Semiconductor Thermal and Electrical Properties Decoupled by Localized Phonon Resonances.

Thermoelectric materials convert heat into electricity through thermally driven charge transport in solids or vice versa for cooling. To compete with conventional energy-conversion technologies, a thermoelectric material must possess the properties of both an electrical conductor and a thermal insulator. However, these properties are normally mutually exclusive because of the interconnection between scattering mechanisms for charge carriers and phonons. Recent theoretical investigations on sub-device scales have revealed that nanopillars attached to a membrane exhibit a multitude of local phonon resonances, spanning the full spectrum, that couple with the heat-carrying phonons in the membrane and cause a reduction in the in-plane thermal conductivity, with no expected change in the electrical properties because the nanopillars are outside the pathway of voltage generation and charge transport. Here this effect is demonstrated experimentally for the first time by investigating device-scale suspended silicon membranes with GaN nanopillars grown on the surface. The nanopillars cause up to 21% reduction in the thermal conductivity while the power factor remains unaffected, thus demonstrating an unprecedented decoupling in the semiconductor's thermoelectric properties. The measured thermal conductivity behavior for coalesced nanopillars and corresponding lattice-dynamics calculations provide evidence that the reductions are mechanistically tied to the phonon resonances. This finding paves the way for high-efficiency solid-state energy recovery and cooling.

Enhancing convolutional neural network predictions of electrocardiograms with left ventricular dysfunction using a novel sub-waveform representation.

Background: Electrocardiogram (ECG) deep learning (DL) has promise to improve the outcomes of patients with cardiovascular abnormalities. In ECG DL, researchers often use convolutional neural networks (CNNs) and traditionally use the full duration of raw ECG waveforms that create redundancies in feature learning and result in inaccurate predictions with large uncertainties. Objective: For enhancing these predictions, we introduced a sub-waveform representation that leverages the rhythmic pattern of ECG waveforms (data-centric approach) rather than changing the CNN architecture (model-centric approach). Results: We applied the proposed representation to a population with 92,446 patients to identify left ventricular dysfunction. We found that the sub-waveform representation increases the performance metrics compared to the full-waveform representation. We observed a 2% increase for area under the receiver operating characteristic curve and 10% increase for area under the precision-recall curve. We also carefully examined three reliability components of explainability, interpretability, and fairness. We provided an explanation for enhancements obtained by heartbeat alignment mechanism. By developing a new scoring system, we interpreted the clinical relevance of ECG features and showed that sub-waveform representation further pushes the scores towards clinical predictions. Finally, we showed that the new representation significantly reduces prediction uncertainties within subgroups that contributes to individual fairness. Conclusion: We expect that this added control over the granularity of ECG data will improve the DL modeling for new artificial intelligence technologies in the cardiovascular space.

Development of a machine learning model using electrocardiogram signals to improve acute pulmonary embolism screening.

Aims: Clinical scoring systems for pulmonary embolism (PE) screening have low specificity and contribute to computed tomography pulmonary angiogram (CTPA) overuse. We assessed whether deep learning models using an existing and routinely collected data modality, electrocardiogram (ECG) waveforms, can increase specificity for PE detection. Methods and results: We create a retrospective cohort of 21 183 patients at moderate- to high suspicion of PE and associate 23 793 CTPAs (10.0% PE-positive) with 320 746 ECGs and encounter-level clinical data (demographics, comorbidities, vital signs, and labs). We develop three machine learning models to predict PE likelihood: an ECG model using only ECG waveform data, an EHR model using tabular clinical data, and a Fusion model integrating clinical data and an embedded representation of the ECG waveform. We find that a Fusion model [area under the receiver-operating characteristic curve (AUROC) 0.81 ± 0.01] outperforms both the ECG model (AUROC 0.59 ± 0.01) and EHR model (AUROC 0.65 ± 0.01). On a sample of 100 patients from the test set, the Fusion model also achieves greater specificity (0.18) and performance (AUROC 0.84 ± 0.01) than four commonly evaluated clinical scores: Wells' Criteria, Revised Geneva Score, Pulmonary Embolism Rule-Out Criteria, and 4-Level Pulmonary Embolism Clinical Probability Score (AUROC 0.50-0.58, specificity 0.00-0.05). The model is superior to these scores on feature sensitivity analyses (AUROC 0.66-0.84) and achieves comparable performance across sex (AUROC 0.81) and racial/ethnic (AUROC 0.77-0.84) subgroups. Conclusion: Synergistic deep learning of ECG waveforms with traditional clinical variables can increase the specificity of PE detection in patients at least at moderate suspicion for PE.

Contrastive learning improves critical event prediction in COVID-19 patients.

Deep learning (DL) models typically require large-scale, balanced training data to be robust, generalizable, and effective in the context of healthcare. This has been a major issue for developing DL models for the coronavirus disease 2019 (COVID-19) pandemic, where data are highly class imbalanced. Conventional approaches in DL use cross-entropy loss (CEL), which often suffers from poor margin classification. We show that contrastive loss (CL) improves the performance of CEL, especially in imbalanced electronic health records (EHR) data for COVID-19 analyses. We use a diverse EHR dataset to predict three outcomes: mortality, intubation, and intensive care unit (ICU) transfer in hospitalized COVID-19 patients over multiple time windows. To compare the performance of CEL and CL, models are tested on the full dataset and a restricted dataset. CL models consistently outperform CEL models, with differences ranging from 0.04 to 0.15 for area under the precision and recall curve (AUPRC) and 0.05 to 0.1 for area under the receiver-operating characteristic curve (AUROC).

Directional thermal channeling: A phenomenon triggered by tight packing of heat sources.

Understanding nanoscale thermal transport is critical for nano-engineered devices such as quantum sensors, thermoelectrics, and nanoelectronics. However, despite overwhelming experimental evidence for nondiffusive heat dissipation from nanoscale heat sources, the underlying mechanisms are still not understood. In this work, we show that for nanoscale heat source spacings that are below the mean free path of the dominant phonons in a substrate, close packing of the heat sources increases in-plane scattering and enhances cross-plane thermal conduction. This leads to directional channeling of thermal transport-a novel phenomenon. By using advanced atomic-level simulations to accurately access the lattice temperature and the phonon scattering and transport properties, we finally explain the counterintuitive experimental observations of enhanced cooling for close-packed heat sources. This represents a distinct fundamental behavior in materials science with far-reaching implications for electronics and future quantum devices.

Contrastive Learning Improves Critical Event Prediction in COVID-19 Patients.

Machine Learning (ML) models typically require large-scale, balanced training data to be robust, generalizable, and effective in the context of healthcare. This has been a major issue for developing ML models for the coronavirus-disease 2019 (COVID-19) pandemic where data is highly imbalanced, particularly within electronic health records (EHR) research. Conventional approaches in ML use cross-entropy loss (CEL) that often suffers from poor margin classification. For the first time, we show that contrastive loss (CL) improves the performance of CEL especially for imbalanced EHR data and the related COVID-19 analyses. This study has been approved by the Institutional Review Board at the Icahn School of Medicine at Mount Sinai. We use EHR data from five hospitals within the Mount Sinai Health System (MSHS) to predict mortality, intubation, and intensive care unit (ICU) transfer in hospitalized COVID-19 patients over 24 and 48 hour time windows. We train two sequential architectures (RNN and RETAIN) using two loss functions (CEL and CL). Models are tested on full sample data set which contain all available data and restricted data set to emulate higher class imbalance.CL models consistently outperform CEL models with the restricted data set on these tasks with differences ranging from 0.04 to 0.15 for AUPRC and 0.05 to 0.1 for AUROC. For the restricted sample, only the CL model maintains proper clustering and is able to identify important features, such as pulse oximetry. CL outperforms CEL in instances of severe class imbalance, on three EHR outcomes with respect to three performance metrics: predictive power, clustering, and feature importance. We believe that the developed CL framework can be expanded and used for EHR ML work in general.

Physics of surface vibrational resonances: pillared phononic crystals, metamaterials, and metasurfaces.

The introduction of engineered resonance phenomena on surfaces has opened a new frontier in surface science and technology. Pillared phononic crystals, metamaterials, and metasurfaces are an emerging class of artificial structured media, featuring surfaces that consist of pillars-or branching substructures-standing on a plate or a substrate. A pillared phononic crystal exhibits Bragg band gaps, while a pillared metamaterial may feature both Bragg band gaps and local resonance hybridization band gaps. These two band-gap phenomena, along with other unique wave dispersion characteristics, have been exploited for a variety of applications spanning a range of length scales and covering multiple disciplines in applied physics and engineering, particularly in elastodynamics and acoustics. The intrinsic placement of pillars on a semi-infinite surface-yielding a metasurface-has similarly provided new avenues for the control and manipulation of wave propagation. Classical waves are admitted in pillared media, including Lamb waves in plates and Rayleigh and Love waves along the surfaces of substrates, ranging in frequency from hertz to several gigahertz. With the presence of the pillars, these waves couple with surface resonances richly creating new phenomena and properties in the subwavelength regime and in some applications at higher frequencies as well. At the nanoscale, it was shown that atomic-scale resonances-stemming from nanopillars-alter the fundamental nature of conductive thermal transport by reducing the group velocities and generating mode localizations across the entire spectrum of the constituent material well into the terahertz regime. In this article, we first overview the history and development of pillared materials, then provide a detailed synopsis of a selection of key research topics that involve the utilization of pillars or similar branching substructures in different contexts. Finally, we conclude by providing a short summary and some perspectives on the state of the field and its promise for further future development.

Knowledge, attitudes, risk perceptions, and practices of adults toward COVID-19: a population and field-based study from Iran.

OBJECTIVES: To determine peoples' knowledge, attitudes, risk perceptions, and practices to provide policymakers pieces of field-based evidence and help them in the management of the COVID-19 epidemic. METHODS: This population-based survey was conducted using multi-stage stratified and cluster sampling in Shiraz, Iran. A total of 1331 persons were interviewed. The questionnaires were completed by face-to-face interviews. Univariable and multivariable (linear regression) data analyses were done using SPSS. RESULTS: The participants answered 63% of questions regarding knowledge, and 78% of questions regarding practice correctly. Only, 4.8% knew about common symptoms of COVID-19 and 7.3% about warning signs that require referral to hospitals. Males, lower educated people, and elders had a lower level of knowledge and poorer practices. Knowledge was also lower in the marginalized (socially deprived) people. Knowledge and practices' correlation was 37%. Overall, 43.6% considered themselves at high risk of COVID-19, and 50% considered it as a severe disease. This disease had negative effects on most participants' routine activities (69.1%). The participants preferred to follow the news from the national TV/Radio, social networks, and foreign satellite channels, respectively. CONCLUSIONS: Encouragement of people to observe preventive measures and decreasing social stress, especially among males, lower educated people, elders, and marginalized groups, are highly recommended.

Attitudes Toward and Experience of Singles with Premarital Sex: A Population-Based Study in Shiraz, Southern Iran.

The population of Iran is young and millions of youths are at risk for unprotected sexual relationships and their consequences. This questionnaire-based study was conducted in Shiraz, southern Iran. Singles were asked about premarital sex (PMS) and sexual health issues. A total of 1076 participants (634 males, 58.9%) with a mean age of 24 ± 5.8 years participated in this study. One out of 2 singles reported PMS and 1 out of 2 singles with PMS reported multiple partners. Median age at first sexual contact was 18 years. Of all singles, 452 (41.9%) were heterosexual, 61 (5.6%) were bisexual, 366 (33.9%) were alcohol users, 252 (23.3%) were smokers, 57 (5.2%) were opium users, and 392 (36.3%) did not know about preventive methods for HIV. Of 528 singles who had PMS, 126 (23.8%) never used a condom, 223 (42.2%) used it inconsistently, and 59 (11.1%) used it mainly against sexually transmitted diseases. In the regression analysis, alcohol use was the strongest associated factor of PMS in singles (OR 4.9, 95% CI 3.3-7.4), followed by lack of religious beliefs (OR 2.3, 95% CI 1.4-3.8). As a result, the PMS situation in our setting is cause for alarm and to protect singles against the risks associated with PMS, a multidisciplinary intervention including improving access to sexual behavioral counseling centers, education about sexual health and especially condom use, abstinence from alcohol use, and commitment to religious values is urgently needed to be established by health policymakers.

Knowledge and Practice of People toward their Rights in Urban Family Physician Program: A Population-Based Study in Shiraz, Southern Iran.

BACKGROUND: Urban family physician program has been launched as a pilot in Fars and Mazandaran provinces of Iran since 2012. Attitudes of policy makers and people toward urban family physician program have become challenging. This study shows what people know and practice toward this program. METHODS: This cross-sectional population-based study was conducted by a multistage randomized sampling in Shiraz, Southern Iran. Knowledge and practice of adults toward urban family physician program were queried through filing the questionnaires. Single and multiple variable analyzes of data were performed. RESULTS: Participation rate was 1257 of 1382 (90.9%), and the mean age of the respondents was 38.1 ± 13.2 years. Of 1257, 634 (50.4%) were men and 882 (70.2%) were married. Peoples' total knowledge toward urban family physician program was 5 ± 2.7 of 19, showed that 1121 (89.2%) had a low level of knowledge. This was correlated positively and in order to being under coverage of this program (P < 0.001), being under coverage of one of the main insurance systems (P = 0.04) and being married (P = 0.002). The mean score of people's practice toward the program was 2.3 ± 0.9 of total score 7, showed that 942 (74%) had poor performance, and it was correlated positively and in order to being under coverage of this program (P < 0.001) and having higher than 1000$ monthly income (P = 0.004). Correlation of people's knowledge and practice toward the program was 24%. CONCLUSIONS: Current evidences show a low level of knowledge, poor practice and weak correlation of knowledge-practice of people toward urban family physician program.