Generative Artificial Intelligence for Chest Radiograph Interpretation in the Emergency Department.

Importance: Multimodal generative artificial intelligence (AI) methodologies have the potential to optimize emergency department care by producing draft radiology reports from input images. Objective: To evaluate the accuracy and quality of AI-generated chest radiograph interpretations in the emergency department setting. Design, Setting, and Participants: This was a retrospective diagnostic study of 500 randomly sampled emergency department encounters at a tertiary care institution including chest radiographs interpreted by both a teleradiology service and on-site attending radiologist from January 2022 to January 2023. An AI interpretation was generated for each radiograph. The 3 radiograph interpretations were each rated in duplicate by 6 emergency department physicians using a 5-point Likert scale. Main Outcomes and Measures: The primary outcome was any difference in Likert scores between radiologist, AI, and teleradiology reports, using a cumulative link mixed model. Secondary analyses compared the probability of each report type containing no clinically significant discrepancy with further stratification by finding presence, using a logistic mixed-effects model. Physician comments on discrepancies were recorded. Results: A total of 500 ED studies were included from 500 unique patients with a mean (SD) age of 53.3 (21.6) years; 282 patients (56.4%) were female. There was a significant association of report type with ratings, with post hoc tests revealing significantly greater scores for AI (mean [SE] score, 3.22 [0.34]; P < .001) and radiologist (mean [SE] score, 3.34 [0.34]; P < .001) reports compared with teleradiology (mean [SE] score, 2.74 [0.34]) reports. AI and radiologist reports were not significantly different. On secondary analysis, there was no difference in the probability of no clinically significant discrepancy between the 3 report types. Further stratification of reports by presence of cardiomegaly, pulmonary edema, pleural effusion, infiltrate, pneumothorax, and support devices also yielded no difference in the probability of containing no clinically significant discrepancy between the report types. Conclusions and Relevance: In a representative sample of emergency department chest radiographs, results suggest that the generative AI model produced reports of similar clinical accuracy and textual quality to radiologist reports while providing higher textual quality than teleradiologist reports. Implementation of the model in the clinical workflow could enable timely alerts to life-threatening pathology while aiding imaging interpretation and documentation.

Probing the Local Electronic Structure in Metal Halide Perovskites through Cobalt Substitution.

Owing to the unique chemical and electronic properties arising from 3d-electrons, substitution with transition metal ions is one of the key routes for engineering new functionalities into materials. While this approach has been used extensively in complex metal oxide perovskites, metal halide perovskites have largely resisted facile isovalent substitution. In this work, it is demonstrated that the substitution of Co2+ into the lattice of methylammonium lead triiodide imparts magnetic behavior to the material while maintaining photovoltaic performance at low concentrations. In addition to comprehensively characterizing its magnetic properties, the Co2+ ions themselves are utilized as probes to sense the local electronic environment of Pb in the perovskite, thereby revealing the nature of their incorporation into the material. A comprehensive understanding of the effect of transition metal incorporation is provided, thereby opening the substitution gateway for developing novel functional perovskite materials and devices for future technologies.

An artificial intelligence platform provides an accurate interpretation of esophageal motility from Functional Lumen Imaging Probe Panometry studies.

BACKGROUND: Functional lumen imaging probe (FLIP) Panometry is performed at the time of sedated endoscopy and evaluates esophageal motility in response to distension. This study aimed to develop and test an automated artificial intelligence (AI) platform that could interpret FLIP Panometry studies. METHODS: The study cohort included 678 consecutive patients and 35 asymptomatic controls that completed FLIP Panometry during endoscopy and high-resolution manometry (HRM). "True" study labels for model training and testing were assigned by experienced esophagologists per a hierarchical classification scheme. The supervised, deep learning, AI model generated FLIP Panometry heatmaps from raw FLIP data and based on convolutional neural networks assigned esophageal motility labels using a two-stage prediction model. Model performance was tested on a 15% held-out test set (n = 103); the remainder of the studies were utilized for model training (n = 610). KEY RESULTS: "True" FLIP labels across the entire cohort included 190 (27%) "normal," 265 (37%) "not normal/not achalasia," and 258 (36%) "achalasia." On the test set, both the Normal/Not normal and the achalasia/not achalasia models achieved an accuracy of 89% (with 89%/88% recall, 90%/89% precision, respectively). Of 28 patients with achalasia (per HRM) in the test set, 0 were predicted as "normal" and 93% as "achalasia" by the AI model. CONCLUSIONS: An AI platform provided accurate interpretation of FLIP Panometry esophageal motility studies from a single center compared with the impression of experienced FLIP Panometry interpreters. This platform may provide useful clinical decision support for esophageal motility diagnosis from FLIP Panometry studies performed at the time of endoscopy.

Deep learning-based artificial intelligence model for identifying swallow types in esophageal high-resolution manometry.

BACKGROUND: This study aimed to build and evaluate a deep learning, artificial intelligence (AI) model to automatically classify swallow types based on raw data from esophageal high-resolution manometry (HRM). METHODS: HRM studies on patients with no history of esophageal surgery were collected including 1,741 studies with 26,115 swallows labeled by swallow type (normal, hypercontractile, weak-fragmented, failed, and premature) by an expert interpreter per the Chicago Classification. The dataset was stratified and split into train/validation/test datasets for model development. Long short-term memory (LSTM), a type of deep-learning AI model, was trained and evaluated. The overall performance and detailed per-swallow type performance were analyzed. The interpretations of the supine swallows in a single study were further used to generate an overall classification of peristalsis. KEY RESULTS: The LSTM model for swallow type yielded accuracies from the train/validation/test datasets of 0.86/0.81/0.83. The model's interpretation for study-level classification of peristalsis yielded accuracy of 0.88 in the test dataset. Among model misclassification, 535/698 (77%) swallows and 25/35 (71%) studies were to adjacent categories, for example, normal to weak or normal to ineffective, respectively. CONCLUSIONS AND INFERENCES: A deep-learning AI model can automatically and accurately identify the Chicago Classification swallow types and peristalsis classification from raw HRM data. While future work to refine this model and incorporate overall manometric diagnoses are needed, this study demonstrates the role that AI will serve in the interpretation and classification of esophageal HRM studies.

Amalgamation of cloud-based colonoscopy videos with patient-level metadata to facilitate large-scale machine learning.

Background and study aims Storage of full-length endoscopic procedures is becoming increasingly popular. To facilitate large-scale machine learning (ML) focused on clinical outcomes, these videos must be merged with the patient-level data in the electronic health record (EHR). Our aim was to present a method of accurately linking patient-level EHR data with cloud stored colonoscopy videos. Methods This study was conducted at a single academic medical center. Most procedure videos are automatically uploaded to the cloud server but are identified only by procedure time and procedure room. We developed and then tested an algorithm to match recorded videos with corresponding exams in the EHR based upon procedure time and room and subsequently extract frames of interest. Results Among 28,611 total colonoscopies performed over the study period, 21,170 colonoscopy videos in 20,420 unique patients (54.2 % male, median age 58) were matched to EHR data. Of 100 randomly sampled videos, appropriate matching was manually confirmed in all. In total, these videos represented 489,721 minutes of colonoscopy performed by 50 endoscopists (median 214 colonoscopies per endoscopist). The most common procedure indications were polyp screening (47.3 %), surveillance (28.9 %) and inflammatory bowel disease (9.4 %). From these videos, we extracted procedure highlights (identified by image capture; mean 8.5 per colonoscopy) and surrounding frames. Conclusions We report the successful merging of a large database of endoscopy videos stored with limited identifiers to rich patient-level data in a highly accurate manner. This technique facilitates the development of ML algorithms based upon relevant patient outcomes.

Microsecond Carrier Lifetimes, Controlled p-Doping, and Enhanced Air Stability in Low-Bandgap Metal Halide Perovskites.

Mixed lead-tin halide perovskites have sufficiently low bandgaps (∼1.2 eV) to be promising absorbers for perovskite-perovskite tandem solar cells. Previous reports on lead-tin perovskites have typically shown poor optoelectronic properties compared to neat lead counterparts: short photoluminescence lifetimes (<100 ns) and low photoluminescence quantum efficiencies (<1%). Here, we obtain films with carrier lifetimes exceeding 1 µs and, through addition of small quantities of zinc iodide to the precursor solutions, photoluminescence quantum efficiencies under solar illumination intensities of 2.5%. The zinc additives also substantially enhance the film stability in air, and we use cross-sectional chemical mapping to show that this enhanced stability is because of a reduction in tin-rich clusters. By fabricating field-effect transistors, we observe that the introduction of zinc results in controlled p-doping. Finally, we show that zinc additives also enhance power conversion efficiencies and the stability of solar cells. Our results demonstrate substantially improved low-bandgap perovskites for solar cells and versatile electronic applications.

Charge-Carrier Cooling and Polarization Memory Loss in Formamidinium Tin Triiodide.

Reports of slow charge-carrier cooling in hybrid metal halide perovskites have prompted hopes of achieving higher photovoltaic cell voltages through hot-carrier extraction. However, observations of long-lived hot charge carriers even at low photoexcitation densities and an orders-of-magnitude spread in reported cooling times have been challenging to explain. Here we present ultrafast time-resolved photoluminescence measurements on formamidinum tin triiodide, showing fast initial cooling over tens of picoseconds and demonstrating that a perceived secondary regime of slower cooling instead derives from electronic relaxation, state-filling, and recombination in the presence of energetic disorder. We identify limitations of some widely used approaches to determine charge-carrier temperature and make use of an improved model which accounts for the full photoluminescence line shape. Further, we do not find any persistent polarization anisotropy in FASnI3 within 270 fs after excitation, indicating that excited carriers rapidly lose both polarization memory and excess energy through interactions with the perovskite lattice.

The Effects of Doping Density and Temperature on the Optoelectronic Properties of Formamidinium Tin Triiodide Thin Films.

Optoelectronic properties are unraveled for formamidinium tin triiodide (FASnI3 ) thin films, whose background hole doping density is varied through SnF2 addition during film fabrication. Monomolecular charge-carrier recombination exhibits both a dopant-mediated part that grows linearly with hole doping density and remnant contributions that remain under tin-enriched processing conditions. At hole densities near 1020 cm-3 , a strong Burstein-Moss effect increases absorption onset energies by ≈300 meV beyond the bandgap energy of undoped FASnI3 (shown to be 1.2 eV at 5 K and 1.35 eV at room temperature). At very high doping densities (1020 cm-3 ), temperature-dependent measurements indicate that the effective charge-carrier mobility is suppressed through scattering with ionized dopants. Once the background hole concentration is nearer 1019 cm-3 and below, the charge-carrier mobility increases with decreasing temperature according to ≈T-1.2 , suggesting that it is limited mostly by intrinsic interactions with lattice vibrations. For the lowest doping concentration of 7.2 × 1018 cm-3 , charge-carrier mobilities reach a value of 67 cm2 V-1 s-1 at room temperature and 470 cm2 V-1 s-1 at 50 K. Intraexcitonic transitions observed in the THz-frequency photoconductivity spectra at 5 K reveal an exciton binding energy of only 3.1 meV for FASnI3 , in agreement with the low bandgap energy exhibited by this perovskite.

Fractional deviations in precursor stoichiometry dictate the properties, performance and stability of perovskite photovoltaic devices.

The last five years have witnessed remarkable progress in the field of lead halide perovskite materials and devices. Examining the existing body of literature reveals staggering inconsistencies in the reported results among different research groups with a particularly wide spread in the photovoltaic performance and stability of devices. In this work we demonstrate that fractional, quite possibly unintentional, deviations in the precursor solution stoichiometry can cause significant changes in the properties of the perovskite layer as well as in the performance and stability of perovskite photovoltaic devices. We show that while the absorbance and morphology of the layers remain largely unaffected, the surface composition and energetics, crystallinity, emission efficiency, energetic disorder and storage stability are all very sensitive to the precise stoichiometry of the precursor solution. Our results elucidate the origin of the irreproducibility and inconsistencies of reported results among different groups as well as the wide spread in device performance even within individual studies. Finally, we propose a simple experimental method to identify the exact stoichiometry of the perovskite layer that researchers can employ to confirm their experiments are performed consistently without unintentional variations in precursor stoichiometry.

A Universal Deposition Protocol for Planar Heterojunction Solar Cells with High Efficiency Based on Hybrid Lead Halide Perovskite Families.

A robust and expedient gas quenching method is developed for the solution deposition of hybrid perovskite thin films. The method offers a reliable standard practice for the fabrication of a non-exhaustive variety of perovskites exhibiting excellent film morphology and commensurate high performance in both regular and inverted structured solar cell architectures.

Assembly of viral hydrogels for three-dimensional conducting nanocomposites.

M13 bacteriophages act as versatile scaffolds capable of organizing single-walled carbon nanotubes and fabricating three-dimensional conducting nanocomposites. The morphological, electrical, and electrochemical properties of the nanocomposites are presented, as well as its ability to disperse and utilize single-walled carbon nanotubes effectively.

Assembly of a bacteriophage-based template for the organization of materials into nanoporous networks.

M13 bacteriophages are assembled via a covalent layer-by-layer process to form a highly nanoporous network capable of organizing nanoparticles and acting as a scaffold for templating metal-oxides. The morphological and optical properties of the film itself are presented as well as its ability to organize and disperse metal nanoparticles.

Versatile three-dimensional virus-based template for dye-sensitized solar cells with improved electron transport and light harvesting.

By genetically encoding affinity for inorganic materials into the capsid proteins of the M13 bacteriophage, the virus can act as a template for the synthesis of nanomaterial composites for use in various device applications. Herein, the M13 bacteriophage is employed to build a multifunctional and three-dimensional scaffold capable of improving both electron collection and light harvesting in dye-sensitized solar cells (DSSCs). This has been accomplished by binding gold nanoparticles (AuNPs) to the virus proteins and encapsulating the AuNP-virus complexes in TiO2 to produce a plasmon-enhanced and nanowire (NW)-based photoanode. The NW morphology exhibits an improved electron diffusion length compared to traditional nanoparticle-based DSSCs, and the AuNPs increase the light absorption of the dye-molecules through the phenomenon of localized surface plasmon resonance. Consequently, we report a virus-templated and plasmon-enhanced DSSC with an efficiency of 8.46%, which is achieved through optimizing both the NW morphology and the concentration of AuNPs loaded into the solar cells. In addition, we propose a theoretical model that predicts the experimentally observed trends of plasmon enhancement.

Tunable localized surface plasmon-enabled broadband light-harvesting enhancement for high-efficiency panchromatic dye-sensitized solar cells.

In photovoltaic devices, light harvesting (LH) and carrier collection have opposite relations with the thickness of the photoactive layer, which imposes a fundamental compromise for the power conversion efficiency (PCE). Unbalanced LH at different wavelengths further reduces the achievable PCE. Here, we report a novel approach to broadband balanced LH and panchromatic solar energy conversion using multiple-core-shell structured oxide-metal-oxide plasmonic nanoparticles. These nanoparticles feature tunable localized surface plasmon resonance frequencies and the required thermal stability during device fabrication. By simply blending the plasmonic nanoparticles with available photoactive materials, the broadband LH of practical photovoltaic devices can be significantly enhanced. We demonstrate a panchromatic dye-sensitized solar cell with an increased PCE from 8.3% to 10.8%, mainly through plasmon-enhanced photoabsorption in the otherwise less harvested region of solar spectrum. This general and simple strategy also highlights easy fabrication, and may benefit solar cells using other photoabsorbers or other types of solar-harvesting devices.

Optical manipulation of nanoparticles and biomolecules in sub-wavelength slot waveguides.

The ability to manipulate nanoscopic matter precisely is critical for the development of active nanosystems. Optical tweezers are excellent tools for transporting particles ranging in size from several micrometres to a few hundred nanometres. Manipulation of dielectric objects with much smaller diameters, however, requires stronger optical confinement and higher intensities than can be provided by these diffraction-limited systems. Here we present an approach to optofluidic transport that overcomes these limitations, using sub-wavelength liquid-core slot waveguides. The technique simultaneously makes use of near-field optical forces to confine matter inside the waveguide and scattering/adsorption forces to transport it. The ability of the slot waveguide to condense the accessible electromagnetic energy to scales as small as 60 nm allows us also to overcome the fundamental diffraction problem. We apply the approach here to the trapping and transport of 75-nm dielectric nanoparticles and lambda-DNA molecules. Because trapping occurs along a line, rather than at a point as with traditional point traps, the method provides the ability to handle extended biomolecules directly. We also carry out a detailed numerical analysis that relates the near-field optical forces to release kinetics. We believe that the architecture demonstrated here will help to bridge the gap between optical manipulation and nanofluidics.