in-depth understanding of reactions and accurate prediction. Finally, the proposed composite method approximates the key features of the available reactions by using Gaussian mixture models, which provide suitable machine learning methods for new reactions. The case studies using the real stereoselective reactions show that the proposed method is effective and provides a solid foundation for further application to other chemical reactions.

A machine learning approach to robustly determine director fields and analyze defects in active nematics.

Active nematics are dense systems of rodlike particles that consume energy to drive motion at the level of the individual particles. They exist in natural systems like biological tissues and artificial materials such as suspensions of self-propelled colloidal particles or synthetic microswimmers. Active nematics have attracted significant attention in recent years due to their spectacular nonequilibrium collective spatiotemporal dynamics, which may enable applications in fields such as robotics, drug delivery, and materials science. The director field, which measures the direction and degree of alignment of the local nematic orientation, is a crucial characteristic of active nematics and is essential for studying topological defects. However, determining the director field is a significant challenge in many experimental systems. Although director fields can be derived from images of active nematics using traditional imaging processing methods, the accuracy of such methods is highly sensitive to the settings of the algorithms. These settings must be tuned from image to image due to experimental noise, intrinsic noise of the imaging technology, and perturbations caused by changes in experimental conditions. This sensitivity currently limits automatic analysis of active nematics. To address this, we developed a machine learning model for extracting reliable director fields from raw experimental images, which enables accurate analysis of topological defects. Application of the algorithm to experimental data demonstrates that the approach is robust and highly generalizable to experimental settings that are different from those in the training data. It could be a promising tool for investigating active nematics and may be generalized to other active matter systems.

Toward Automatic Inference of Glycan Linkages Using MSn and Machine Learning─Proof of Concept Using Sialic Acid Linkages.

Glycosidic linkages in oligosaccharides play essential roles in determining their chemical properties and biological activities. MSn has been widely used to infer glycosidic linkages but requires a substantial amount of starting material, which limits its application. In addition, there is a lack of rigorous research on what MSn protocols are proper for characterizing glycosidic linkages. In this work, to deliver high-quality experimental data and analysis results, we propose a machine learning-based framework to establish appropriate MSn protocols and build effective data analysis methods. We demonstrate the proof-of-principle by applying our approach to elucidate sialic acid linkages (α2'-3' and α2'-6') in a set of sialyllactose standards and NIST sialic acid-containing N-glycans as well as identify several protocol configurations for producing high-quality experimental data. Our companion data analysis method achieves nearly 100% accuracy in classifying α2'-3' vs α2'-6' using MS5, MS4, MS3, or even MS2 spectra alone. The ability to determine glycosidic linkages using MS2 or MS3 is significant as it requires substantially less sample, enabling linkage analysis for quantity-limited natural glycans and synthesized materials, as well as shortens the overall experimental time. MS2 is also more amenable than MS3/4/5 to automation when coupled to direct infusion or LC-MS. Additionally, our method can predict the ratio of α2'-3' and α2'-6' in a mixture with 8.6% RMSE (root-mean-square error) across data sets using MS5 spectra. We anticipate that our framework will be generally applicable to analysis of other glycosidic linkages.

Lipoaspirate fluid derived factors and extracellular vesicles accelerate wound healing in a rat burn model.

Background: The regenerative capabilities of derivatives derived from the fat layer of lipoaspirate have been demonstrated. However, the large volume of lipoaspirate fluid has not attracted extensive attention in clinical applications. In this study, we aimed to isolate the factors and extracellular vesicles from human lipoaspirate fluid and evaluate their potential therapeutic efficacy. Methods: Lipoaspirate fluid derived factors and extracellular vesicles (LF-FVs) were prepared from human lipoaspirate and characterized by nanoparticle tracking analysis, size-exclusion chromatography and adipokine antibody arrays. The therapeutic potential of LF-FVs was evaluated on fibroblasts in vitro and rat burn model in vivo. Wound healing process was recorded on days 2, 4, 8, 10, 12 and 16 post-treatment. The scar formation was analyzed by histology, immunofluorescent staining and scar-related gene expression at day 35 post-treatment. Results: The results of nanoparticle tracking analysis and size-exclusion chromatography indicated that LF-FVs were enriched with proteins and extracellular vesicles. Specific adipokines (adiponectin and IGF-1) were detected in LF-FVs. In vitro, LF-FVs augmented the proliferation and migration of fibroblasts in a dose-dependent manner. In vivo, the results showed that LF-FVs significantly accelerated burn wound healing. Moreover, LF-FVs improved the quality of wound healing, including regenerating cutaneous appendages (hair follicles and sebaceous glands) and decreasing scar formation in the healed skin. Conclusion: LF-FVs were successfully prepared from lipoaspirate liquid, which were cell-free and enriched with extracellular vesicles. Additionally, they were found to improve wound healing in a rat burn model, suggesting that LF-FVs could be potentially used for wound regeneration in clinical settings.

De novo glycan sequencing by electronic excitation dissociation MS2-guided MS3 analysis on an Omnitrap-Orbitrap hybrid instrument.

Comprehensive de novo glycan sequencing remains an elusive goal due to the structural diversity and complexity of glycans. Present strategies employing collision-induced dissociation (CID) and higher energy collisional dissociation (HCD)-based multi-stage tandem mass spectrometry (MSn) or MS/MS combined with sequential exoglycosidase digestions are inherently low-throughput and difficult to automate. Compared to CID and HCD, electron transfer dissociation (ETD) and electron capture dissociation (ECD) each generate more cross-ring cleavages informative about linkage positions, but electronic excitation dissociation (EED) exceeds the information content of all other methods and is also applicable to analysis of singly charged precursors. Although EED can provide extensive glycan structural information in a single stage of MS/MS, its performance has largely been limited to FTICR MS, and thus it has not been widely adopted by the glycoscience research community. Here, the effective performance of EED MS/MS was demonstrated on a hybrid Orbitrap-Omnitrap QE-HF instrument, with high sensitivity, fragmentation efficiency, and analysis speed. In addition, a novel EED MS2-guided MS3 approach was developed for detailed glycan structural analysis. Automated topology reconstruction from MS2 and MS3 spectra could be achieved with a modified GlycoDeNovo software. We showed that the topology and linkage configurations of the Man9GlcNAc2 glycan can be accurately determined from first principles based on one EED MS2 and two CID-EED MS3 analyses, without reliance on biological knowledge, a structure database or a spectral library. The presented approach holds great promise for autonomous, comprehensive and de novo glycan sequencing.

Graph-Graph Similarity Network.

Graph learning aims to predict the label for an entire graph. Recently, graph neural network (GNN)-based approaches become an essential strand to learning low-dimensional continuous embeddings of entire graphs for graph label prediction. While GNNs explicitly aggregate the neighborhood information and implicitly capture the topological structure for graph representation, they ignore the relationships among graphs. In this article, we propose a graph-graph (G2G) similarity network to tackle the graph learning problem by constructing a SuperGraph through learning the relationships among graphs. Each node in the SuperGraph represents an input graph, and the weights of edges denote the similarity between graphs. By this means, the graph learning task is then transformed into a classical node label propagation problem. Specifically, we use an adversarial autoencoder to align embeddings of all the graphs to a prior data distribution. After the alignment, we design the G2G similarity network to learn the similarity between graphs, which functions as the adjacency matrix of the SuperGraph. By running node label propagation algorithms on the SuperGraph, we can predict the labels of graphs. Experiments on five widely used classification benchmarks and four public regression benchmarks under a fair setting demonstrate the effectiveness of our method.

Identification of thermogenesis-related lncRNAs in small extracellular vesicles derived from adipose tissue.

BACKGROUND: Brown adipose tissue (BAT) is considered as a primary location of adaptive thermogenesis and the thermogenic activities of brown adipocytes are also connected to generating heat and counteracting obesity. Recent studies revealed that BAT could secrete certain batokines-like factors especially small extracellular vesicles (sEVs), which contributed to the systemic consequences of BAT activities. As a newly emerging class of mediators, some long non-coding RNAs (lncRNAs) have exhibited metabolic regulatory effects in adipocyte development. However, besides the well-studied lncRNAs, the lncRNAs carried by sEVs derived from brown adipose tissue (sEV-BAT) have not been identified yet.  RESULTS: In this study, we demonstrated that sEV-BAT could induce beige adipocyte differentiation both in ASCs and 3T3-L1 cells, while sEV-WAT had no corresponding effects. The lncRNA microarray assay on sEV-WAT and sEV-BAT revealed a total of 563 types of known lncRNAs were identified to be differentially expressed, among which 232 lncRNAs were upregulated and 331 lncRNAs were downregulated in sEV-BAT. Three novel candidates (AK029592, humanlincRNA1030 and ENSMUST00000152284) were selected for further validation. LncRNA-mRNA network analysis revealed candidate lncRNAs were largely embedded in cellular metabolic pathways. During adipogenic and thermogenic phenotype differentiation in ASCs and 3T3-L1 cells, only the expressions of AK029592 were upregulated. The three lncRNAs were all relatively enriched in brown adipose tissues and brown adipocytes. In different adipocytes, sEV and adipose tissue, the expression of AK029592 and ENSMUST00000152284 were remarkably decreased in obese mice compared to lean mice, while obesity state could not change the expression of humanlincRNA1030. CONCLUSION: Collectively, our profiling study provided a comprehensive catalog for the study of lncRNAs specifically carried by sEV-BAT and indicated the potential regulatory role of certain sEV-BAT lncRNAs in thermogenesis.

Enzyme-Responsive Peptide Thioesters for Targeting Golgi Apparatus.

The Golgi apparatus (GA) is the hub of intracellular trafficking, but selectively targeting GA remains a challenge. We show an unconventional types of peptide thioesters, consisting of an aminoethyl thioester and acting as substrates of thioesterases, for instantly targeting the GA of cells. The peptide thioesters, above or below their critical micelle concentrations, enter cells mainly via caveolin-mediated endocytosis or macropinocytosis, respectively. After being hydrolyzed by GA-associated thioesterases, the resulting thiopeptides form dimers and accumulate in the GA. After saturating the GA, the thiopeptides are enriched in the endoplasmic reticulum (ER). Their buildup in ER and GA disrupts protein trafficking, thus leading to cell death via multiple pathways. The peptide thioesters target the GA of a wide variety of cells, including human, murine, and Drosophila cells. Changing d-diphenylalanine to l-diphenylalanine in the peptide maintains the GA-targeting ability. In addition, targeting GA redirects protein (e.g., NRAS) distribution. This work illustrates a thioesterase-responsive and redox-active molecular platform for targeting the GA and controlling cell fates.

A Self-Assembling Probe for Imaging the States of Golgi Apparatus in Live Single Cells.

Despite the enormous progress in genomics and proteomics, it is still challenging to assess the states of organelles in living cells with high spatiotemporal resolution. Based on our recent finding of enzyme-instructed self-assembly of a thiophosphopeptide that targets the Golgi Apparatus (GA) instantly, we use the thiophosphopeptide, which is enzymatically responsive and redox active, as an integrative probe for revealing the state of the GA of live cells at the single cell level. By imaging the probe in the GA of live cells over time, our results show that the accumulation of the probe at the GA depends on cell types. By comparison to a conventional Golgi probe, this self-assembling probe accumulates at the GA much faster and are sensitive to the expression of alkaline phosphatases. In addition, subtle changes of the fluorophore results in slightly different GA responses. This work illustrates a novel class of active molecular probes that combine enzyme-instructed self-assembly and redox reaction for high-resolution imaging of the states of subcellular organelles over a large area and extended times.

GlycoDeNovo2: An Improved MS/MS-Based De Novo Glycan Topology Reconstruction Algorithm.

Glycan structure identification is essential to understanding the roles of glycans in various biological processes. Previously, we developed GlycoDeNovo, a de novo algorithm for reconstructing glycan topologies from tandem mass spectra (MS/MS). In this work, we introduce GlycoDeNovo2 that contains two major improvements to GlycoDeNovo. First, we use the precursor mass measured for a peak that likely corresponds to a glycan to determine its potential compositions, which are used to constrain the search space, enable parallel computation, and hence speed up topology reconstruction. Second, we developed a procedure to calculate the empirical p-value of a reconstructed topology candidate. Experimental results are provided to demonstrate the effectiveness of GlycoDeNovo2.

Crash Prediction Using Deep Learning in a Disorienting Spaceflight Analog Balancing Task.

Were astronauts forced to land on the surface of Mars using manual control of their vehicle, they would not have familiar gravitational cues because Mars' gravity is only 0.38 g. They could become susceptible to spatial disorientation, potentially causing mission ending crashes. In our earlier studies, we secured blindfolded participants into a Multi-Axis Rotation System (MARS) device that was programmed to behave like an inverted pendulum. Participants used a joystick to stabilize around the balance point. We created a spaceflight analog condition by having participants dynamically balance in the horizontal roll plane, where they did not tilt relative to the gravitational vertical and therefore could not use gravitational cues to determine their position. We found 90% of participants in our spaceflight analog condition reported spatial disorientation and all of them showed it in their data. There was a high rate of crashing into boundaries that were set at ± 60° from the balance point. Our goal was to see whether we could use deep learning to predict the occurrence of crashes before they happened. We used stacked gated recurrent units (GRU) to predict crash events 800 ms in advance with an AUC (area under the curve) value of 99%. When we prioritized reducing false negatives we found it resulted in more false positives. We found that false negatives occurred when participants made destabilizing joystick deflections that rapidly moved the MARS away from the balance point. These unpredictable destabilizing joystick deflections, which occurred in the duration of time after the input data, are likely a result of spatial disorientation. If our model could work in real time, we calculated that immediate human action would result in the prevention of 80.7% of crashes, however, if we accounted for human reaction times (∼400 ms), only 30.3% of crashes could be prevented, suggesting that one solution could be an AI taking temporary control of the spacecraft during these moments.

Diverse RNAs in adipose-derived extracellular vesicles and their therapeutic potential.

Adipose tissue, which is considered an energy storage and active endocrine organ, produces and secretes a large amount of adipokines to regulate distant targets through blood circulation, especially extracellular vesicles (EVs). As cell-derived, membranous nanoparticles, EVs have recently garnered great attention as novel mediators in establishing intercellular communications as well as in accelerating interorgan crosstalk. Studies have revealed that the RNAs, including coding RNAs (messenger RNAs) and noncoding RNAs (long noncoding RNAs, microRNAs, and circular RNAs) are key bioactive cargoes of EV functions in various pathophysiological processes, such as cell differentiation, metabolic homeostasis, immune signal transduction, and cancer. Moreover, certain EV-contained RNAs have gradually been recognized as novel biomarkers, prognostic indicators, or even therapeutic nanodrugs of diseases. Therefore, in this review, we comprehensively summarize different classes of RNAs presented in adipose-derived EVs and discuss their therapeutic potential according to the latest research progress to provide valuable knowledge in this area.

MicroRNAs in Transforming Growth Factor-Beta Signaling Pathway Associated With Fibrosis Involving Different Systems of the Human Body.

Fibrosis, a major cause of morbidity and mortality, is a histopathological manifestation of many chronic inflammatory diseases affecting different systems of the human body. Two types of transforming growth factor beta (TGF-ß) signaling pathways regulate fibrosis: the canonical TGF-ß signaling pathway, represented by SMAD-2 and SMAD-3, and the noncanonical pathway, which functions without SMAD-2/3 participation and currently includes TGF-ß/mitogen-activated protein kinases, TGF-ß/SMAD-1/5, TGF-ß/phosphatidylinositol-3-kinase/Akt, TGF-ß/Janus kinase/signal transducer and activator of transcription protein-3, and TGF-ß/rho-associated coiled-coil containing kinase signaling pathways. MicroRNA (miRNA), a type of non-coding single-stranded small RNA, comprises approximately 22 nucleotides encoded by endogenous genes, which can regulate physiological and pathological processes in fibrotic diseases, particularly affecting organs such as the liver, the kidney, the lungs, and the heart. The aim of this review is to introduce the characteristics of the canonical and non-canonical TGF-ß signaling pathways and to classify miRNAs with regulatory effects on these two pathways based on the influenced organ. Further, we aim to summarize the limitations of the current research of the mechanisms of fibrosis, provide insights into possible future research directions, and propose therapeutic options for fibrosis.

Network-based virus-host interaction prediction with application to SARS-CoV-2.

COVID-19, caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has quickly become a global health crisis since the first report of infection in December of 2019. However, the infection spectrum of SARS-CoV-2 and its comprehensive protein-level interactions with hosts remain unclear. There is a massive amount of underutilized data and knowledge about RNA viruses highly relevant to SARS-CoV-2 and proteins of their hosts. More in-depth and more comprehensive analyses of that knowledge and data can shed new light on the molecular mechanisms underlying the COVID-19 pandemic and reveal potential risks. In this work, we constructed a multi-layer virus-host interaction network to incorporate these data and knowledge. We developed a machine-learning-based method to predict virus-host interactions at both protein and organism levels. Our approach revealed five potential infection targets of SARS-CoV-2 and 19 highly possible interactions between SARS-CoV-2 proteins and human proteins in the innate immune pathway.

Therapeutic potential of small extracellular vesicles derived from lipoma tissue in adipose tissue regeneration-an in vitro and in vivo study.

OBJECTIVE: To explore the adipogenic effects of the small extracellular vesicles derived from the lipoma tissues (sEV-LT), and to find a new cell-free therapeutic approach for adipose tissue regeneration. METHODS: Adipose tissue-derived stem cells (ADSCs) and small extracellular vesicles derived from the adipose tissues (sEV-AT) were isolated from human adipose tissue, while sEV-LT were isolated from human lipomatous tissue. ADSCs were characterized by using flow cytometric analysis and adipogenic and osteogenic differentiation assays. sEV was identified by electron microscopy, nanoparticle tracking, and western blotting. ADSCs were treated with sEV-LT and sEV-AT, respectively. Fluorescence confocal microscopy was used to investigate whether sEV-LT and sEV-AT could be taken by ADSCs. The proliferation and migration abilities and adipogenic differentiation assay of ADSCs were evaluated by CCK-8 assays, scratch test, and oil red O staining test, and the expression levels of adipogenic-related genes C/EBP-δ, PPARγ2, and Adiponectin in ADSCs were assessed by real-time quantitative PCR (RT-PCR). The sEV-LT and sEV-AT transplantation tubes were implanted subcutaneously in SD rats, and the neotissues were qualitatively and histologically evaluated at 2, 4, 8, and 12 weeks after transplantation. Hematoxylin and eosin (H&E) staining was subsequently used to observe and compare the adipogenesis and angiogenesis in neotissues, while immunohistochemistry was used to examine the expression and the distribution of C/EBP-α, PPARγ, Adiponectin, and CD31 at the 4th week. RESULTS: The in vitro experiments showed that both sEV-LT and sEV-AT could be taken up by ADSCs via endocytosis. The scratch experiment and CCK-8 experiment showed that the migration area and proliferation number of ADSCs in sEV-LT group and sEV-AT group were significantly higher than those in the non-sEV group (p < 0.05). Compared with sEV-AT group, sEV-LT group had larger migration area and proliferation number of ADSCs (p < 0.05). Oil red O staining and RT-PCR experiments showed that, compared with the non-sEVs group, the lipid droplets and the mRNA expression levels of adipogenesis-related genes PPARγ2 and Adiponectin of ADSCs in sEV-LT group and sEV-AT group were significantly upregulated (p < 0.05); however, there was no statistical significance in the expression level of C/EBP-δ (p > 0.05). In addition, no significant difference in the amount of lipid droplets and adipogenesis-related genes between the sEV-LT groups and sEV-AT was seen (p > 0.05). At 2, 4, 8, and 12 weeks, the adipocyte area and the number of capillaries in neotissues in the sEV-LT groups and sEV-AT groups were significantly increased compared with the Matrigel group (p < 0.05); however, there was no dramatic difference between sEV-LT groups and sEV-AT groups (p > 0.05). At the 4th week, neotissues in the sEV-LT groups and sEV-AT groups all showed upregulated expression of C/EBP-α, PPARγ, Adiponectin, and CD31 protein, while neotissues in the Matrigel group only showed positive expression of CD31 protein. CONCLUSIONS: This study demonstrated that sEV-LT exerted promotion effects on adipose tissue regeneration by accelerating the proliferation, migration, and adipogenic differentiation of ADSCs in vitro and recruiting adipocytes and promoting angiogenesis in vivo. The sEV-LT could serve as an alternative cell-free therapeutic strategy for generating adipose tissue, thus providing a promising application prospect in tissue engineering.

Machine learning forecasting of active nematics.

Active nematics are a class of far-from-equilibrium materials characterized by local orientational order of force-generating, anisotropic constitutes. Traditional methods for predicting the dynamics of active nematics rely on hydrodynamic models, which accurately describe idealized flows and many of the steady-state properties, but do not capture certain detailed dynamics of experimental active nematics. We have developed a deep learning approach that uses a Convolutional Long-Short-Term-Memory (ConvLSTM) algorithm to automatically learn and forecast the dynamics of active nematics. We demonstrate our purely data-driven approach on experiments of 2D unconfined active nematics of extensile microtubule bundles, as well as on data from numerical simulations of active nematics.

IMMIGRATE: A Margin-Based Feature Selection Method with Interaction Terms.

Traditional hypothesis-margin researches focus on obtaining large margins and feature selection. In this work, we show that the robustness of margins is also critical and can be measured using entropy. In addition, our approach provides clear mathematical formulations and explanations to uncover feature interactions, which is often lack in large hypothesis-margin based approaches. We design an algorithm, termed IMMIGRATE (Iterative max-min entropy margin-maximization with interaction terms), for training the weights associated with the interaction terms. IMMIGRATE simultaneously utilizes both local and global information and can be used as a base learner in Boosting. We evaluate IMMIGRATE in a wide range of tasks, in which it demonstrates exceptional robustness and achieves the state-of-the-art results with high interpretability.

Characterizing Individual Differences in a Dynamic Stabilization Task Using Machine Learning.

INTRODUCTION: Being able to identify individual differences in skilled motor learning during disorienting conditions is important for spaceflight, military aviation, and rehabilitation.METHODS: Blindfolded subjects (N = 34) were strapped into a device that behaved like an inverted pendulum in the horizontal roll plane and were instructed to use a joystick to stabilize themselves across two experimental sessions on consecutive days. Subjects could not use gravitational cues to determine their angular position and many soon became spatially disoriented.RESULTS: Most demonstrated minimal learning, poor performance, and a characteristic pattern of positional drifting during horizontal roll plane balancing. To understand the wide range of individual differences observed, we used a Bayesian Gaussian Mixture method to cluster subjects into three statistically distinct groups that represent Proficient, Somewhat Proficient, and Not Proficient performance. We found that subjects in the Not Proficient group exhibited a suboptimal strategy of using very stereotyped large magnitude joystick deflections. We also used a Gaussian Naive Bayes method to create predictive classifiers. As early as the second block of experimentation (out of ten), we could predict a subject's final group with 80% accuracy.DISCUSSION: Our findings indicate that machine learning can help predict individual performance and learning in a disorienting dynamic stabilization task and identify suboptimal strategies in Not Proficient subjects, which could lead to personalized and more effective training programs.Vimal VP, Zheng H, Hong P, Fakharzadeh LN, Lackner JR, DiZio P. Characterizing individual differences in a dynamic stabilization task using machine learning. Aerosp Med Hum Perform. 2020; 91(6):479-488.

Atypical Myosin Tunes Dendrite Arbor Subdivision.

Dendrite arbor pattern determines the functional characteristics of a neuron. It is founded on primary branch structure, defined through cell intrinsic and transcription-factor-encoded mechanisms. Developing arbors have extensive acentrosomal microtubule dynamics, and here, we report an unexpected role for the atypical actin motor Myo6 in creating primary branch structure by specifying the position, polarity, and targeting of these events. We carried out in vivo time-lapse imaging of Drosophila adult sensory neuron differentiation, integrating machine-learning-based quantification of arbor patterning with molecular-level tracking of cytoskeletal remodeling. This revealed that Myo6 and the transcription factor Knot regulate transient surges of microtubule polymerization at dendrite tips; they drive retrograde extension of an actin filament array that specifies anterograde microtubule polymerization and guides these microtubules to subdivide the tip into multiple branches. Primary branches delineate functional compartments; this tunable branching mechanism is key to define and diversify dendrite arbor compartmentalization.