Detection of Personal and Family History of Suicidal Thoughts and Behaviors using Deep Learning and Natural Language Processing: A Multi-Site Study.

Objective: Personal and family history of suicidal thoughts and behaviors (PSH and FSH, respectively) are significant risk factors associated with future suicide events. These are often captured in narrative clinical notes in electronic health records (EHRs). Collaboratively, Weill Cornell Medicine (WCM), Northwestern Medicine (NM), and the University of Florida (UF) developed and validated deep learning (DL)-based natural language processing (NLP) tools to detect PSH and FSH from such notes. The tool's performance was further benchmarked against a method relying exclusively on ICD-9/10 diagnosis codes. Materials and Methods: We developed DL-based NLP tools utilizing pre-trained transformer models Bio_ClinicalBERT and GatorTron, and compared them with expert-informed, rule-based methods. The tools were initially developed and validated using manually annotated clinical notes at WCM. Their portability and performance were further evaluated using clinical notes at NM and UF. Results: The DL tools outperformed the rule-based NLP tool in identifying PSH and FHS. For detecting PSH, the rule-based system obtained an F1-score of 0.75 ± 0.07, while the Bio_ClinicalBERT and GatorTron DL tools scored 0.83 ± 0.09 and 0.84 ± 0.07, respectively. For detecting FSH, the rule-based NLP tool's F1-score was 0.69 ± 0.11, compared to 0.89 ± 0.10 for Bio_ClinicalBERT and 0.92 ± 0.07 for GatorTron. For the gold standard corpora across the three sites, only 2.2% (WCM), 9.3% (NM), and 7.8% (UF) of patients reported to have an ICD-9/10 diagnosis code for suicidal thoughts and behaviors prior to the clinical notes report date. The best performing GatorTron DL tool identified 93.0% (WCM), 80.4% (NM), and 89.0% (UF) of patients with documented PSH, and 85.0%(WCM), 89.5%(NM), and 100%(UF) of patients with documented FSH in their notes. Discussion: While PSH and FSH are significant risk factors for future suicide events, little effort has been made previously to identify individuals with these history. To address this, we developed a transformer based DL method and compared with conventional rule-based NLP approach. The varying effectiveness of the rule-based tools across sites suggests a need for improvement in its dictionary-based approach. In contrast, the performances of the DL tools were higher and comparable across sites. Furthermore, DL tools were fine-tuned using only small number of annotated notes at each site, underscores its greater adaptability to local documentation practices and lexical variations. Conclusion: Variations in local documentation practices across health care systems pose challenges to rule-based NLP tools. In contrast, the developed DL tools can effectively extract PSH and FSH information from unstructured clinical notes. These tools will provide clinicians with crucial information for assessing and treating patients at elevated risk for suicide who are rarely been diagnosed.

Machine learning identification of patient clusters for cutaneous melanoma.

BACKGROUND: Cutaneous melanoma is a cancer arising in melanocyte skin cells and is the deadliest form of skin cancer worldwide. Although some risk factors are known, accurate prediction of disease progression and probability for metastasis are difficult to ascertain, given the complexity of the disease and the absence of reliable predictive markers. Since early detection and treatment are essential to enhance survival, this study utilizing machine learning (ML) aims to further delineate additional risk factors associated with cutaneous melanoma. METHODS: A Bayesian Gaussian Mixture ML model was created with data from 2056 patients diagnosed with cutaneous melanoma and then used to group the patients into six Clusters based on a Silhouette Score analysis. A t-distributed stochastic neighbor embedding (t-SNE) model was used to visualize the six Clusters. RESULTS: Statistical analysis revealed that Cluster 4 showed a significantly higher rate of metastatic disease, as well as higher Breslow depth at diagnosis, compared to the other five Clusters. Compared to the other five Clusters, patients represented in Cluster 4 also had lower healthcare utilization, fewer dermatology clinic visits, fewer primary care providers, and less frequent colonoscopies and mammograms, and were more likely to smoke and less likely to have a prior diagnosis of basal cell carcinoma. CONCLUSIONS: This study uncovers gaps in healthcare utilization of services among patient groups with cutaneous melanoma as well as possible implications for management of disease progression. Data-driven analyses emphasize the importance of routine clinic visits to dermatologists and/or primary care physicians (PCPs) for early detection and management of cutaneous melanoma. The findings from this study demonstrate that unsupervised ML methodology may serve to define the best candidate patients to benefit from enhanced dermatology/primary care which, in turn, is expected to improve outcomes for cutaneous melanoma.

Geographic Disparities in Re-triage Destinations Among Seriously Injured Californians.

Objective: To quantify geographic disparities in sub-optimal re-triage of seriously injured patients in California. Summary of Background Data: Re-triage is the emergent transfer of seriously injured patients from the emergency departments of non-trauma and low-level trauma centers to, ideally, high-level trauma centers. Some patients are re-triaged to a second non-trauma or low-level trauma center (sub-optimal) instead of a high-level trauma center (optimal). Methods: This was a retrospective observational cohort study of seriously injured patients, defined by an Injury Severity Score > 15, re-triaged in California (2009-2018). Re-triages within one day of presentation to the sending center were considered. The sub-optimal re-triage rate was quantified at the state, regional trauma coordinating committees (RTCC), local emergency medical service agencies, and sending center level. A generalized linear mixed-effects regression quantified the association of sub-optimality with the RTCC of the sending center. Geospatial analyses demonstrated geographic variations in sub-optimal re-triage rates and calculated alternative re-triage destinations. Results: There were 8,882 re-triages of seriously injured patients and 2,680 (30.2 %) were sub-optimal. Sub-optimally re-triaged patients had 1.5 higher odds of transfer to a third short-term acute care hospital and 1.25 increased odds of re-admission within 60 days from discharge. The sub-optimal re-triage rates increased from 29.3 % in 2009 to 38.6 % in 2018. 56.0 % of non-trauma and low-level trauma centers had at least one sub-optimal re-triage. The Southwest RTCC accounted for the largest proportion (39.8 %) of all sub-optimal re-triages in California. Conclusion: High population density geographic areas experienced higher sub-optimal re-triage rates.

Re-triage moderates association between state trauma funding and lower mortality of trauma patients.

BACKGROUND: Severely injured patients who are re-triaged (emergently transferred from an emergency department to a high-level trauma center) experience lower in-hospital mortality. Patients in states with trauma funding also experience lower in-hospital mortality. This study examines the interaction of re-triage, state trauma funding, and in-hospital mortality. STUDY DESIGN: Severely injured patients (Injury Severity Score (ISS) >15) were identified from 2016 to 2017 Healthcare Cost and Utilization Project State Emergency Department Databases and State Inpatient Databases in five states (FL, MA, MD, NY, WI). Data were merged with the American Hospital Association Annual Survey and state trauma funding data. Patients were linked across hospital encounters to determine if they were appropriately field triaged, field under-triaged, optimally re-triaged, or sub-optimally re-triaged. A hierarchical logistic regression modeling in-hospital mortality was used to quantify the effect of re-triage on the association between state trauma funding and in-hospital mortality, while adjusting for patient and hospital characteristics. RESULTS: A total of 241,756 severely injured patients were identified. Median age was 52 years (IQR: 28, 73) and median ISS was 17 (IQR: 16, 25). Two states (MA, NY) allocated no funding, while three states (WI, FL, MD) allocated $0.09-$1.80 per capita. Patients in states with trauma funding were more broadly distributed across trauma center levels, with a higher proportion of patients brought to Level III, IV, or non-trauma centers, compared to patients in states without trauma funding (54.0% vs. 41.1%, p < 0.001). Patients in states with trauma funding were more often re-triaged, compared to patients in states without trauma funding (3.7% vs. 1.8%, p < 0.001). Patients who were optimally re-triaged in states with trauma funding experienced 0.67 lower adjusted odds of in-hospital mortality (95% CI: 0.50-0.89), compared to patients in states without trauma funding. We found that re-triage significantly moderated the association between state trauma funding and lower in-hospital mortality (p = 0.018). CONCLUSION: Severely injured patients in states with trauma funding are more often re-triaged and experience lower odds of mortality. Re-triage of severely injured patients may potentiate the mortality benefit of increased state trauma funding.

Evaluation of the portability of computable phenotypes with natural language processing in the eMERGE network.

The electronic Medical Records and Genomics (eMERGE) Network assessed the feasibility of deploying portable phenotype rule-based algorithms with natural language processing (NLP) components added to improve performance of existing algorithms using electronic health records (EHRs). Based on scientific merit and predicted difficulty, eMERGE selected six existing phenotypes to enhance with NLP. We assessed performance, portability, and ease of use. We summarized lessons learned by: (1) challenges; (2) best practices to address challenges based on existing evidence and/or eMERGE experience; and (3) opportunities for future research. Adding NLP resulted in improved, or the same, precision and/or recall for all but one algorithm. Portability, phenotyping workflow/process, and technology were major themes. With NLP, development and validation took longer. Besides portability of NLP technology and algorithm replicability, factors to ensure success include privacy protection, technical infrastructure setup, intellectual property agreement, and efficient communication. Workflow improvements can improve communication and reduce implementation time. NLP performance varied mainly due to clinical document heterogeneity; therefore, we suggest using semi-structured notes, comprehensive documentation, and customization options. NLP portability is possible with improved phenotype algorithm performance, but careful planning and architecture of the algorithms is essential to support local customizations.

Extracting social determinants of health from electronic health records using natural language processing: a systematic review.

OBJECTIVE: Social determinants of health (SDoH) are nonclinical dispositions that impact patient health risks and clinical outcomes. Leveraging SDoH in clinical decision-making can potentially improve diagnosis, treatment planning, and patient outcomes. Despite increased interest in capturing SDoH in electronic health records (EHRs), such information is typically locked in unstructured clinical notes. Natural language processing (NLP) is the key technology to extract SDoH information from clinical text and expand its utility in patient care and research. This article presents a systematic review of the state-of-the-art NLP approaches and tools that focus on identifying and extracting SDoH data from unstructured clinical text in EHRs. MATERIALS AND METHODS: A broad literature search was conducted in February 2021 using 3 scholarly databases (ACL Anthology, PubMed, and Scopus) following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A total of 6402 publications were initially identified, and after applying the study inclusion criteria, 82 publications were selected for the final review. RESULTS: Smoking status (n = 27), substance use (n = 21), homelessness (n = 20), and alcohol use (n = 15) are the most frequently studied SDoH categories. Homelessness (n = 7) and other less-studied SDoH (eg, education, financial problems, social isolation and support, family problems) are mostly identified using rule-based approaches. In contrast, machine learning approaches are popular for identifying smoking status (n = 13), substance use (n = 9), and alcohol use (n = 9). CONCLUSION: NLP offers significant potential to extract SDoH data from narrative clinical notes, which in turn can aid in the development of screening tools, risk prediction models, and clinical decision support systems.

Divergent Trends in Emergency Department Presentations Amid the Novel Coronavirus Disease 2019 Pandemic in Chicago, Illinois.

Excess deaths during the coronavirus disease 2019 (COVID-19) pandemic have been largely attributed to cardiovascular disease (CVD); however, patterns in CVD hospitalizations after the first surge of the pandemic have not well-documented. Our brief report, examining trends in health care avoidance documents that CVD hospitalizations decreased in Chicago before significant burden of COVID-19 cases or deaths and normalized during the first COVID-19 surge. These data may help to inform health care systems responses in the coming months while mobilizing vaccinations to the population at large.

The effect of number of healthcare visits on study sample selection in electronic health record data.

INTRODUCTION: Few studies have addressed how to select a study sample when using electronic health record (EHR) data. OBJECTIVE: To examine how changing criterion for number of visits in EHR data required for inclusion in a study sample would impact one basic epidemiologic measure: estimates of disease period prevalence. METHODS: Year 2016 EHR data from three Midwestern health systems (Northwestern Medicine in Illinois, University of Iowa Health Care, and Froedtert & the Medical College of Wisconsin, all regional tertiary health care systems including hospitals and clinics) was used to examine how alternate definitions of the study sample, based on number of healthcare visits in one year, affected measures of disease period prevalence. In 2016, each of these health systems saw between 160,000 and 420,000 unique patients. Curated collections of ICD-9, ICD-10, and SNOMED codes (from CMS-approved electronic clinical quality measures) were used to define three diseases: acute myocardial infarction, asthma, and diabetic nephropathy). RESULTS: Across all health systems, increasing the minimum required number of visits to be included in the study sample monotonically increased crude period prevalence estimates. The rate at which prevalence estimates increased with number of visits varied across sites and across diseases. CONCLUSIONS: In addition to providing thorough descriptions of case definitions, when using EHR data authors must carefully describe how a study sample is identified and report data for a range of sample definitions, including minimum number of visits, so that others can assess the sensitivity of reported results to sample definition in EHR data.

A predictive tool for identification of SARS-CoV-2 PCR-negative emergency department patients using routine test results.

BACKGROUND: Testing for COVID-19 remains limited in the United States and across the world. Poor allocation of limited testing resources leads to misutilization of health system resources, which complementary rapid testing tools could ameliorate. OBJECTIVE: To predict SARS-CoV-2 PCR positivity based on complete blood count components and patient sex. STUDY DESIGN: A retrospective case-control design for collection of data and a logistic regression prediction model was used. Participants were emergency department patients > 18 years old who had concurrent complete blood counts and SARS-CoV-2 PCR testing. 33 confirmed SARS-CoV-2 PCR positive and 357 negative patients at Stanford Health Care were used for model training. Validation cohorts consisted of emergency department patients > 18 years old who had concurrent complete blood counts and SARS-CoV-2 PCR testing in Northern California (41 PCR positive, 495 PCR negative), Seattle, Washington (40 PCR positive, 306 PCR negative), Chicago, Illinois (245 PCR positive, 1015 PCR negative), and South Korea (9 PCR positive, 236 PCR negative). RESULTS: A decision support tool that utilizes components of complete blood count and patient sex for prediction of SARS-CoV-2 PCR positivity demonstrated a C-statistic of 78 %, an optimized sensitivity of 93 %, and generalizability to other emergency department populations. By restricting PCR testing to predicted positive patients in a hypothetical scenario of 1000 patients requiring testing but testing resources limited to 60 % of patients, this tool would allow a 33 % increase in properly allocated resources. CONCLUSIONS: A prediction tool based on complete blood count results can better allocate SARS-CoV-2 testing and other health care resources such as personal protective equipment during a pandemic surge.

Prediction of seebeck coefficient for compounds without restriction to fixed stoichiometry: A machine learning approach.

The regression model-based tool is developed for predicting the Seebeck coefficient of crystalline materials in the temperature range from 300 K to 1000 K. The tool accounts for the single crystal versus polycrystalline nature of the compound, the production method, and properties of the constituent elements in the chemical formula. We introduce new descriptive features of crystalline materials relevant for the prediction the Seebeck coefficient. To address off-stoichiometry in materials, the predictive tool is trained on a mix of stoichiometric and nonstoichiometric materials. The tool is implemented into a web application (http://info.eecs.northwestern.edu/SeebeckCoefficientPredictor) to assist field scientists in the discovery of novel thermoelectric materials. © 2017 Wiley Periodicals, Inc.

Molecularly Tunable Fluorescent Quantum Defects.

We describe the chemical creation of molecularly tunable fluorescent quantum defects in semiconducting carbon nanotubes through covalently bonded surface functional groups that are themselves nonemitting. By variation of the surface functional groups, the same carbon nanotube crystal is chemically converted to create more than 30 distinct fluorescent nanostructures with unique near-infrared photoluminescence that is molecularly specific, systematically tunable, and significantly brighter than that of the parent semiconductor. This novel exciton-tailoring chemistry readily occurs in aqueous solution and creates functional defects on the sp(2) carbon lattice with highly predictable C-C bonding from virtually any iodine-containing hydrocarbon precursor. Our new ability to control nanostructure excitons through a single surface functional group opens up exciting possibilities for postsynthesis chemical engineering of carbon nanomaterials and suggests that the rational design and creation of a large variety of molecularly tunable quantum emitters-for applications ranging from in vivo bioimaging and chemical sensing to room-temperature single-photon sources-can now be anticipated.

Molecular-Level Engineering of Adhesion in Carbon Nanomaterial Interfaces.

Weak interfilament van der Waals interactions are potentially a significant roadblock in the development of carbon nanotube- (CNT-) and graphene-based nanocomposites. Chemical functionalization is envisioned as a means of introducing stronger intermolecular interactions at nanoscale interfaces, which in turn could enhance composite strength. This paper reports measurements of the adhesive energy of CNT-graphite interfaces functionalized with various coverages of arylpropionic acid. Peeling experiments conducted in situ in a scanning electron microscope show significantly larger adhesive energies compared to previously obtained measurements for unfunctionalized surfaces (Roenbeck et al. ACS Nano 2014, 8 (1), 124-138). Surprisingly, however, the adhesive energies are significantly higher when both surfaces have intermediate coverages than when one surface is densely functionalized. Atomistic simulations reveal a novel functional group interdiffusion mechanism, which arises for intermediate coverages in the presence of water. This interdiffusion is not observed when one surface is densely functionalized, resulting in energy trends that correlate with those observed in experiments. This unique intermolecular interaction mechanism, revealed through the integrated experimental-computational approach presented here, provides significant insights for use in the development of next-generation nanocomposites.

Shear and friction between carbon nanotubes in bundles and yarns.

We perform a detailed density functional theory assessment of the factors that determine shear interactions between carbon nanotubes (CNTs) within bundles and in related CNT and graphene structures including yarns, providing an explanation for the shear force measured in recent experiments (Filleter, T. etal. Nano Lett. 2012, 12, 73). The potential energy barriers separating AB stacked structures are found to be irrelevant to the shear analysis for bundles and yarns due to turbostratic stacking, and as a result, the tube-tube shear strength for pristine CNTs is estimated to be <0.24 MPa, that is, extremely small. Instead, it is pinning due to the presence of defects and functional groups at the tube ends that primarily cause resistance to shear when bundles are fractured in weak vacuum (∼10(-5) Torr). Such defects and groups are estimated to involve 0.55 eV interaction energies on average, which is much larger than single-atom vacancy defects (approximately 0.039 eV). Furthermore, because graphitic materials are stiff they have large coherence lengths, and this means that push-pull effects result in force cancellation for vacancy and other defects that are internal to the CNTs. Another important factor is the softness of cantilever structures relative to the stiff CNTs in the experiments, as this contributes to elastic instability transitions that account for significant dissipation during shear that has been observed. The application of these results to the mechanical behavior of yarns is discussed, providing general guidelines for the manufacture of strong yarns composed of CNTs.

Electron induced single strand break and cyclization: a DFT study on the radiosensitization mechanism of the nucleotide of 8-bromoguanine.

Cleavage of the O-P bond in 8-bromo-2'-deoxyguanosine-3',5'-diphosphate (BrdGDP), considered as a model of single strand break (SSB) in labelled double-stranded DNA (ds DNA), is investigated at the B3LYP/6-31++G(d,p) level. The thermodynamic and kinetic characteristics of the formation of SSB are compared to those related to the 5',8-cycloguanosine lesion. The first reaction step, common to both damage types, which is the formation of the reactive guanyl radical, proceeds with a barrier-free or low-barrier release of the bromide anion. The guanyl radical is then stabilized by hydrogen atom transfer from the C3' or C5' sites of the 2'-deoxyribose moiety to its C8 center. The C3' path, via the O-P bond cleavage, leads to a ketone derivative (the SSB model), while the C5' path is more likely to yield 5',8-cycloguanosine.

In situ scanning electron microscope peeling to quantify surface energy between multiwalled carbon nanotubes and graphene.

Understanding atomic interactions between constituents is critical to the design of high-performance nanocomposites. Here, we report an experimental-computational approach to investigate the adhesion energy between as-produced arc discharge multiwalled carbon nanotubes (MWCNTs) and graphene. An in situ scanning electron microscope (SEM) experiment is used to peel MWCNTs from graphene grown on copper foils. The force during peeling is obtained by monitoring the deflection of a cantilever. Finite element and molecular mechanics simulations are performed to assist the data analysis and interpretation of the results. A finite element analysis of the experimental configuration is employed to confirm the applicability of Kendall's peeling model to obtain the adhesion energy. Molecular mechanics simulations are used to estimate the effective contact width at the MWCNT-graphene interface. The measured surface energy is γ = 0.20 ± 0.09 J·m(-2) or γ = 0.36 ± 0.16 J·m(-2), depending on the assumed conformation of the tube cross section during peeling. The scatter in the data is believed to result from an amorphous carbon coating on the MWCNTs, observed using transmission electron microscopy (TEM), and the surface roughness of graphene as characterized by atomic force microscopy (AFM).

Novel view on the mechanism of water-assisted proton transfer in the DNA bases: bulk water hydration.

In the present work, the conventional static ab initio picture of a water-assisted mechanism of the tautomerization of Nucleic Acid Bases (NABs) in an aqueous environment is enhanced by the classical and Car-Parrinello molecular dynamics simulations. The inclusion of the dynamical contribution is vital because the formation and longevity of the NAB-water bridge complexes represent decisive factors for further tautomerization. The results of both molecular dynamic techniques indicate that the longest time when such complexes exist is significantly shorter than the time required for proton transfer suggested by the static ab initio level of theory. New rate constants of tautomerization corrected for the dynamic effect of environment are proposed based on the first principles molecular dynamics data. Those values are used for the evaluation of a water-assisted mechanism that is feasible in such biological systems as E. coli cell.

New insight on structural properties of hydrated nucleic acid bases from ab initio molecular dynamics.

The correlation between hydration of Nucleic Acid Bases (NABs) and their conformational flexibility was analyzed based on the results of Car-Parrinello Molecular Dynamics (CPMD) simulations of NABs in bulk water environment. Correlations with quantum chemical results were drawn whenever it was possible. Statistical analysis confirmed that hydration causes bond length alteration in NABs and formation of zwitter-ionic resonance structures. In contrast to the gas phase, bulk hydration results in restricted mobility of amino group and increase in population of its planar-like conformations. At the same time, rings of all NABs become almost equally flexible in the dynamic aqueous environment. Therefore, each NAB possesses a non-planar effective conformation of pyrimidine ring despite the fact that planar geometry corresponds to minimum on the potential energy surface.

Hydration of nucleic acid bases: a Car-Parrinello molecular dynamics approach.

Comprehensive study on interactions between nucleic acid bases (NABs) and bulk water environment has been performed with use of Car-Parrinello molecular dynamics. Detailed analysis of average number, lifetimes and mobility of water molecules, orientation and 3D organization of hydrogen bond network in the first hydration shell of adenine, guanine, cytosine and thymine has been carried out. Effect of hydration by bulk water environment has been compared with the data from polyhydrated complexes of NABs. During bulk water hydration the presence of mixed Hw...N/Hw...pi type of bonding is detected for imino nitrogen atoms. The formation of three hydrogen bonds to carbonyl groups reflects the significance of polarizing effects of aqueous environments. Hydration of hydrophobic sites revealed the presence of extremely weak bonding. Hydration of C6-H6 site of thymine is standing significantly apart from the hydration of other hydrophobic sites. An average coordination numbers of adenine, guanine, cytosine and thymine in bulk water environment are 6.87, 8.52, 6.12 and 6.42 water molecules, correspondingly. The lifetime of water molecules in the first hydration shell varies from 1 to 3 ps. Some differences in hydration studied by CPMD (bulk water) and quantum chemical (less than 20 water molecules) methods indicate a significant effect of the second hydration shell on structure and properties of the first hydration shell for the considered compounds.