Robust deep learning object recognition models rely on low frequency information in natural images.

Machine learning models have difficulty generalizing to data outside of the distribution they were trained on. In particular, vision models are usually vulnerable to adversarial attacks or common corruptions, to which the human visual system is robust. Recent studies have found that regularizing machine learning models to favor brain-like representations can improve model robustness, but it is unclear why. We hypothesize that the increased model robustness is partly due to the low spatial frequency preference inherited from the neural representation. We tested this simple hypothesis with several frequency-oriented analyses, including the design and use of hybrid images to probe model frequency sensitivity directly. We also examined many other publicly available robust models that were trained on adversarial images or with data augmentation, and found that all these robust models showed a greater preference to low spatial frequency information. We show that preprocessing by blurring can serve as a defense mechanism against both adversarial attacks and common corruptions, further confirming our hypothesis and demonstrating the utility of low spatial frequency information in robust object recognition.

In silico dynamics of COVID-19 phenotypes for optimizing clinical management.

Understanding the underlying mechanisms of COVID-19 progression and the impact of various pharmaceutical interventions is crucial for the clinical management of the disease. We developed a comprehensive mathematical framework based on the known mechanisms of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, incorporating the renin-angiotensin system and ACE2, which the virus exploits for cellular entry, key elements of the innate and adaptive immune responses, the role of inflammatory cytokines, and the coagulation cascade for thrombus formation. The model predicts the evolution of viral load, immune cells, cytokines, thrombosis, and oxygen saturation based on patient baseline condition and the presence of comorbidities. Model predictions were validated with clinical data from healthy people and COVID-19 patients, and the results were used to gain insight into identified risk factors of disease progression including older age; comorbidities such as obesity, diabetes, and hypertension; and dysregulated immune response. We then simulated treatment with various drug classes to identify optimal therapeutic protocols. We found that the outcome of any treatment depends on the sustained response rate of activated CD8+ T cells and sufficient control of the innate immune response. Furthermore, the best treatment-or combination of treatments-depends on the preinfection health status of the patient. Our mathematical framework provides important insight into SARS-CoV-2 pathogenesis and could be used as the basis for personalized, optimal management of COVID-19.

Domain-driven models yield better predictions at lower cost than reservoir computers in Lorenz systems.

Recent advances in computing algorithms and hardware have rekindled interest in developing high-accuracy, low-cost surrogate models for simulating physical systems. The idea is to replace expensive numerical integration of complex coupled partial differential equations at fine time scales performed on supercomputers, with machine-learned surrogates that efficiently and accurately forecast future system states using data sampled from the underlying system. One particularly popular technique being explored within the weather and climate modelling community is the echo state network (ESN), an attractive alternative to other well-known deep learning architectures. Using the classical Lorenz 63 system, and the three tier multi-scale Lorenz 96 system (Thornes T, Duben P, Palmer T. 2017 Q. J. R. Meteorol. Soc. 143, 897-908. (doi:10.1002/qj.2974)) as benchmarks, we realize that previously studied state-of-the-art ESNs operate in two distinct regimes, corresponding to low and high spectral radius (LSR/HSR) for the sparse, randomly generated, reservoir recurrence matrix. Using knowledge of the mathematical structure of the Lorenz systems along with systematic ablation and hyperparameter sensitivity analyses, we show that state-of-the-art LSR-ESNs reduce to a polynomial regression model which we call Domain-Driven Regularized Regression (D2R2). Interestingly, D2R2 is a generalization of the well-known SINDy algorithm (Brunton SL, Proctor JL, Kutz JN. 2016 Proc. Natl Acad. Sci. USA 113, 3932-3937. (doi:10.1073/pnas.1517384113)). We also show experimentally that LSR-ESNs (Chattopadhyay A, Hassanzadeh P, Subramanian D. 2019 (http://arxiv.org/abs/1906.08829)) outperform HSR ESNs (Pathak J, Hunt B, Girvan M, Lu Z, Ott E. 2018 Phys. Rev. Lett. 120, 024102. (doi:10.1103/PhysRevLett.120.024102)) while D2R2 dominates both approaches. A significant goal in constructing surrogates is to cope with barriers to scaling in weather prediction and simulation of dynamical systems that are imposed by time and energy consumption in supercomputers. Inexact computing has emerged as a novel approach to helping with scaling. In this paper, we evaluate the performance of three models (LSR-ESN, HSR-ESN and D2R2) by varying the precision or word size of the computation as our inexactness-controlling parameter. For precisions of 64, 32 and 16 bits, we show that, surprisingly, the least expensive D2R2 method yields the most robust results and the greatest savings compared to ESNs. Specifically, D2R2 achieves 68 × in computational savings, with an additional 2 × if precision reductions are also employed, outperforming ESN variants by a large margin. This article is part of the theme issue 'Machine learning for weather and climate modelling'.

Pediatric Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2): Clinical Presentation, Infectivity, and Immune Responses.

OBJECTIVES: As schools plan for re-opening, understanding the potential role children play in the coronavirus infectious disease 2019 (COVID-19) pandemic and the factors that drive severe illness in children is critical. STUDY DESIGN: Children ages 0-22 years with suspected severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection presenting to urgent care clinics or being hospitalized for confirmed/suspected SARS-CoV-2 infection or multisystem inflammatory syndrome in children (MIS-C) at Massachusetts General Hospital were offered enrollment in the Massachusetts General Hospital Pediatric COVID-19 Biorepository. Enrolled children provided nasopharyngeal, oropharyngeal, and/or blood specimens. SARS-CoV-2 viral load, ACE2 RNA levels, and serology for SARS-CoV-2 were quantified. RESULTS: A total of 192 children (mean age, 10.2 ± 7.0 years) were enrolled. Forty-nine children (26%) were diagnosed with acute SARS-CoV-2 infection; an additional 18 children (9%) met the criteria for MIS-C. Only 25 children (51%) with acute SARS-CoV-2 infection presented with fever; symptoms of SARS-CoV-2 infection, if present, were nonspecific. Nasopharyngeal viral load was highest in children in the first 2 days of symptoms, significantly higher than hospitalized adults with severe disease (P = .002). Age did not impact viral load, but younger children had lower angiotensin-converting enzyme 2 expression (P = .004). Immunoglobulin M (IgM) and Immunoglobulin G (IgG) to the receptor binding domain of the SARS-CoV-2 spike protein were increased in severe MIS-C (P < .001), with dysregulated humoral responses observed. CONCLUSIONS: This study reveals that children may be a potential source of contagion in the SARS-CoV-2 pandemic despite having milder disease or a lack of symptoms; immune dysregulation is implicated in severe postinfectious MIS-C.

Prognostic machine learning models for COVID-19 to facilitate decision making.

An increasing number of COVID-19 cases worldwide has overwhelmed the healthcare system. Physicians are struggling to allocate resources and to focus their attention on high-risk patients, partly because early identification of high-risk individuals is difficult. This can be attributed to the fact that COVID-19 is a novel disease and its pathogenesis is still partially understood. However, machine learning algorithms have the capability to analyse a large number of parameters within a short period of time to identify the predictors of disease outcome. Implementing such an algorithm to predict high-risk individuals during the early stages of infection would be helpful in decision making for clinicians such that irreversible damage could be prevented. Here, we propose recommendations to develop prognostic machine learning models using electronic health records so that a real-time risk score can be developed for COVID-19.

Increasing Medical Student Interest in Nephrology.

BACKGROUND: Interest in nephrology careers is declining, possibly due to perceptions of the field and/or training aspects. Understanding practices of medical schools successfully instilling nephrology interest could inform efforts to attract leading candidates to the specialty. METHODS: The American Society of Nephrology Workforce Committee's Best Practices Project was one of several initiatives to increase nephrology career interest. Board-certified nephrologists graduating medical school between 2002 and 2009 were identified in the American Medical Association Masterfile and their medical schools ranked by production. Renal educators from the top 10 producing institutions participated in directed focus groups inquiring about key factors in creating nephrology career interest, including aspects of their renal courses, clinical rotations, research activities, and faculty interactions. Thematic content analysis of the transcripts (with inductive reasoning implementing grounded theory) was performed to identify factors contributing to their programs' success. RESULTS: The 10 schools identified were geographically representative, with similar proportions of graduates choosing internal medicine (mean 26%) as the national graduating class (26% in the 2017 residency Match). Eighteen educators from 9 of these 10 institutions participated. Four major themes were identified contributing to these schools' success: (1) nephrology faculty interaction with medical students; (2) clinical exposure to nephrology and clinical relevance of renal pathophysiology materials; (3) use of novel educational modalities; and (4) exposure, in particular early exposure, to the breadth of nephrology practice. CONCLUSION: Early and consistent exposure to a range of clinical nephrology experiences and nephrology faculty contact with medical students are important to help generate interest in the specialty.

Enhancing Nephrology Career Interest through the ASN Kidney TREKS Program.

The Kidney Tutored Research and Education for Kidney Students (TREKS) Program is a product of the American Society of Nephrology (ASN) Workforce Committee that seeks to connect medical and graduate students to nephrology. This program starts with a weeklong camp-like course introducing participants to renal physiology through classic and modern experiments. Next, each student is matched with a nephrology mentor at his or her home institution to foster a better understanding of a nephrology career. Lastly, the students are encouraged to participate in scholarly activities and attend the ASN Kidney Week. Now in its third year, with a total of 84 participants, survey data suggest early success of the program, with a self-reported 40% increased interest in nephrology fellowship and/or research careers. In addition, students give high ratings to the course components and mentorship pairings. Continued student tracking will be necessary to determine the long-term program effect.

The 90-degree rule in posterior ptosis surgery.

To propose a guideline for ptosis clamp positioning to minimize the risk of globe injury during posterior ptosis surgery. Measurements of 20 consecutive patients, 40 eyelids, undergoing bilateral posterior ptosis repair surgery were taken; as a surrogate for needle tip position, measurement of the distance from the clamp base to the ocular surface was taken using a caliper with the clamp held at 90-degrees to the ocular surface and again at 45-degrees to the ocular surface. These measurements were compared to geometric predictions of the distance from the clamp base to the ocular surface. The average distance from the clamp base to the ocular surface when the clamp is held 90-degrees to the ocular surface was 7 mm, this distance decreases to 5 mm when the clamp is held 45° to the ocular surface. This coincides well with geometric predictions. Posterior ptosis surgery overall has an excellent safety profile; however, complications are possible, perhaps the most severe of which is inadvertent globe and/or corneal injury. The more acute the angle the ptosis clamp is held, the closer the clamp base, and subsequently the needle tip, is to the ocular surface as would be predicted geometrically. This coincides with closer proximity of the needle to the ocular surface during surgery. The theoretical risk of globe injury should decrease as the distance of the needle from the globe increases, and this distance is greatest when the clamp is held at a 90-degree angle to the ocular surface. This distinction becomes particularly important to consider in large eye morphology patients where the distance from the needle to the globe can approach 2 mm when the clamp is held at 45-degrees.

Translational symmetry in convolutions with localized kernels causes an implicit bias toward high frequency adversarial examples.

Adversarial attacks are still a significant challenge for neural networks. Recent efforts have shown that adversarial perturbations typically contain high-frequency features, but the root cause of this phenomenon remains unknown. Inspired by theoretical work on linear convolutional models, we hypothesize that translational symmetry in convolutional operations together with localized kernels implicitly bias the learning of high-frequency features, and that this is one of the main causes of high frequency adversarial examples. To test this hypothesis, we analyzed the impact of different choices of linear and non-linear architectures on the implicit bias of the learned features and adversarial perturbations, in spatial and frequency domains. We find that, independently of the training dataset, convolutional operations have higher frequency adversarial attacks compared to other architectural parameterizations, and that this phenomenon is exacerbated with stronger locality of the kernel (kernel size) end depth of the model. The explanation for the kernel size dependence involves the Fourier Uncertainty Principle: a spatially-limited filter (local kernel in the space domain) cannot also be frequency-limited (local in the frequency domain). Using larger convolution kernel sizes or avoiding convolutions (e.g., by using Vision Transformers or MLP-style architectures) significantly reduces this high-frequency bias. Looking forward, our work strongly suggests that understanding and controlling the implicit bias of architectures will be essential for achieving adversarial robustness.

Disruption of neural periodicity predicts clinical response after deep brain stimulation for obsessive-compulsive disorder.

Recent advances in surgical neuromodulation have enabled chronic and continuous intracranial monitoring during everyday life. We used this opportunity to identify neural predictors of clinical state in 12 individuals with treatment-resistant obsessive-compulsive disorder (OCD) receiving deep brain stimulation (DBS) therapy ( NCT05915741 ). We developed our neurobehavioral models based on continuous neural recordings in the region of the ventral striatum in an initial cohort of five patients and tested and validated them in a held-out cohort of seven additional patients. Before DBS activation, in the most symptomatic state, theta/alpha (9 Hz) power evidenced a prominent circadian pattern and a high degree of predictability. In patients with persistent symptoms (non-responders), predictability of the neural data remained consistently high. On the other hand, in patients who improved symptomatically (responders), predictability of the neural data was significantly diminished. This neural feature accurately classified clinical status even in patients with limited duration recordings, indicating generalizability that could facilitate therapeutic decision-making.

Feedback inhibition of ENaC during acute sodium loading in vivo.

The epithelial Na(+) channel (ENaC) is tightly regulated by sodium intake to maintain whole body sodium homeostasis. In addition, ENaC is inhibited by high levels of intracellular Na(+) [Na(+)](i), presumably to prevent cell Na(+) overload and swelling. However, it is not clear if this regulation is relevant in vivo. We show here that in rats, an acute (4 h) oral sodium load decreases whole-cell amiloride-sensitive currents (I(Na)) in the cortical collecting duct (CCD) even when plasma aldosterone levels are maintained high by infusing the hormone. This was accompanied by decreases in whole-kidney cleaved α-ENaC (2.6 fold), total ß-ENaC (1.7 fold), and cleaved γ-ENaC (6.2 fold). In addition, cell-surface ß- and γ-ENaC expression was measured using in situ biotinylation. There was a decrease in cell-surface core-glycosylated (2.2 fold) and maturely glycosylated (4.9 fold) ß-ENaC and cleaved γ-ENaC (4.7 fold). There were no significant changes for other apical sodium transporters. To investigate the role of increases in Na(+) entry and presumably [Na(+)](i) on ENaC, animals were infused with amiloride prior to and during sodium loading. Blocking Na(+) entry did not inhibit the effect of resalting on I(Na). However, amiloride did prevent decreases in ENaC expression, an effect that was not mimicked by hydrochlorothiazide administration. Na(+) entry and presumably [Na(+)](i) can regulate ENaC expression but does not fully account for the aldosterone-independent decrease in I(Na) during an acute sodium load.

Estimating kidney function in patients with cancer: A narrative review.

AIM: Accurate evaluation of glomerular filtration rate (GFR) is crucial in Oncology as drug eligibility and dosing depend on estimates of GFR. However, there are no clear guidelines on the optimal method of determining kidney function in patients with cancer. We aimed to summarize the evidence on estimation of kidney function in patients with cancer. METHODS: We searched PubMed for literature discussing the performance of GFR estimating equations in patients with malignancy to create a table of the evidence for creatinine- and cystatin c-based equations. We further reviewed novel estimation techniques such as panel eGFR, real-time measured GFR, and functional magnetic resonance imaging. RESULTS: The commonly used GFR estimating equations were derived from populations of patients without cancer. These equations may be less applicable in Oncology due to severe sarcopenia, inflammation, and other physiologic changes in patients with cancer. The Cockcroft-Gault equation currently dominates in clinical Oncology despite significant limitations and accumulating evidence for use of the CKD-EPICr formula. Additional considerations in the practice of Oncology include a recently developed equation (CamGFRv2, also called the Janowitz formula) and the use of cystatin c-based equations to overcome some of the barriers to accurate GFR estimation based on creatinine alone. CONCLUSION: Overall, we suggest using the CKD-EPI equations (either cystatin c or creatinine-based) among patients with cancer in routine clinical practice and measured GFR for patients at a critical threshold for treatment decisions.

Human ureteric bud organoids recapitulate branching morphogenesis and differentiate into functional collecting duct cell types.

Directed differentiation of human pluripotent stem cells (hPSCs) into functional ureteric and collecting duct (CD) epithelia is essential to kidney regenerative medicine. Here we describe highly efficient, serum-free differentiation of hPSCs into ureteric bud (UB) organoids and functional CD cells. The hPSCs are first induced into pronephric progenitor cells at 90% efficiency and then aggregated into spheres with a molecular signature similar to the nephric duct. In a three-dimensional matrix, the spheres form UB organoids that exhibit branching morphogenesis similar to the fetal UB and correct distal tip localization of RET expression. Organoid-derived cells incorporate into the UB tips of the progenitor niche in chimeric fetal kidney explant culture. At later stages, the UB organoids differentiate into CD organoids, which contain >95% CD cell types as estimated by single-cell RNA sequencing. The CD epithelia demonstrate renal electrophysiologic functions, with ENaC-mediated vectorial sodium transport by principal cells and V-type ATPase proton pump activity by FOXI1-induced intercalated cells.

Decoding Depression Severity From Intracranial Neural Activity.

BACKGROUND: Disorders of mood and cognition are prevalent, disabling, and notoriously difficult to treat. Fueling this challenge in treatment is a significant gap in our understanding of their neurophysiological basis. METHODS: We recorded high-density neural activity from intracranial electrodes implanted in depression-relevant prefrontal cortical regions in 3 human subjects with severe depression. Neural recordings were labeled with depression severity scores across a wide dynamic range using an adaptive assessment that allowed sampling with a temporal frequency greater than that possible with typical rating scales. We modeled these data using regularized regression techniques with region selection to decode depression severity from the prefrontal recordings. RESULTS: Across prefrontal regions, we found that reduced depression severity is associated with decreased low-frequency neural activity and increased high-frequency activity. When constraining our model to decode using a single region, spectral changes in the anterior cingulate cortex best predicted depression severity in all 3 subjects. Relaxing this constraint revealed unique, individual-specific sets of spatiospectral features predictive of symptom severity, reflecting the heterogeneous nature of depression. CONCLUSIONS: The ability to decode depression severity from neural activity increases our fundamental understanding of how depression manifests in the human brain and provides a target neural signature for personalized neuromodulation therapies.