Voice outcomes in high-grade Reinke's edema: Comparing microflap excision and microdebrider surgery.

Objectives: Patients presenting with hoarseness and diagnosed with high-grade Reinke's edema (RE) will often require surgical intervention for polypoid changes of the true vocal folds. We compared patient outcomes in patients who had microflap or microdebrider excision surgeries. Methods: Patients with the diagnosis of grade II or grade III RE based on laryngoscopy or videostroboscopy who failed conservative management underwent surgery using the standard excision practice of the primary surgeon. Voice outcomes were compared using VHI-30 (Voice Handicap Index), V-RQOL (Voice-Related Quality of Life), and MPT (maximum phonation time) preoperatively and at 1-month and 6-months postoperatively. Results: Of the 115 patients included, there were 46 RE grade II patients and 69 RE grade III patients with 52 patient undergoing microflap surgery and 63 patients undergoing microdebrider surgery. Both procedures resulted in significant improvement in VHI-30, V-RQOL, and MPT at 1-month and 6-months postoperatively. The microdebrider group had better 6-month VHI scores (40.84) than the microflap group (44.54) (CI -7.27 to -0.12). The microdebrider group also had better 6-month V-RQOL measures (62.56) than the microflap group (57.79) (CI 0.38-9.16). Conclusion: Both microflap excision and microdebrider excision for high-grade RE lesions resulted in significant improvement in VHI-30, V-RQOL, and MPT at 1-month and 6-months postoperatively with the microdebrider excision group scoring statistically significantly better at 6 months in comparison to the microflap group. Overall, the results support the use of both surgical modalities for treating high-grade RE patients. Level of evidence: 3.

Overview of the 2022 n2c2 shared task on contextualized medication event extraction in clinical notes.

BACKGROUND: An accurate medication history, foundational for providing quality medical care, requires understanding of medication change events documented in clinical notes. However, extracting medication changes without the necessary clinical context is insufficient for real-world applications. METHODS: To address this need, Track 1 of the 2022 National NLP Clinical Challenges focused on extracting the context for medication changes documented in clinical notes using the Contextualized Medication Event Dataset. Track 1 consisted of 3 subtasks: extracting medication mentions from clinical notes (NER), determining whether a medication change is being discussed (Event), and determining the action, negation, temporality, certainty, and actor for any change events (Context). Participants were allowed to participate in any one or more of the subtasks. RESULTS: A total of 32 teams with participants from 19 countries submitted a total of 211 systems across all subtasks. Most teams formulated NER as a token classification task and Event and Context as multi-class classification tasks, using transformer-based large language models. Overall, performance for NER was high across submitted systems. However, performance for Event and Context were much lower, often due to indirectly stated change events with no clear action verb, events requiring farther textual clues for understanding, and medication mentions with multiple change events. CONCLUSIONS: This shared task showed that while NLP research on medication extraction is relatively mature, understanding of contextual information surrounding medication events in clinical notes is still an open problem requiring further research to achieve the end goal of supporting real-world clinical applications.

Extracting medication changes in clinical narratives using pre-trained language models.

An accurate and detailed account of patient medications, including medication changes within the patient timeline, is essential for healthcare providers to provide appropriate patient care. Healthcare providers or the patients themselves may initiate changes to patient medication. Medication changes take many forms, including prescribed medication and associated dosage modification. These changes provide information about the overall health of the patient and the rationale that led to the current care. Future care can then build on the resulting state of the patient. This work explores the automatic extraction of medication change information from free-text clinical notes. The Contextual Medication Event Dataset (CMED) is a corpus of clinical notes with annotations that characterize medication changes through multiple change-related attributes, including the type of change (start, stop, increase, etc.), initiator of the change, temporality, change likelihood, and negation. Using CMED, we identify medication mentions in clinical text and propose three novel high-performing BERT-based systems that resolve the annotated medication change characteristics. We demonstrate that our proposed systems improve medication change classification performance over the initial work exploring CMED.

The Use of Race and Socioeconomic Status Variables in Published Otolaryngologic Research.

OBJECTIVE: To characterize the use of race and socioeconomic status (SES) variables in clinical otolarynogologic research. METHODS: Databases were queried for all articles published in 2016 issues of 5 major otolaryngologic journals. One thousand, one hundred and forty of 1593 articles abstracted met inclusion criteria for analysis. RESULTS: In total, 244 (21.4%) studies specified race as a variable. The subspecialty of Head and Neck cancer specified race at statistically higher rates compared to other subspecialties (P = .002). Two hundred nine (34.0%) domestic studies specified race compared to 35 (6.7%) international studies. Of the 244 studies that specified race, 79 (32.4%) defined race using racial and ethnic categories interchangeably. Two hundred twenty-four (91.8%) studies reported data by race, 145 (59.4%) analyzed the data, and 112 (45.9%) discussed race-based results.In total, 94 (8.2%) studies specified SES. All subspecialties specified SES at statistically similar rates. Seventy (11.4%) domestic studies specified SES compared to 24 (4.6%) international studies. Of the 94 studies that specified SES, 42 (44.7%) defined SES using insurance status, 35 (37.2%) used education, and 32 (34.0%) used income. Seventy-eight (83.0%) studies reported data by SES, 71 (75.5%) analyzed the data, and 68 (72.3%) discussed SES-based results. CONCLUSION: In clinical otolaryngologic research, the study of race and SES is limited. To improve quality of research and patient care for all patients, investigators should clearly justify their use of race and SES variables, carefully select their measures of race and SES (if the use of these variables is justified), and study race/SES-based data beyond just a superficial level.

Postoperative Complications After Total Thyroidectomy for Patients With Graves' Disease.

OBJECTIVE: To identify the rates and types of postoperative complications in patients with and without Graves' disease undergoing total thyroidectomy using the National Surgical Quality Improvement Program (NSQIP) database. STUDY DESIGN: Retrospective cohort study. SETTING: All hospitals participating in NSQIP from 2007 to 2017. METHODS: Thyroidectomy data were abstracted from the NSQIP database from 2007 to 2017 using related Current Procedural Terminology codes. Exclusion criteria included diagnosis of malignancy and partial thyroidectomy. Patients with a diagnosis of Graves' disease were compared against the control group, which consisted of other nononcologic diagnoses. Statistical analysis including matched pair analysis was performed. RESULTS: Unmatched data demonstrated that patients with Graves' disease who underwent total thyroidectomy (n = 5495) had a higher rate of readmission (odds ratio [OR], 1.41; 95% CI, 1.16-1.73) and rate of reoperation (OR, 2.29; 95% CI, 1.88-2.79) in comparison to control patients (n = 24,213). They also had a higher rate of postoperative complication (OR, 1.54; 95% CI, 1.23-1.93) especially for wound-related outcomes (OR, 1.88; 95% CI, 1.32-2.69), readmission for postoperative hypocalcemia (OR, 2.12; 95% CI, 1.54-2.92), and reoperation for hematoma or hemorrhage (OR, 1.88; 95% CI, 1.32-2.69). A matched-pair analysis of the data also demonstrated similar significant results. CONCLUSION: Patients with Graves' disease undergoing total thyroidectomy are at higher risk of complications in comparison to those who do not have Graves' disease, likely due to sequelae of the disease. However, overall rates were low, suggesting that the procedure remains relatively low risk and should continue to be offered to select patients who meet criteria for surgery.

Capturing Individual-level Social Determinants from Clinical Text.

Knowledge of social determinants of health (SDOH), which refer to nonmedical factors influencing health outcomes, can help providers improve patient care. However, SDOH are often documented in unstructured notes, making them more inaccessible. Although previous works have attempted SDOH extraction from clinical notes, most efforts defined SDOH more narrowly and focused on the note's social history (SH) section, where social factors are traditionally documented. Here, we introduce a new SDOH dataset covering a broad range of SDOH content that is annotated over entire notes. We characterize what, where, and how SDOH information is documented in clinical text, present baseline systems using a token classification and generative approach, and investigate whether training only on the SH section can effectively extract SDOH from the entire note. The final dataset, consisting of 2,007 annotations covering 7 open-ended SDOH domains over 500 notes, will be publicly released to encourage further research in this area.

Antimicrobial use in Canadian acute-care hospitals: Findings from three national point-prevalence surveys between 2002 and 2017.

OBJECTIVES: The Canadian Nosocomial Infection Surveillance Program conducted point-prevalence surveys in acute-care hospitals in 2002, 2009, and 2017 to identify trends in antimicrobial use. METHODS: Eligible inpatients were identified from a 24-hour period in February of each survey year. Patients were eligible (1) if they were admitted for ≥48 hours or (2) if they had been admitted to the hospital within a month. Chart reviews were conducted. We calculated the prevalence of antimicrobial use as follows: patients receiving ≥1 antimicrobial during survey period per number of patients surveyed × 100%. RESULTS: In each survey, 28-47 hospitals participated. In 2002, 2,460 (36.5%; 95% CI, 35.3%-37.6%) of 6,747 surveyed patients received ≥1 antimicrobial. In 2009, 3,566 (40.1%, 95% CI, 39.0%-41.1%) of 8,902 patients received ≥1 antimicrobial. In 2017, 3,936 (39.6%, 95% CI, 38.7%-40.6%) of 9,929 patients received ≥1 antimicrobial. Among patients who received ≥1 antimicrobial, penicillin use increased 36.8% between 2002 and 2017, and third-generation cephalosporin use increased from 13.9% to 18.1% (P < .0001). Between 2002 and 2017, fluoroquinolone use decreased from 25.7% to 16.3% (P < .0001) and clindamycin use decreased from 25.7% to 16.3% (P < .0001) among patients who received ≥1 antimicrobial. Aminoglycoside use decreased from 8.8% to 2.4% (P < .0001) and metronidazole use decreased from 18.1% to 9.4% (P < .0001). Carbapenem use increased from 3.9% in 2002 to 6.1% in 2009 (P < .0001) and increased by 4.8% between 2009 and 2017 (P = .60). CONCLUSIONS: The prevalence of antimicrobial use increased between 2002 and 2009 and then stabilized between 2009 and 2017. These data provide important information for antimicrobial stewardship programs.

Hierarchy-aware Adverse Reaction Embeddings for Signal Detection.

Post-market drug surveillance monitors new and evolving treatments for their effectiveness and safety in real-world conditions. A large amount of drug safety surveillance data is captured by spontaneous reporting systems such as the FAERS. Developing automated methods to identify actionable safety signals from these databases is an active area of research. In this paper, we propose two novel network representation learning methods (HARE and T-HARE) for signal detection that jointly utilize association information between drugs and medical outcomes from the FAERS and ancestral information in medical ontologies. We evaluate these methods using two publicly available reference datasets, EU-ADR and OMOP corpus. Experimental results showed that the proposed methods significantly outper-formed standard methodologies based on disproportionality metrics and the existing state-of-the-art aer2vec method with statistically significant improvements on both EU-ADR and OMOP datasets. Through quantitative and qualitative analysis, we demonstrate the potential of the proposed methods for effective signal detection.

Reducing Physicians' Cognitive Load During Chart Review: A Problem-Oriented Summary of the Patient Electronic Record.

Overabundance of information within electronic health records (EHRs) has resulted in a need for automated systems to mitigate the cognitive burden on physicians utilizing today's EHR systems. We present ProSPER, a Problem-oriented Summary of the Patient Electronic Record that displays a patient summary centered around an auto-generated problem list and disease-specific views for chronic conditions. ProSPER was developed using 1,500 longitudinal patient records from two large multi-specialty medical groups in the United States, and leverages multiple natural language processing (NLP) components targeting various fundamental (e.g. syntactic analysis), clinical (e.g. adverse drug event extraction) and summarizing (e.g. problem list generation) tasks. We report evaluation results for each component and discuss how specific components address existing physician challenges in reviewing EHR data. This work demonstrates the need to leverage holistic information in EHRs to build a comprehensive summarization application, and the potential for NLP-based applications to support physicians and improve clinical care.

Toward Understanding Clinical Context of Medication Change Events in Clinical Narratives.

Understanding medication events in clinical narratives is essential to achieving a complete picture of a patient's medication history. While prior research has explored identification of medication changes in clinical notes, due to the longitudinal and narrative nature of clinical documentation, extraction of medication change alone without the necessary clinical context is insufficient for use in real-world applications, such as medication timeline generation and medication reconciliation. Here, we present a framework to capture multi-dimensional context of medication changes documented in clinical notes. We define specific contextual aspects pertinent to medication change events (i.e. Action, Negation, Temporality, Certainty, and Actor), describe the annotation process and challenges encountered while creating the dataset, and explore models based on state-of-the-art transformers to automate the task. The resulting dataset, Contextualized Medication Event Dataset (CMED), consisting of 9,013 medications annotated over 500 clinical notes, will be released to the community as a shared task in 2021-2022.

Identification of Semantically Similar Sentences in Clinical Notes: Iterative Intermediate Training Using Multi-Task Learning.

BACKGROUND: Although electronic health records (EHRs) have been widely adopted in health care, effective use of EHR data is often limited because of redundant information in clinical notes introduced by the use of templates and copy-paste during note generation. Thus, it is imperative to develop solutions that can condense information while retaining its value. A step in this direction is measuring the semantic similarity between clinical text snippets. To address this problem, we participated in the 2019 National NLP Clinical Challenges (n2c2)/Open Health Natural Language Processing Consortium (OHNLP) clinical semantic textual similarity (ClinicalSTS) shared task. OBJECTIVE: This study aims to improve the performance and robustness of semantic textual similarity in the clinical domain by leveraging manually labeled data from related tasks and contextualized embeddings from pretrained transformer-based language models. METHODS: The ClinicalSTS data set consists of 1642 pairs of deidentified clinical text snippets annotated in a continuous scale of 0-5, indicating degrees of semantic similarity. We developed an iterative intermediate training approach using multi-task learning (IIT-MTL), a multi-task training approach that employs iterative data set selection. We applied this process to bidirectional encoder representations from transformers on clinical text mining (ClinicalBERT), a pretrained domain-specific transformer-based language model, and fine-tuned the resulting model on the target ClinicalSTS task. We incrementally ensembled the output from applying IIT-MTL on ClinicalBERT with the output of other language models (bidirectional encoder representations from transformers for biomedical text mining [BioBERT], multi-task deep neural networks [MT-DNN], and robustly optimized BERT approach [RoBERTa]) and handcrafted features using regression-based learning algorithms. On the basis of these experiments, we adopted the top-performing configurations as our official submissions. RESULTS: Our system ranked first out of 87 submitted systems in the 2019 n2c2/OHNLP ClinicalSTS challenge, achieving state-of-the-art results with a Pearson correlation coefficient of 0.9010. This winning system was an ensembled model leveraging the output of IIT-MTL on ClinicalBERT with BioBERT, MT-DNN, and handcrafted medication features. CONCLUSIONS: This study demonstrates that IIT-MTL is an effective way to leverage annotated data from related tasks to improve performance on a target task with a limited data set. This contribution opens new avenues of exploration for optimized data set selection to generate more robust and universal contextual representations of text in the clinical domain.

Extraction of Information Related to Drug Safety Surveillance From Electronic Health Record Notes: Joint Modeling of Entities and Relations Using Knowledge-Aware Neural Attentive Models.

BACKGROUND: An adverse drug event (ADE) is commonly defined as "an injury resulting from medical intervention related to a drug." Providing information related to ADEs and alerting caregivers at the point of care can reduce the risk of prescription and diagnostic errors and improve health outcomes. ADEs captured in structured data in electronic health records (EHRs) as either coded problems or allergies are often incomplete, leading to underreporting. Therefore, it is important to develop capabilities to process unstructured EHR data in the form of clinical notes, which contain a richer documentation of a patient's ADE. Several natural language processing (NLP) systems have been proposed to automatically extract information related to ADEs. However, the results from these systems showed that significant improvement is still required for the automatic extraction of ADEs from clinical notes. OBJECTIVE: This study aims to improve the automatic extraction of ADEs and related information such as drugs, their attributes, and reason for administration from the clinical notes of patients. METHODS: This research was conducted using discharge summaries from the Medical Information Mart for Intensive Care III (MIMIC-III) database obtained through the 2018 National NLP Clinical Challenges (n2c2) annotated with drugs, drug attributes (ie, strength, form, frequency, route, dosage, duration), ADEs, reasons, and relations between drugs and other entities. We developed a deep learning-based system for extracting these drug-centric concepts and relations simultaneously using a joint method enhanced with contextualized embeddings, a position-attention mechanism, and knowledge representations. The joint method generated different sentence representations for each drug, which were then used to extract related concepts and relations simultaneously. Contextualized representations trained on the MIMIC-III database were used to capture context-sensitive meanings of words. The position-attention mechanism amplified the benefits of the joint method by generating sentence representations that capture long-distance relations. Knowledge representations were obtained from graph embeddings created using the US Food and Drug Administration Adverse Event Reporting System database to improve relation extraction, especially when contextual clues were insufficient. RESULTS: Our system achieved new state-of-the-art results on the n2c2 data set, with significant improvements in recognizing crucial drug-reason (F1=0.650 versus F1=0.579) and drug-ADE (F1=0.490 versus F1=0.476) relations. CONCLUSIONS: This study presents a system for extracting drug-centric concepts and relations that outperformed current state-of-the-art results and shows that contextualized embeddings, position-attention mechanisms, and knowledge graph embeddings effectively improve deep learning-based concepts and relation extraction. This study demonstrates the potential for deep learning-based methods to help extract real-world evidence from unstructured patient data for drug safety surveillance.

Timely and Efficient AI Insights on EHR: System Design.

A patient's electronic health record (EHR) contains extensive documentation of the patient's medical history but is difficult for clinicians to review and find what they are looking for under the time constraints of the clinical setting. Although recent advances in artificial intelligence (AI) in healthcare have shown promise in enhancing clinical diagnosis and decision-making in clinicians' day-to-day tasks, the problem of how to implement and scale such computationally expensive analytics remains an open issue. In this work, we present a system architecture that generates AI-based insights from analysis of the entire patient medical record for a multispecialty outpatient facility of over 700,000 patients. Our resulting system is able to generate insights efficiently while handling complexities of scheduling to deliver the results in a timely manner, and handle more than 30,000 updates per day while achieving desirable operating cost-performance goals.

Sex distribution and sex data handling in published otolaryngology research.

OBJECTIVES/HYPOTHESIS: To characterize the sex distribution and sex data handling in published otolaryngology research. STUDY DESIGN: Published research data analysis. METHODS: The total number of male and female participants, study characteristics, and sex data handling were abstracted from all original studies containing human participants published in five major otolaryngology journals from January 1, 2016 to December 31, 2016 and January 1, 2006 to December 31, 2006. RESULTS: Of the 1,128 studies from 2016 included in the analysis, 88.5% specified the sex of participants. There were 3,605,636 (42.1%) men and 4,515,508 (52.8%) women, with 429,006 (5.0%) participants unspecified. However, the average proportions of male and female participants (wherein studies are weighted the same, regardless of number of participants) were 0.579 and 0.421, respectively. Studies from the United States had a significantly higher proportion of women than studies from outside the United States. Subspecialties varied significantly in proportions. Average sex proportions in 2016 remained similar to those in 2006. For all studies, fewer than 40% of studies used any sex data for reporting of outcomes, for any sex-related analysis, or for discussion of results. CONCLUSIONS: There was a higher average proportion of male participants than female. Studies originating in the United States included a greater number of female participants than those originating elsewhere, a possible result of explicit sex-inclusion policies governing research in the United States. Inclusion of women did not changed from 2006 to 2016, but analysis of sex data improved. Improvement of reporting, analysis, and discussion with regard to sex would benefit otolaryngology research and improve treatment for both sexes. LEVEL OF EVIDENCE: NA Laryngoscope, 129:E420-E427, 2019.

EXOGENOUS TESTOSTERONE DOES NOT INDUCE OR EXACERBATE THE METABOLIC FEATURES ASSOCIATED WITH PCOS AMONG TRANSGENDER MEN.

OBJECTIVE: Polycystic ovary syndrome (PCOS) is a complex condition which can include menstrual irregularity, metabolic derangement, and increased androgen levels. The mechanism of PCOS is unknown. Some suggest that excess production of androgens by the ovaries may cause or exacerbate the metabolic findings. The purpose of this study was to assess the role of increased testosterone on metabolic parameters for individuals presumed to be chromosomally female by examination of these parameters in hormone-treated transgender men. METHODS: In 2015 and 2016, we asked all transgender men who visited the Endocrinology Clinic at Boston Medical Center treated with testosterone for consent for a retrospective anonymous chart review. Of the 36 men, 34 agreed (94%). Serum metabolic factors and body mass index (BMI) levels for each patient were graphed over time, from initiation of therapy through 6 years of treatment. Bivariate analyses were conducted to analyze the impact of added testosterone. RESULTS: Regressions measuring the impact of testosterone demonstrated no significant changes in levels of glycated hemoglobin (HbA1c), triglycerides, or low-density-lipoprotein cholesterol. There was a statistically significant decrease in BMI with increasing testosterone. There was also a statistically significant decrease in high-density lipoprotein levels upon initiation of testosterone therapy. CONCLUSION: Testosterone therapy in transgender men across a wide range of doses and over many years did not result in the dyslipidemia or abnormalities in HbA1c seen with PCOS. Instead, treatment of transgender men with testosterone resulted only in a shift of metabolic biomarkers toward the average physiologic male body. ABBREVIATIONS: BMI = body mass index; HbA1c = glycated hemoglobin; HDL = high-density lipoprotein; LDL = low-density lipoprotein; PCOS = polycystic ovary syndrome.

ESTROGEN LEVELS DO NOT RISE WITH TESTOSTERONE TREATMENT FOR TRANSGENDER MEN.

OBJECTIVE: Existing transgender treatment guidelines suggest that for transmasculine treatment, there is a possible need for estrogen-lowering strategies adjunct to testosterone therapy. Further, guidelines advocate consideration of prophylactic female reproductive tissue surgeries for transgender men to avoid the possibility of estrogen-related health risks. Despite the paucity of objective data, some transgender men seek conversion inhibitors. We sought to determine estradiol levels in transgender men treated with testosterone therapy and the change in those levels with treatment, if any. METHODS: Estradiol levels were extracted from the electronic medical records of 34 anonymized transgender men treated with testosterone therapy at the Endocrinology Clinic at Boston Medical Center. Data were sufficient to observe 6 years of follow-up. RESULTS: With increased testosterone levels in trans-gender men, a significant decrease in estradiol levels was noted. There was a significant negative correlation between testosterone levels and body mass index, which may serve to explain part of the mechanism for the fall in estradiol levels. Even though the fall in estradiol levels was significant statistically, the actual levels remained within the normal male range, even with 6 years of follow-up. CONCLUSION: These data suggest that when exogenous testosterone is used to achieve normal serum male testosterone levels for transgender men, it is converted to normal male levels of estradiol, with some decline in those estradiol levels that might be attributable to a fall in fat mass. There appears to be no role for aromatase conversion inhibitors or other estrogen-reducing strategies in trans-gender men. Abbreviation: BMI = body mass index.

Overview of the Molecular Steps in Steroidogenesis of the GABAergic Neurosteroids Allopregnanolone and Pregnanolone.

Allopregnanolone and pregnanolone-neurosteroids synthesized from progesterone in the brain, adrenal gland, ovary and testis-have been implicated in a range of neuropsychiatric conditions including seizure disorders, post-traumatic stress disorder, major depression, post-partum depression, pre-menstrual dysphoric disorder, chronic pain, Parkinson's disease, Alzheimer's disease, neurotrauma, and stroke. Allopregnanolone and pregnanolone equipotently facilitate the effects of gamma-amino-butyric acid (GABA) at GABAA receptors, and when sulfated, antagonize N-methyl-D-aspartate receptors. They play myriad roles in neurophysiological homeostasis and adaptation to stress while exerting anxiolytic, antidepressant, anti-nociceptive, anticonvulsant, anti-inflammatory, sleep promoting, memory stabilizing, neuroprotective, pro-myelinating, and neurogenic effects. Given that these neurosteroids are synthesized de novo on demand, this review details the molecular steps involved in the biochemical conversion of cholesterol to allopregnanolone and pregnanolone within steroidogenic cells. Although much is known about the early steps in neurosteroidogenesis, less is known about transcriptional, translational, and post-translational processes in allopregnanolone- and pregnanolone-specific synthesis. Further research to elucidate these mechanisms as well as to optimize the timing and dose of interventions aimed at altering the synthesis or levels of these neurosteroids is much needed. This should include the development of novel therapeutics for the many neuropsychiatric conditions to which dysregulation of these neurosteroids contributes.

TESTOSTERONE LEVELS ACHIEVED BY MEDICALLY TREATED TRANSGENDER WOMEN IN A UNITED STATES ENDOCRINOLOGY CLINIC COHORT.

OBJECTIVE: Most transgender women depend on medical treatment alone to lower testosterone levels in order to align physical appearance with gender identity. The medical regimen in the United States typically includes spironolactone and estrogens. The purpose of this cross-sectional study was to assess the testosterone suppression achieved among transgender women treated with spironolactone and estrogens. METHODS: Testosterone and estradiol levels were extracted from the electronic medical records of 98 anonymized transgender women treated with oral spironolactone and oral estrogen therapy at the Endocrinology Clinic at Boston Medical Center. RESULTS: Patients starting therapy required about 9 months to reach a steady-state testosterone, with significant heterogeneity of levels achieved among patients. Patients with normal body mass index (BMI) had higher testosterone levels, whereas patients with obese BMI had lower testosterone levels throughout treatment. Stratification of patients by age or spironolactone dosage revealed no significant difference in testosterone levels achieved. At steady state, patients in the highest suppressing quartile were able to achieve testosterone levels of 27 ng/dL, with a standard deviation of 21 ng/dL. Measured serum estradiol levels did not change over time and did not correlate with dosage of estradiol administered. CONCLUSION: Among a cohort of transgender women treated with spironolactone and estrogen, the highest suppressing quartile could reliably achieve testosterone levels in the female range at virtually all times. The second highest suppressing quartile could not achieve female levels but remained below the male range virtually all of the time. One quartile was unable to achieve any significant suppression. ABBREVIATIONS: BMC = Boston Medical Center BMI = body mass index CPY = cyproterone acetate LC-MS/MS = liquid chromatography-tandem mass spectrometry Q = quartile.

Ground Truth Creation for Complex Clinical NLP Tasks - an Iterative Vetting Approach and Lessons Learned.

Natural language processing (NLP) holds the promise of effectively analyzing patient record data to reduce cognitive load on physicians and clinicians in patient care, clinical research, and hospital operations management. A critical need in developing such methods is the "ground truth" dataset needed for training and testing the algorithms. Beyond localizable, relatively simple tasks, ground truth creation is a significant challenge because medical experts, just as physicians in patient care, have to assimilate vast amounts of data in EHR systems. To mitigate potential inaccuracies of the cognitive challenges, we present an iterative vetting approach for creating the ground truth for complex NLP tasks. In this paper, we present the methodology, and report on its use for an automated problem list generation task, its effect on the ground truth quality and system accuracy, and lessons learned from the effort.

SemanticFind: Locating What You Want in a Patient Record, Not Just What You Ask For.

We present a new model of patient record search, called SemanticFind, which goes beyond traditional textual and medical synonym matches by locating patient data that a clinician would want to see rather than just what they ask for. The new model is implemented by making extensive use of the UMLS semantic network, distributional semantics, and NLP, to match query terms along several dimensions in a patient record with the returned matches organized accordingly. The new approach finds all clinically related concepts without the user having to ask for them. An evaluation of the accuracy of SemanticFind shows that it found twice as many relevant matches compared to those found by literal (traditional) search alone, along with very high precision and recall. These results suggest potential uses for SemanticFind in clinical practice, retrospective chart reviews, and in automated extraction of quality metrics.