Automated stratification of trauma injury severity across multiple body regions using multi-modal, multi-class machine learning models

OBJECTIVE

The timely stratification of trauma injury severity can enhance the quality of trauma care but it requires intense manual annotation from certified trauma coders. The objective of this study is to develop machine learning models for the stratification of trauma injury severity across various body regions using clinical text and structured electronic health records (EHRs) data.

MATERIALS AND METHODS

Our study utilized clinical documents and structured EHR variables linked with the trauma registry data to create 2 machine learning models with different approaches to representing text. The first one fuses concept unique identifiers (CUIs) extracted from free text with structured EHR variables, while the second one integrates free text with structured EHR variables. Temporal validation was undertaken to ensure the models' temporal generalizability. Additionally, analyses to assess the variable importance were conducted.

RESULTS

Both models demonstrated impressive performance in categorizing leg injuries, achieving high accuracy with macro-F1 scores of over 0.8. Additionally, they showed considerable accuracy, with macro-F1 scores exceeding or near 0.7, in assessing injuries in the areas of the chest and head. We showed in our variable importance analysis that the most important features in the model have strong face validity in determining clinically relevant trauma injuries.

DISCUSSION

The CUI-based model achieves comparable performance, if not higher, compared to the free-text-based model, with reduced complexity. Furthermore, integrating structured EHR data improves performance, particularly when the text modalities are insufficiently indicative.

CONCLUSIONS

Our multi-modal, multiclass models can provide accurate stratification of trauma injury severity and clinically relevant interpretations.

Development and external validation of multimodal postoperative acute kidney injury risk machine learning models

Objectives

To develop and externally validate machine learning models using structured and unstructured electronic health record data to predict postoperative acute kidney injury (AKI) across inpatient settings.

Materials and Methods

Data for adult postoperative admissions to the Loyola University Medical Center (2009-2017) were used for model development and admissions to the University of Wisconsin-Madison (2009-2020) were used for validation. Structured features included demographics, vital signs, laboratory results, and nurse-documented scores. Unstructured text from clinical notes were converted into concept unique identifiers (CUIs) using the clinical Text Analysis and Knowledge Extraction System. The primary outcome was the development of Kidney Disease Improvement Global Outcomes stage 2 AKI within 7 days after leaving the operating room. We derived unimodal extreme gradient boosting machines (XGBoost) and elastic net logistic regression (GLMNET) models using structured-only data and multimodal models combining structured data with CUI features. Model comparison was performed using the receiver operating characteristic curve (AUROC), with Delong's test for statistical differences.

Results

The study cohort included 138 389 adult patient admissions (mean [SD] age 58 [16] years; 11 506 [8%] African-American; and 70 826 [51%] female) across the 2 sites. Of those, 2959 (2.1%) developed stage 2 AKI or higher. Across all data types, XGBoost outperformed GLMNET (mean AUROC 0.81 [95% confidence interval (CI), 0.80-0.82] vs 0.78 [95% CI, 0.77-0.79]). The multimodal XGBoost model incorporating CUIs parameterized as term frequency-inverse document frequency (TF-IDF) showed the highest discrimination performance (AUROC 0.82 [95% CI, 0.81-0.83]) over unimodal models (AUROC 0.79 [95% CI, 0.78-0.80]).

Discussion

A multimodality approach with structured data and TF-IDF weighting of CUIs increased model performance over structured data-only models.

Conclusion

These findings highlight the predictive power of CUIs when merged with structured data for clinical prediction models, which may improve the detection of postoperative AKI.

DR.BENCH: Diagnostic Reasoning Benchmark for Clinical Natural Language Processing

The meaningful use of electronic health records (EHR) continues to progress in the digital era with clinical decision support systems augmented by artificial intelligence. A priority in improving provider experience is to overcome information overload and reduce the cognitive burden so fewer medical errors and cognitive biases are introduced during patient care. One major type of medical error is diagnostic error due to systematic or predictable errors in judgement that rely on heuristics. The potential for clinical natural language processing (cNLP) to model diagnostic reasoning in humans with forward reasoning from data to diagnosis and potentially reduce cognitive burden and medical error has not been investigated. Existing tasks to advance the science in cNLP have largely focused on information extraction and named entity recognition through classification tasks. We introduce a novel suite of tasks coined as Diagnostic Reasoning Benchmarks, Dr.Bench, as a new benchmark for developing and evaluating cNLP models with clinical diagnostic reasoning ability. The suite includes six tasks from ten publicly available datasets addressing clinical text understanding, medical knowledge reasoning, and diagnosis generation. DR.BENCH is the first clinical suite of tasks designed to be a natural language generation framework to evaluate pre-trained language models for diagnostic reasoning. The goal of DR. BENCH is to advance the science in cNLP to support downstream applications in computerized diagnostic decision support and improve the efficiency and accuracy of healthcare providers during patient care. We fine-tune and evaluate the state-of-the-art generative models on DR.BENCH. Experiments show that with domain adaptation pre-training on medical knowledge, the model demonstrated opportunities for improvement when evaluated in DR. BENCH. We share DR. BENCH as a publicly available GitLab repository with a systematic approach to load and evaluate models for the cNLP community. We also discuss the carbon footprint produced during the experiments and encourage future work on DR.BENCH to report the carbon footprint.

A Natural Language Processing Pipeline to Identify COVID-19 Outbreaks from Contact Tracing Interview Forms for Public Health Departments

BACKGROUND

In Wisconsin, COVID-19 case interview forms contain free text fields that need to be mined to identify potential outbreaks for targeted policy making. We developed an automated pipeline to ingest the free text into a pre-trained neural language model to identify businesses and facilities as outbreaks.

OBJECTIVE

We aimed to examine the precision and recall of our natural language processing pipeline against existing outbreaks and potentially new clusters.

METHODS

Data on cases of COVID-19 were extracted from the Wisconsin Electronic Disease Surveillance System (WEDSS) for Dane County between July 1, 2020, and June 30, 2021. Features from the case interview forms were fed into a Bidirectional Encoder Representations from Transformers (BERT) model that was fine-tuned for named entity recognition (NER). We also developed a novel location mapping tool to provide addresses for relevant NERs. Precision and recall were measured against manually verified outbreaks and valid addresses in WEDSS.

RESULTS

There were 46,798 cases of COVID-19 with 4,183,273 total BERT tokens and 15,051 unique tokens. The recall and precision of the NER tool were 0.67 (95 % CI 0.66-0.68) and 0.55 (95 % CI: 0.54-0.57), respectively. For the location mapping tool, the recall and precision were 0.93 (95% CI: 0.92-0.95) and 0.93 (95% CI: 0.92-0.95), respectively. Across monthly intervals, the NER tool identified more potential clusters than were verified in the WEDSS system.

CONCLUSIONS

We developed a novel pipeline of tools that identified existing outbreaks and novel clusters with associated addresses. Our pipeline ingests data from a statewide database and may be deployed to assist local health departments for targeted interventions.

Intramuscular vaccination of guinea pigs with the live-attenuated human herpes simplex vaccine VC2 stimulates a transcriptional profile of vaginal Th17 and regulatory Tr1 responses

Herpes simplex virus is a common causative agent of oral and genital diseases. Novel vaccines and therapeutics are needed to combat herpes infections especially after the failure of subunit vaccines in human clinical trials. We have shown that the live-attenuated HSV-1 VC2 vaccine strain is unable to establish latency in vaccinated animals and produces a robust immune response capable of completely protecting mice against lethal vaginal HSV-1 or HSV-2 infections. The guinea pig represents the best small animal model of genital HSV-2 disease. Reported here, twenty-one female Hartley guinea pigs received intramuscular injection with either the VC2 vaccine, or equal volume of conditioned tissue culture media. Animals received 2 booster vaccinations at 21 day intervals following the initial vaccination. After vaccination, animals were challenged with the highly virulent HSV-2 (G) strain. Histologically, VC2 vaccinated animals had little to no apparent inflammation/disease following challenge. Unvaccinated animals developed moderate to severe erosive and ulcerative vaginitis. Quantitative reverse-transcriptase PCR analysis in VC2 vaccinated and challenged animals identified transcriptional signatures of Th17 and regulatory Tr1 cells associated with the inflammatory response primed by VC2 vaccination. Treatment of cultured human vaginal epithelial cells (VK2 cells) with a combination of IL-17A and IL-22 resulted in the significant induction of beta-defensin 3 expression. Further, treatment of VK2 cells with IL-17A, IL-22, IL-36 or beta-defensin 3 resulted in diminished HSV-2 replication. Overall, these results suggest that intramuscular vaccination with the live-attenuated vaccine VC2 primes a mucosal immune response predisposing the adaptive expression of transcripts associated with a Th17 response to challenge and these responses contribute to antiviral immunity.

Sequencing-relative to hybridization-based transcriptomics approaches better define Mycobacterium tuberculosis stress-response regulons

Mycobacterium tuberculosis (Mtb) infections cause tuberculosis (TB), an infectious disease which causes ∼1.5 million deaths annually. The ability of this pathogen to evade, escape and encounter immune surveillance is fueled by its adaptability. Thus, Mtb induces a transition in its transcriptome in response to environmental changes. Global transcriptome profiling has been key to our understanding of how Mtb responds to the different stress conditions it faces during its life cycle. While this was initially achieved using microarray technology, RNAseq is now widely employed. It is important to understand the correlation between the large amount of microarray based transcriptome data, which continues to shape our understanding of Mtb stress networks, and newer data being generated using RNAseq. We assessed how well the two platforms correlate using three well-defined stress conditions: diamide, hypoxia, and re-aeration. The data used here was generated by different individuals over time using distinct samples, providing a stringent test of platform correlation. While correlation between microarrays and sequencing was high upon diamide treatment, which causes a rapid reprogramming of the transcriptome, RNAseq allowed a better definition of the hypoxic response, characterized by subtle changes in the magnitude of gene-expression. RNAseq also allows for the best cross-platform reproducibility.

Cigarette Smoke Extract-Exposed Methicillin-Resistant Staphylococcus aureus Regulates Leukocyte Function for Pulmonary Persistence

Cigarette smoke (CS) predisposes exposed individuals to respiratory infections not only by suppressing immune response but also by enhancing the virulence of pathogenic bacteria. As per our observations, in methicillin-resistant Staphylococcus aureus strain USA300, CS extract (CSE) potentiates biofilm formation via the down-regulation of quorum-sensing regulon accessory gene regulator. Because accessory gene regulator is a global regulator of the staphylococcal virulome, in the present study we sought to identify the effects of CS exposure on staphylococcal gene expression using RNAseq. Comparative analysis of RNAseq profiles revealed the up-regulation of important virulence genes encoding surface adhesins (fibronectin- and fibrinogen-binding proteins A and B and clumping factor B) and proteins involved in immune evasion, such as staphylocoagulase, staphylococcal protein A, and nuclease. In concurrence with the RNAseq data, we observed: (1) significant up-regulation of the ability of CSE-exposed USA300 to evade phagocytosis by macrophages and neutrophils, a known function of staphylococcal protein A; and (2) twofold higher (P < 0.001) number of CSE-exposed USA300 escaping neutrophil extracellular trap-mediated killing by neutrophils as a result of CS-mediated induction of nuclease. Importantly, in three different mouse strains, C57BL6/J, Balb/C, and A/J, we observed significantly higher pulmonary bacterial burden in animals infected with CSE-exposed USA300 as compared with medium-exposed control USA300. Taken together, these observations indicate that bioactive chemicals in CS induce hypervirulence by augmenting the ability of USA300 to evade bactericidal functions of leukocytes, such as phagocytosis and neutrophil extracellular trap-mediated killing.

Spontaneous Rupture of the Iliac Vein

The fourth known case of spontaneous iliac vein rupture is reported. Prodromal symptoms, etiology, and treatment are dis cussed. The three previously reported cases are reviewed.

Adolescents anticipated experience of screening for genital herpes

Screening asymptomatic adolescents for genital herpes will require sensitivity to adolescents' developmental needs. Twenty-four adolescents (age range 1619 years) were interviewed to explore their perceptions of screening. In general, adolescents thought screening for genital herpes would be viewed as taking care of themselves, although there were concerns that their peers might view getting screened as implying that the adolescent was infected or sexually promiscuous. Most adolescents expected their parents to respond positively, but typically thought that younger adolescents should obtain parental consent for screening, and that adolescents should inform their parents of positive test results. Adolescents wanted to be screened in settings that provided confidentiality and by non-judgemental care providers, but they differed on the setting which they believed would accomplish this. Some recommended mass screening (for example, screening all of those in a certain year at school), presumably as a way to reduce embarrassment and/or stigma. Screening programmes that are adolescent-friendly and accessible, and address adolescents' specific concerns regarding managing the information, can be created.

A phase I study of daily carboplatin and simultaneous accelerated, hyperfractionated chest irradiation in patients with regionally inoperable non-small cell lung cancer

PURPOSE

This Phase I study was designed to determine the maximally tolerated dose (MTD) of daily low dose carboplatin with concurrent accelerated hyperfractionated radiotherapy (AHFX) in patients with locally advanced non-small-cell lung cancer. Patients also received consolidation chemotherapy with carboplatin. Secondary objectives were to determine the response rate, response duration, sites of first relapse, and survival.

METHODS AND MATERIALS

Thirty patients received daily carboplatin at doses of 25 or 30 mg/m2. Concurrent radiotherapy was given in 1.5 Gy fractions twice daily for a total dose of 60 Gy. Following chemoradiotherapy, patients received four cycles of carboplatin at 350 mg/m2.

RESULTS

Grade 4 esophagitis developed in 2 of 6 (33%) patients receiving 30 mg/m2 of daily carboplatin and was dose limiting. The remaining 24 patients received carboplatin at 25 mg/m2, with 3 patients developing Grade 4 esophagitis (13%). One of 22 patients who received consolidation carboplatin developed Grade 4 thrombocytopenia. An objective response was observed in 70% of patients (2 complete and 17 partial). Sites of failure were local (7 patients), distant (7 patients), and both (3 patients). The median time to progression was 8.3 months, with a median survival time of 18.3 months. The 1- and 2-year survival rates were 63 and 49%, respectively.

CONCLUSIONS

Esophagitis was dose limiting when 30 mg/m2 of daily carboplatin was administered with AHFX. At the MTD of 25 mg/m2 of daily carboplatin plus AHFX followed by four cycles of carboplatin, the regimen was shown to be safe and as active or more active than other regimens. Thus, further studies with this regimen are warranted.