Diagnostic Yield of Computed Tomography for Cancer Detection in a Tertiary Referral Population of Idiopathic Inflammatory Myositis Patients.

OBJECTIVE: To inform guidance for cancer detection in patients with idiopathic inflammatory myopathy (IIM), we evaluated the diagnostic yield of computed tomography (CT) imaging for cancer screening/surveillance within distinct IIM subtypes and myositis-specific autoantibody strata. METHODS: We conducted a single-center, retrospective cohort study in IIM patients. Overall diagnostic yield (number of cancers diagnosed/number of tests performed), percentage of false positives (number of biopsies performed not leading to cancer diagnosis/number of tests performed), and test characteristics were determined on CT of the chest and abdomen/pelvis. RESULTS: Within the first 3 years since IIM symptom onset, a total of 9 of 1,011 (0.9%) chest CT scans and 12 of 657 (1.8%) abdomen/pelvis CT scans detected cancer. Diagnostic yields for both CT of the chest and CT of the abdomen/pelvis were highest in dermatomyositis, specifically anti-transcription intermediary factor 1γ (2.9% and 2.4% for CT of the chest and abdomen/pelvis, respectively). The highest percentage of false positives was in patients with antisynthetase syndrome (ASyS) (4.4%) and immune-mediated necrotizing myopathy (4.4%) on CT of the chest, and ASyS (3.8%) on CT of the abdomen/pelvis. Patients ages <40 years old at IIM onset had both low diagnostic yields (0% and 0.5%) and high false-positive rates (1.9% and 4.4%) for CT of the chest and abdomen/pelvis, respectively. CONCLUSION: In a tertiary referral cohort of IIM patients, CT imaging has a wide range of diagnostic yield and frequency of false positives for contemporaneous cancer. These findings suggest that cancer detection strategies targeted according to IIM subtype, autoantibody positivity, and age may maximize cancer detection while minimizing the harms and costs of over-screening.

Evaluating Language Characteristics and Related Gender Bias in Letters of Recommendation for Hand Surgery Fellowship.

PURPOSE: Letters of recommendation (LORs) function as an indicator of competence and future potential for a trainee. Our purpose was to evaluate gender bias in hand surgery fellowship applicant LORs. METHODS: This was a retrospective study of all LORs submitted to a hand surgery fellowship program between 2015 and 2020. Demographic data about applicants and letter writers were collected. Linguistic analysis was performed using a text analysis software, and results were evaluated with nonparametric tests, multiple linear regression models, and a mixed effects regression model. RESULTS: Letters of recommendation were analyzed; 720 letters for 188 (23.4%) female applicants and 2,337 letters for 616 (76.6%) male applicants. Compared with LORs written for men, those written for women had more references to categories of anxiety (eg, worried and fearful) and affiliation (eg, ally and friend). Letters for women had more "clout." In subgroup analysis, letters for female plastic surgery applicants had more words signaling power, whereas recommendations for female applicants from orthopedic residencies had more words related to anxiety, achievement, work, and leisure. CONCLUSIONS: Letters of recommendation written for female residents applying to hand fellowship had more references to anxiety but were written with higher clout and more words of affiliation. Subgroup analysis looking at orthopedic and plastic surgery applicants separately found a mixed picture. Overall, these LORs written for applicants to hand fellowship had no notable specific patterns of gender bias in our analyses. CLINICAL RELEVANCE: Because programs look to train the next generation of hand surgeons, alerting letter readers to trends in implicit bias may help in the selection of qualified applicants. Bringing topics of implicit bias forward may help writers think more critically about word choice and topics.

The META tool optimizes metagenomic analyses across sequencing platforms and classifiers.

A major challenge in the field of metagenomics is the selection of the correct combination of sequencing platform and downstream metagenomic analysis algorithm, or "classifier". Here, we present the Metagenomic Evaluation Tool Analyzer (META), which produces simulated data and facilitates platform and algorithm selection for any given metagenomic use case. META-generated in silico read data are modular, scalable, and reflect user-defined community profiles, while the downstream analysis is done using a variety of metagenomic classifiers. Reported results include information on resource utilization, time-to-answer, and performance. Real-world data can also be analyzed using selected classifiers and results benchmarked against simulations. To test the utility of the META software, simulated data was compared to real-world viral and bacterial metagenomic samples run on four different sequencers and analyzed using 12 metagenomic classifiers. Lastly, we introduce "META Score": a unified, quantitative value which rates an analytic classifier's ability to both identify and count taxa in a representative sample.

Physician Use of Stigmatizing Language in Patient Medical Records.

Importance: Negative attitudes toward patients can adversely impact health care quality and contribute to health disparities. Stigmatizing language written in a patient's medical record can perpetuate negative attitudes and influence decision-making of clinicians subsequently caring for that patient. Objective: To identify and describe physician language in patient health records that may reflect, or engender in others, negative and positive attitudes toward the patient. Design, Setting, and Participants: This qualitative study analyzed randomly selected encounter notes from electronic medical records in the ambulatory internal medicine setting at an urban academic medical center. The 600 encounter notes were written by 138 physicians in 2017. Data were analyzed in 2019. Main Outcomes and Measures: Common linguistic characteristics reflecting an overall positive or negative attitude toward the patient. Results: A total of 138 clinicians wrote encounter notes about 507 patients. Of these patients, 350 (69%) were identified as female, 406 (80%) were identified as Black/African American, and 76 (15%) were identified as White. Of 600 encounter notes included in this study, there were 5 major themes representing negative language and 6 themes representing positive language. The majority of negative language was not explicit and generally fell into one or more of the following categories: (1) questioning patient credibility, (2) expressing disapproval of patient reasoning or self-care, (3) stereotyping by race or social class, (4) portraying the patient as difficult, and (5) emphasizing physician authority over the patient. Positive language was more often more explicit and included (1) direct compliments, (2) expressions of approval, (3) self-disclosure of the physician's own positive feelings toward the patient, (4) minimization of blame, (5) personalization, and (6) highlighting patient authority for their own decisions. Conclusions and Relevance: This qualitative study found that physicians express negative and positive attitudes toward patients when documenting in the medical record. Although often not explicit, this language could potentially transmit bias and affect the quality of care that patients subsequently receive. These findings suggest that increased physician awareness when writing and reading medical records is needed to prevent the perpetuation of negative bias in medical care.

Testimonial Injustice: Linguistic Bias in the Medical Records of Black Patients and Women.

BACKGROUND: Black Americans and women report feeling doubted or dismissed by health professionals. OBJECTIVE: To identify linguistic mechanisms by which physicians communicate disbelief of patients in medical records and then to explore racial and gender differences in the use of such language. DESIGN: Cross-sectional. SETTING/PARTICIPANTS: All notes for patients seen in an academic ambulatory internal medicine practice in 2017. MAIN MEASURES: A content analysis of 600 clinic notes revealed three linguistic features suggesting disbelief: (1) quotes (e.g., had a "reaction" to the medication); (2) specific "judgment words" that suggest doubt (e.g., "claims" or "insists"); and (3) evidentials, a sentence construction in which patients' symptoms or experience is reported as hearsay. We used natural language processing to evaluate the prevalence of these features in the remaining notes and tested differences by race and gender, using mixed-effects regression to account for clustering of notes within patients and providers. KEY RESULTS: Our sample included 9251 notes written by 165 physicians about 3374 unique patients. Most patients were identified as Black (74%) and female (58%). Notes written about Black patients had higher odds of containing at least one quote (OR 1.48, 95% CI 1.20-1.83) and at least one judgment word (OR 1.25, 95% CI 1.02-1.53), and used more evidentials (ß 0.32, 95% CI 0.17-0.47), compared to notes of White patients. Notes about female vs. male patients did not differ in terms of judgment words or evidentials but had a higher odds of containing at least one quote (OR 1.22, 95% CI 1.05-1.44). CONCLUSIONS: Black patients may be subject to systematic bias in physicians' perceptions of their credibility, a form of testimonial injustice. This is another potential mechanism for racial disparities in healthcare quality that should be further investigated and addressed.

Are There Gender-based Differences in Language in Letters of Recommendation to an Orthopaedic Surgery Residency Program?

BACKGROUND: Letters of recommendation are considered one of the most important factors for whether an applicant is selected for an interview for orthopaedic surgery residency programs. Language differences in letters describing men versus women candidates may create differential perceptions by gender. Given the gender imbalance in orthopaedic surgery, we sought to determine whether there are differences in the language of letters of recommendation by applicant gender. QUESTIONS/PURPOSES: (1) Are there differences in word count and word categories in letters of recommendation describing women and men applicants, regardless of author gender? (2) Is author gender associated with word category differences in letters of recommendation? (3) Do authors of different academic rank use different words to describe women versus men applicants? METHODS: Using a linguistic analysis in a retrospective study, we analyzed all letters of recommendation (2834 letters) written for all 738 applicants with completed Electronic Residency Application Service applications submitted to the Johns Hopkins Orthopaedic Surgery Residency program during the 2018 to 2019 cycle to determine differences in word category use among applicants by gender, authors by gender, and authors by academic rank. Thirty nine validated word categories from the Linguistic Inquiry and Word Count dictionary along with seven additional word categories from previous publications were used in this analysis. The occurrence of words in each word category was divided by the number of words in the letter to obtain a word frequency for each letter. We calculated the mean word category frequency across all letters and analyzed means using non-parametric tests. For comparison of two groups, a p value threshold of 0.05 was used. For comparison of multiple groups, the Bonferroni correction was used to calculate an adjusted p value (p = 0.00058). RESULTS: Letters of recommendation for women applicants were slightly longer compared with those for men applicants (366 ± 188 versus 339 ± 199 words; p = 0.003). When comparing word category differences by applicant gender, letters for women applicants had slightly more "achieve" words (0.036 ± 0.015 versus 0.035 ± 0.018; p < 0.0001). Letters for men had more use of their first name (0.016 ± 0.013 versus 0.014 ± 0.009; p < 0.0001), and more "young" words (0.001 ± 0.003 versus 0.000 ± 0.001; p < 0.0001) than letters for women applicants. These differences were very small as each 0.001 difference in mean word frequency was equivalent to one more additional word from the word category appearing when comparing three letters for women to three letters for men. For differences in letters by author gender, there were no word category differences between men and women authors. Finally, when looking at author academic rank, letters for men applicants written by professors had slightly more "research" terms (0.011 ± 0.010) than letters written by associate professors (0.010 ± 0.010) or faculty of other rank (0.009 ± 0.011; p < 0.0001), a finding not observed in letters written for women. CONCLUSIONS: Although there were some minor differences favoring women, language in letters of recommendation to an academic orthopaedic surgery residency program were overall similar between men and women applicants. CLINICAL RELEVANCE: Given the similarity in language between men and women applicants, increasing women applicants may be a more important factor in addressing the gender gap in orthopaedics.

Predicting adverse drug events from personal health messages.

Adverse drug events (ADEs) remain a large problem in the United States, being the fourth leading cause of death, despite post market drug surveillance. Much post consumer drug surveillance relies on self-reported "spontaneous" patient data. Previous work has performed datamining over the FDA's Adverse Event Reporting System (AERS) and other spontaneous reporting systems to identify drug interactions and drugs correlated with high rates of serious adverse events. However, safety problems have resulted from the lack of post marketing surveillance information about drugs, with underreporting rates of up to 98% within such systems. We explore the use of online health forums as a source of data to identify drugs for further FDA scrutiny. In this work we aggregate individuals' opinions and review of drugs similar to crowd intelligence3. We use natural language processing to group drugs discussed in similar ways and are able to successfully identify drugs withdrawn from the market based on messages discussing them before their removal.

BeeSpace Navigator: exploratory analysis of gene function using semantic indexing of biological literature.

With the rapid decrease in cost of genome sequencing, the classification of gene function is becoming a primary problem. Such classification has been performed by human curators who read biological literature to extract evidence. BeeSpace Navigator is a prototype software for exploratory analysis of gene function using biological literature. The software supports an automatic analogue of the curator process to extract functions, with a simple interface intended for all biologists. Since extraction is done on selected collections that are semantically indexed into conceptual spaces, the curation can be task specific. Biological literature containing references to gene lists from expression experiments can be analyzed to extract concepts that are computational equivalents of a classification such as Gene Ontology, yielding discriminating concepts that differentiate gene mentions from other mentions. The functions of individual genes can be summarized from sentences in biological literature, to produce results resembling a model organism database entry that is automatically computed. Statistical frequency analysis based on literature phrase extraction generates offline semantic indexes to support these gene function services. The website with BeeSpace Navigator is free and open to all; there is no login requirement at www.beespace.illinois.edu for version 4. Materials from the 2010 BeeSpace Software Training Workshop are available at www.beespace.illinois.edu/bstwmaterials.php.

Identifying overrepresented concepts in gene lists from literature: a statistical approach based on Poisson mixture model.

BACKGROUND: Large-scale genomic studies often identify large gene lists, for example, the genes sharing the same expression patterns. The interpretation of these gene lists is generally achieved by extracting concepts overrepresented in the gene lists. This analysis often depends on manual annotation of genes based on controlled vocabularies, in particular, Gene Ontology (GO). However, the annotation of genes is a labor-intensive process; and the vocabularies are generally incomplete, leaving some important biological domains inadequately covered. RESULTS: We propose a statistical method that uses the primary literature, i.e. free-text, as the source to perform overrepresentation analysis. The method is based on a statistical framework of mixture model and addresses the methodological flaws in several existing programs. We implemented this method within a literature mining system, BeeSpace, taking advantage of its analysis environment and added features that facilitate the interactive analysis of gene sets. Through experimentation with several datasets, we showed that our program can effectively summarize the important conceptual themes of large gene sets, even when traditional GO-based analysis does not yield informative results. CONCLUSIONS: We conclude that the current work will provide biologists with a tool that effectively complements the existing ones for overrepresentation analysis from genomic experiments. Our program, Genelist Analyzer, is freely available at: http://workerbee.igb.uiuc.edu:8080/BeeSpace/Search.jsp.

Measuring population health using personal health messages.

Personal health messages - inter patient communications within online communities; represent a new path towards providing continuous information about patient derived health status. We apply natural language processing techniques to personal health messages from online message boards to demonstrate the ability to track trends in people's positive or negative opinion (sentiment) regarding particular drugs over time. The significant changes in sentiment correspond to FDA announcements and other publicity. We envision such analysis as a scalable tool for pharmacovigilance hypothesis generation for possible adverse drug reactions.