Structured information extraction from scientific text with large language models.

Extracting structured knowledge from scientific text remains a challenging task for machine learning models. Here, we present a simple approach to joint named entity recognition and relation extraction and demonstrate how pretrained large language models (GPT-3, Llama-2) can be fine-tuned to extract useful records of complex scientific knowledge. We test three representative tasks in materials chemistry: linking dopants and host materials, cataloging metal-organic frameworks, and general composition/phase/morphology/application information extraction. Records are extracted from single sentences or entire paragraphs, and the output can be returned as simple English sentences or a more structured format such as a list of JSON objects. This approach represents a simple, accessible, and highly flexible route to obtaining large databases of structured specialized scientific knowledge extracted from research papers.

Text Mining the Literature to Inform Experiments and Rationalize Impurity Phase Formation for BiFeO3.

We used data-driven methods to understand the formation of impurity phases in BiFeO3 thin-film synthesis through the sol-gel technique. Using a high-quality dataset of 331 synthesis procedures and outcomes extracted manually from 177 scientific articles, we trained decision tree models that reinforce important experimental heuristics for the avoidance of phase impurities but ultimately show limited predictive capability. We find that several important synthesis features, identified by our model, are often not reported in the literature. To test our ability to correctly impute missing synthesis parameters, we attempted to reproduce nine syntheses from the literature with varying degrees of "missingness". We demonstrate how a text-mined dataset can be made useful by informing new controlled experiments and forming a better understanding for impurity phase formation in this complex oxide system.

An autonomous laboratory for the accelerated synthesis of novel materials.

To close the gap between the rates of computational screening and experimental realization of novel materials1,2, we introduce the A-Lab, an autonomous laboratory for the solid-state synthesis of inorganic powders. This platform uses computations, historical data from the literature, machine learning (ML) and active learning to plan and interpret the outcomes of experiments performed using robotics. Over 17 days of continuous operation, the A-Lab realized 41 novel compounds from a set of 58 targets including a variety of oxides and phosphates that were identified using large-scale ab initio phase-stability data from the Materials Project and Google DeepMind. Synthesis recipes were proposed by natural-language models trained on the literature and optimized using an active-learning approach grounded in thermodynamics. Analysis of the failed syntheses provides direct and actionable suggestions to improve current techniques for materials screening and synthesis design. The high success rate demonstrates the effectiveness of artificial-intelligence-driven platforms for autonomous materials discovery and motivates further integration of computations, historical knowledge and robotics.

Role of Intravascular Imaging in Complex Percutaneous Coronary Intervention: A Meta-Analysis of Randomized Controlled Trials.

Intravascular imaging (IVI) during percutaneous coronary intervention (PCI) has been shown to improve clinical outcomes. However, data is limited in complex PCI and the adoption remains low. We aimed to conduct a meta-analysis of all available randomized controlled trials comparing IVI with conventional angiography in patients who underwent complex PCI. The primary outcomes of interest were major adverse cardiovascular events, all-cause death, cardiovascular death, myocardial infarction, stent thrombosis, target lesion revascularization and target vessel revascularization. Random-effects model was used to calculate pooled risk ratios (RRs) and 95% confidence intervals (CIs). A total of 10 randomized controlled trials comprising 6,368 patients with 3,452 in the IVI group and 2,916 in the angiography group were included. The mean duration of follow up was 2 years, mean age was 65 years and 73% of patients were men. As compared with PCI with routine angiography, the IVI-guided PCI group had significantly lower risks of major adverse cardiovascular events (RR 0.65, 95% CI 0.56 to 0.75, p <0.00001), stent thrombosis (RR 0.57, 95% CI 0.36 to 0.92, p = 0.02), cardiovascular deaths (RR 0.46, 95% CI 0.31 to 0.68, p = 0.0001), target lesion revascularization (RR 0.61, 95% CI 0.48 to 0.78, p <0.0001) and target vessel revascularization (RR 0.62, 95% CI 0.48 to 0.80, p = 0.0003). All-cause deaths and MI were similar in the 2 groups. In conclusion, among patients who underwent complex PCI, IVI reduces adverse events, importantly stent thrombosis and repeat revascularizations, compared with angiography alone guided PCI.

Closed-loop cathode recycling in solid-state batteries enabled by supramolecular electrolytes.

Deconstructing solid-state batteries (SSBs) to physically separated cathode and solid-electrolyte particles remains intensive, as does the remanufacturing of cathodes and separators from the recovered materials. To address this challenge, we designed supramolecular organo-ionic (ORION) electrolytes that are viscoelastic solids at battery operating temperatures (-40° to 45°C) yet are viscoelastic liquids above 100°C, which enables both the fabrication of high-quality SSBs and the recycling of their cathodes at end of life. SSBs implementing ORION electrolytes alongside Li metal anodes and either LFP or NMC cathodes were operated for hundreds of cycles at 45°C with less than 20% capacity fade. Using a low-temperature solvent process, we isolated the cathode from the electrolyte and demonstrated that refurbished cells recover 90% of their initial capacity and sustain it for an additional 100 cycles with 84% capacity retention in their second life.

Duodenal Adenocarcinoma With Suspected Brain Metastasis.

Duodenal adenocarcinoma (DA) is a rare tumor. We present the case of an 84-year-old lady who presented with episodic emesis with progressive dysphagia to solids and liquids. She also noted a significant weight loss of 31kg over four months. She was reported to have multiple brain masses three months before this admission. A computed tomography (CT) scan showed a heterogeneous mass (8cm) in the left retroperitoneum, inseparable from the duodenum. Additional peritoneal nodules and enlarged retroperitoneal lymph nodes were suspicious for metastases. Esophagogastroduodenoscopy revealed extrinsic compression of the stomach by the tumor. A large friable distal duodenal mass (fourth part) partially obstructed the lumen, which was biopsied. Pathology results demonstrated high-grade dysplasia but did not confirm malignancy. The patient's carcinoembryonic antigen (CEA) was elevated, but cancer antigens (CA)125 and CA19-9 were normal. A percutaneous biopsy of the mass revealed enteric-type adenocarcinoma. Immunohistochemistry showed that the tumor was positive for caudal-type homeobox (CDX)2, negative for special AT-rich sequence-binding protein (SATB)2, and patchy positive for cytokeratin (CK)7 and CK20 staining. The collective evidence suggested a duodenal primary. The patient opted for hospice and died in three days. We lack pathological evidence, but the patient's brain masses were suspicious of metastases. This would be one of the few reported cases of DA with possible brain metastases.

An Unusual Cause of Persistent Tachycardia: Atypical Neuroleptic Malignant Syndrome.

Neuroleptic malignant syndrome (NMS) is a rare, life-threatening emergency caused more commonly by typical antipsychotics. However, unusual presentations of NMS are intermittently reported with the use of atypical antipsychotics. We present the case of a 42-year-old gentleman with schizoaffective and bipolar disorder who was admitted for change in mentation and lithium toxicity. His mentation did not improve despite being dialyzed and the resolution of lithium level to baseline. He developed persistent tachycardia and hyperthermia, initially attributed to Streptococcal infection. But despite appropriate antibiotic therapy, his clinical symptoms did not improve. An extensive workup for his neurological symptoms, including lumbar puncture, 5-hydroxy indole acetic acid urine test, and brain magnetic resonance imaging, was inconclusive of any underlying etiology. Given the suspicion of atypical NMS, bromocriptine 2.5 mg three times daily was initiated. This led to the gradual resolution of his symptoms and a return to his baseline mental status. Diagnosing atypical NMS can be challenging and must be differentiated from similar disorders. Lithium toxicity can predispose patients to develop NMS.

Association of Androgen Deprivation Therapy with Metabolic Disease in Prostate Cancer Patients: An Updated Meta-Analysis.

BACKGROUND: Androgen deprivation therapy (ADT), a backbone treatment for advanced prostate cancer (PC), is known to have a variety of metabolic side effects. We conducted an updated meta-analysis to quantify the metabolic risks of ADT. MATERIALS AND METHODS: We searched PubMed, Web of Science, and Scopus in May of 2022 for studies investigating the risk of metabolic syndrome (MetS), diabetes, and hypertension from ADT in PC patients using keywords. Only full-length studies with a control group of PC patients not on ADT were included. All results compatible with each outcome domain in each included study were sought. For included studies, relative risk (RR) was pooled using a random effects model and a trim-fill approach was used to adjust for publication bias. RESULTS: 1,846 records were screened, of which 19 were found suitable for data extraction. Five studies, including 891 patients, were evaluated for MetS as an outcome, with the random effects model showing a pooled RR of 1.60 ([95% Confidence Interval (CI), 1.06-2.42]; P=0.03) for patients on ADT while twelve studies, including 336,330 patients, examined diabetes as an outcome, and the random effects model showed a RR of 1.43 ([95% CI, 1.28-1.59]; P< 0.01). After adjustment for publication bias, ADT was associated with a 25% increased risk for diabetes but was not associated with MetS. 4 studies, including 7,051 patients, examined hypertension as an outcome, and the random effects model showed a RR of 1.30 ([95% CI, 1.08-1.55]; P=0.18) in ADT patients. CONCLUSION: In patients with PC, ADT was not associated with MetS and the association with diabetes was not as strong as previously reported. Our novel meta-analysis of hypertension showed that ADT increased the risk of hypertension by 30%. These results should be understood in the context of collaborating care between a patient's oncologist and primary care provider to optimize care.

Carotid Artery Stenting Versus Carotid Artery Endarterectomy in Asymptomatic Severe Carotid Stenosis: An Updated Meta-Analysis.

Carotid artery stenting (CAS) and carotid artery endarterectomy (CEA) are revascularization options for the management of severe carotid disease in asymptomatic patients. We aimed to compare the peri-procedural outcomes of the two modalities. A systematic review of the databases PUBMED, EBSCO, and Cochrane Library was performed. All the studies that reported periprocedural outcomes (within 30 days) in asymptomatic carotid stenosis patients were included in the meta-analysis. Random effects models with inverse-variance weighting were used to estimate pooled risk ratios (RRs) to compare the outcomes. Fifteen studies (including seven randomized controlled trials) met the inclusion criteria. A total of 15251 patients were included, out of which 6419 (42%) underwent CAS and 8832 (57.9%) underwent CEA. There was no statistical difference in the primary composite outcome of death/stroke/myocardial infarction (MI) (RR 1.02, 95% CI [0.69-1.51], p 0.93). No difference was found in the secondary outcome of all-cause mortality. CAS was associated with a slightly lower risk of MI and cranial nerve palsy. CAS was associated with a slightly higher risk of stroke with no difference in the occurrence of disabling stroke or ipsilateral stroke. In general terms, the study confirms equipoise in the two treatment strategies with a higher risk of MI and cranial nerve palsy with CEA and a higher risk of non-disabling stroke with CAS.

Optical emissivity dataset of multi-material heterogeneous designs generated with automated figure extraction.

Optical device design is typically an iterative optimization process based on a good initial guess from prior reports. Optical properties databases are useful in this process but difficult to compile because their parsing requires finding relevant papers and manually converting graphical emissivity curves to data tables. Here, we present two contributions: one is a dataset of thermal emissivity records with design-related parameters, and the other is a software tool for automated colored curve data extraction from scientific plots. We manually collected 64 papers with 176 figures reporting thermal emissivity and automatically retrieved 153 colored curve data records. The automated figure analysis software pipeline uses Faster R-CNN for axes and legend object detection, EasyOCR for axes numbering recognition, and k-means clustering for colored curve retrieval. Additionally, we manually extracted geometry, materials, and method information from the text to add necessary metadata to each emissivity curve. Finally, we analyzed the dataset to determine the dominant classes of emissivity curves and determine the underlying design parameters leading to a type of emissivity profile.

Screening of bimetallic electrocatalysts for water purification with machine learning.

Electrocatalysis provides a potential solution to NO3 - pollution in wastewater by converting it to innocuous N2 gas. However, materials with excellent catalytic activity are typically limited to expensive precious metals, hindering their commercial viability. In response to this challenge, we have conducted the most extensive computational search to date for electrocatalysts that can facilitate NO3 - reduction reaction, starting with 59 390 candidate bimetallic alloys from the Materials Project and Automatic-Flow databases. Using a joint machine learning- and computation-based screening strategy, we evaluated our candidates based on corrosion resistance, catalytic activity, N2 selectivity, cost, and the ability to synthesize. We found that only 20 materials will satisfy all criteria in our screening strategy, all of which contain varying amounts of Cu. Our proposed list of candidates is consistent with previous materials investigated in the literature, with the exception of Cu-Co and Cu-Ag based compounds that merit further investigation.

Machine-Learning Rationalization and Prediction of Solid-State Synthesis Conditions.

There currently exist no quantitative methods to determine the appropriate conditions for solid-state synthesis. This not only hinders the experimental realization of novel materials but also complicates the interpretation and understanding of solid-state reaction mechanisms. Here, we demonstrate a machine-learning approach that predicts synthesis conditions using large solid-state synthesis data sets text-mined from scientific journal articles. Using feature importance ranking analysis, we discovered that optimal heating temperatures have strong correlations with the stability of precursor materials quantified using melting points and formation energies (ΔG f , ΔH f ). In contrast, features derived from the thermodynamics of synthesis-related reactions did not directly correlate to the chosen heating temperatures. This correlation between optimal solid-state heating temperature and precursor stability extends Tamman's rule from intermetallics to oxide systems, suggesting the importance of reaction kinetics in determining synthesis conditions. Heating times are shown to be strongly correlated with the chosen experimental procedures and instrument setups, which may be indicative of human bias in the data set. Using these predictive features, we constructed machine-learning models with good performance and general applicability to predict the conditions required to synthesize diverse chemical systems.

Comparison of Percutaneous Coronary Intervention Outcomes Among Patients With Obstructive Sleep Apnea, Chronic Obstructive Pulmonary Disease Overlap, and Pickwickian Syndrome (Obesity Hypoventilation Syndrome).

Background Obstructive sleep apnea (OSA) is often present in coronary artery disease patients and confers a high risk of complications following percutaneous coronary interventions (PCI). The impact of two commonly associated comorbid conditions, chronic obstructive pulmonary disease (COPD) and obesity hypoventilation syndrome (OHS, Pickwickian syndrome) in OSA patients undergoing PCI has never been studied. Methods The National Inpatient Sample (NIS; 2007-2014) was queried using the International Classification of Diseases, Clinical Modification 9 (ICD-9-CM) codes to compare baseline characteristics, comorbidities, and outcomes in adults undergoing PCI with OSA, COPD-overlap syndrome, and OSA+OHS. Results Of a total of 4,792,177 PCI-related inpatient encounters, OSA, OSA-COPD overlap syndrome, and OSA+OHS were found to be present in 153,706 (median age 62 years, 79.4% male), 65135 (median age 65 years, 66.0% male), and 2291 (median age 63 years, 58.2% males) patients, respectively. The OHS+OSA cohort, when compared to the COPD-OSA and OSA cohorts, was found to have the worst outcomes in terms of all-cause mortality (2.8% vs. 1.5% vs. 1.1%), hospital stay (median 6 vs. 3 vs. 2 days), hospital charges ($147, 209 vs. $101,416 vs. $87,983). Complications, including cardiogenic shock (7.3% vs. 3.4% vs. 2.6%), post-procedural myocardial infarction (11.2% vs. 7.1% vs. 6.0%), iatrogenic cardiac complications (6.1% vs. 3.5% vs. 3.7%), respiratory failure, acute kidney injury, infections, and pulmonary embolism, were also significantly higher in patients with OHS+OSA. Adjusted multivariable analysis revealed equivalent results with OHS+OSA having worse outcomes than OSA-COPD and OSA. Conclusion Concomitant OHS and COPD were linked to worse clinical outcomes in patients with OSA undergoing PCI. Future prospective studies are warranted to fully understand related pathophysiology, evaluate and validate long-term outcomes, and formulate effective preventive and management strategies.

Length of stay and cost of care associated with admissions for atrial fibrillation among patients with cancer.

BACKGROUND: The aim of this study is to assess the burden of AF-related hospitalizations inclusive of inflation-adjusted cost-of-care and length-of-stay (LOS) among cancer patients and the impact of direct current cardioversion (DCCV) on these outcomes. METHODS: Using the National Inpatient Sample (NIS), patients hospitalized with either a primary or secondary diagnosis of AF and comorbid cancer were identified and both cost of hospitalization and LOS were evaluated for each group. Subgroup analyses were performed for specific cancer types (breast, lung, colon, prostate and lymphoma), and those receiving DCCV. RESULTS: The prevalence of co-morbid AF was 8.2 million (16%) and 35.5 million (10%) among those with vs. those without cancer, respectively (odds ratio = 1.6, 95% confidence interval = 1.5-1.7; P < 0.001). Over time, both primary and prevalent AF admissions among those with comorbid cancer increased from 1.1% and 12.3% in 2003 to 1.5% and 21% in 2015, respectively. The total cost of hospitalization increased 94.4% among those with AF and comorbid cancer compared to 23.9% among those without cancer. Among the subgroup of patients with comorbid cancer and primary admission for AF undergoing DCCV, length of stay (2.7 vs. 2.2 days; P < 0.001, model 1) and cost of care ($7,093 vs. 6,152; P < 0.001) were both significantly higher. CONCLUSIONS: AF related admissions are increasing for all populations especially amongst those patients with a comorbid diagnosis of cancer, including all cancer subtypes evaluated. Among those patients who underwent DCCV, cancer patients had longer length of stay and increased health care costs.

Text-mined dataset of gold nanoparticle synthesis procedures, morphologies, and size entities.

Gold nanoparticles are highly desired for a range of technological applications due to their tunable properties, which are dictated by the size and shape of the constituent particles. Many heuristic methods for controlling the morphological characteristics of gold nanoparticles are well known. However, the underlying mechanisms controlling their size and shape remain poorly understood, partly due to the immense range of possible combinations of synthesis parameters. Data-driven methods can offer insight to help guide understanding of these underlying mechanisms, so long as sufficient synthesis data are available. To facilitate data mining in this direction, we have constructed and made publicly available a dataset of codified gold nanoparticle synthesis protocols and outcomes extracted directly from the nanoparticle materials science literature using natural language processing and text-mining techniques. This dataset contains 5,154 data records, each representing a single gold nanoparticle synthesis article, filtered from a database of 4,973,165 publications. Each record contains codified synthesis protocols and extracted morphological information from a total of 7,608 experimental and 12,519 characterization paragraphs.

Quantifying the advantage of domain-specific pre-training on named entity recognition tasks in materials science.

A bottleneck in efficiently connecting new materials discoveries to established literature has arisen due to an increase in publications. This problem may be addressed by using named entity recognition (NER) to extract structured summary-level data from unstructured materials science text. We compare the performance of four NER models on three materials science datasets. The four models include a bidirectional long short-term memory (BiLSTM) and three transformer models (BERT, SciBERT, and MatBERT) with increasing degrees of domain-specific materials science pre-training. MatBERT improves over the other two BERTBASE-based models by 1%∼12%, implying that domain-specific pre-training provides measurable advantages. Despite relative architectural simplicity, the BiLSTM model consistently outperforms BERT, perhaps due to its domain-specific pre-trained word embeddings. Furthermore, MatBERT and SciBERT models outperform the original BERT model to a greater extent in the small data limit. MatBERT's higher-quality predictions should accelerate the extraction of structured data from materials science literature.

Thermal fluids with high specific heat capacity through reversible Diels-Alder reactions.

Thermal fluids are used as heat transfer fluids and thermal energy storage media in many energy technologies ranging from solar thermal heating to battery thermal management. The heat capacity of state-of-the-art thermal fluids remains ∼50% of that of water (which suffers from a limited operation range between 0°C and 100°C), and their viscosities are typically more than one order of magnitude higher than that of water. Our results demonstrate that the heat capacity of the proposed thermochemical fluid is significantly higher than that of state-of-the-art thermal fluids over a broad temperature range and is also higher than that of water between 60°C and 90°C. The viscosity of our liquid is only 3 times higher than that of water, and the operating temperature range is between -90°C and 135°C. Furthermore, a model was developed allowing for novel design of thermochemical thermal fluids in the future with even higher heat capacity.

Experimental validation of high thermoelectric performance in RECuZnP2 predicted by high-throughput DFT calculations.

Accurate density functional theory calculations of the interrelated properties of thermoelectric materials entail high computational cost, especially as crystal structures increase in complexity and size. New methods involving ab initio scattering and transport (AMSET) and compressive sensing lattice dynamics are used to compute the transport properties of quaternary CaAl2Si2-type rare-earth phosphides RECuZnP2 (RE = Pr, Nd, Er), which were identified to be promising thermoelectrics from high-throughput screening of 20 000 disordered compounds. Experimental measurements of the transport properties agree well with the computed values. Compounds with stiff bulk moduli (>80 GPa) and high speeds of sound (>3500 m s-1) such as RECuZnP2 are typically dismissed as thermoelectric materials because they are expected to exhibit high lattice thermal conductivity. However, RECuZnP2 exhibits not only low electrical resistivity, but also low lattice thermal conductivity (∼1 W m-1 K-1). Contrary to prior assumptions, polar-optical phonon scattering was revealed by AMSET to be the primary mechanism limiting the electronic mobility of these compounds, raising questions about existing assumptions of scattering mechanisms in this class of thermoelectric materials. The resulting thermoelectric performance (zT of 0.5 for ErCuZnP2 at 800 K) is among the best observed in phosphides and can likely be improved with further optimization.

Multifunctional Dental Composite with Piezoelectric Nanofillers for Combined Antibacterial and Mineralization Effects.

After nearly seven decades of development, dental composite restorations continue to show limited clinical service. The triggering point for restoration failure is the degradation of the bond at the tooth-biomaterial interface from chemical, biological, and mechanical sources. Oral biofilms form at the bonded interfaces, producing enzymes and acids that demineralize hard tissues and damage the composite. Removing bacteria from bonded interfaces and remineralizing marginal gaps will increase restorations' clinical service. To address this need, we propose for the first time the use of piezoelectric nanoparticles of barium titanate (BaTiO3) as a multifunctional bioactive filler in dental resin composites, offering combined antibacterial and (re)mineralization effects. In this work, we developed and characterized the properties of dental piezoelectric resin composites, including the degree of conversion and mechanical and physical properties, for restorative applications. Moreover, we evaluated the antibacterial and mineralization responses of piezoelectric composites in vitro. We observed a significant reduction in biofilm growth (up to 90%) and the formation of thick and dense layers of calcium phosphate minerals in piezoelectric composites compared to control groups. The antibacterial mechanism was also revealed. Additionally, we developed a unique approach evaluating the bond strength of dentin-adhesive-composite interfaces subjected to simultaneous attacks from bacteria and cyclic mechanical loading operating in synergy. Our innovative bioactive multifunctional composite provides an ideal technology for restorative applications using a single filler with combined long-lasting nonrechargeable antibacterial/remineralization effects.

A framework for quantifying uncertainty in DFT energy corrections.

In this work, we demonstrate a method to quantify uncertainty in corrections to density functional theory (DFT) energies based on empirical results. Such corrections are commonly used to improve the accuracy of computational enthalpies of formation, phase stability predictions, and other energy-derived properties, for example. We incorporate this method into a new DFT energy correction scheme comprising a mixture of oxidation-state and composition-dependent corrections and show that many chemical systems contain unstable polymorphs that may actually be predicted stable when uncertainty is taken into account. We then illustrate how these uncertainties can be used to estimate the probability that a compound is stable on a compositional phase diagram, thus enabling better-informed assessments of compound stability.