Non-home discharge after orthopedic trauma is associated with worse outcomes, increased costs, and greater resource utilization. Existing prediction tools often rely on hospital course variables unavailable at presentation or are limited to specific fracture populations. This study aimed to develop and validate the Fracture Orthopedic Risk of Non-Home Discharge (FORD) Score, a bedside tool using only emergency department-available variables to predict non-home discharge in adult fracture patients. A retrospective cohort study was conducted of adult fracture patients treated at an ACS-verified Level I trauma center from 2015 to 2023. Patients were randomly split into derivation (67%) and validation (33%) cohorts. Candidate predictors available immediately upon patient arrival were evaluated using univariate logistic regression, followed by multivariate logistic regression after collinearity assessment. Independent predictors were converted into an integer-based point system to construct the FORD Score. Model discrimination, calibration, and classification performance were assessed in the validation cohort and compared with established trauma severity measures. The final cohort included 8422 patients, of whom 8.1% had non-home discharge. Fifteen independent predictors comprised the FORD Score, including age, physiologic abnormalities, fracture characteristics, and transport mode. In the validation cohort, FORD demonstrated good discrimination (AUROC 0.818, 95% CI 0.791-0.846) and excellent calibration. At the optimal threshold (score ≥4), sensitivity was 74.1%, specificity 75.8%, PPV 21.3%, and NPV 97.1%. FORD outperformed GTOS-II (AUROC 0.777; DeLong p = 0.018) and TRIAGES (AUROC 0.746; p < 0.001). Non-home discharge rates ranged from 1.7% in the lowest risk group to 34.1% in the highest, a 20-fold gradient. The FORD Score is a validated bedside tool that accurately predicts non-home discharge in adult orthopedic trauma patients using only admission data, enabling early discharge planning and optimized resource allocation.
To evaluate whether biomarkers of systemic inflammation and alterations in redox homeostasis, which we refer to as oxidative stress, associate with musculoskeletal injury (MSKI) during US Army Basic Combat Training (BCT) and to characterise longitudinal changes in these biomarkers throughout BCT. This prospective observational study included 206 Army trainees (51% female) undergoing BCT. Blood samples were collected and analysed for high-sensitivity C reactive protein (hsCRP), free oxygen radical test (FORT), free oxygen radical defence (FORD) and Oxidative Stress Index (OSI=FORT/FORD). Injuries were identified using the International Classification of Diseases, Tenth Revision (ICD-10) codes. Mixed-effects logistic regression models assessed associations between biomarker levels and injury diagnosis over three timeframes (0-7 days before draw, 1-7 days following draw and 8-14 days following draw). Models included both chronic (between-person) and acute (within-person) biomarker components. Inflammation and oxidative stress biomarker associations with MSKI were strongest in the 0-7 days before and 0-7 days after injury diagnosis. Acute elevation in hsCRP (OR=1.41, 95% CI 1.03 to 1.93, p=0.034) was associated with 41% higher odds of injury diagnosis within the next 7 days. Chronically high hsCRP and OSI were also associated with increased MSKI risk (OR 2.27, 95% CI 1.22 to 4.19, p=0.01 and OR=1.72, 95% CI 1.01 to 2.92, p=0.046, respectively). Elevated hsCRP and OSI were temporally associated with MSKI diagnoses during BCT, with the strongest associations in the week surrounding diagnosis. These associative findings may reflect heightened physiological stress and early injury-related tissue stress and repair responses.
A 15-year-old right-handed boy presented 7 weeks after a hyperextension injury while playing basketball with a suspected metacarpophalangeal (MCP) joint dislocation to his right thumb and subsequent reduction by his mother at the time of injury. Initial radiographs were negative for acute fracture or dislocation. However, pain and stiffness persisted in the MCP joint, and subsequent magnetic resonance imaging demonstrated an incarcerated radial sesamoid and a proximal ulnar collateral ligament (UCL) injury with migration of the ulnar sesamoid under the proximal UCL, a finding not previously documented in the literature. He underwent surgical excision of the sesamoid bone and repair of the UCL injury with successful improvement in symptoms and return to all activities, including basketball, by his 4-month postoperative visit. This case demonstrates that it is imperative to scrutinize radiographs and consider advanced imaging when evaluating a locked MCP joint following a hyperextension injury.
暂无摘要(点击查看详情)
Triple-negative breast cancer (TNBC) is an aggressive subtype associated with higher recurrence rates and inferior survival compared with hormone receptor-positive disease. The addition of immune checkpoint inhibition to neoadjuvant chemotherapy has improved pathologic complete response (pCR) and event-free survival in early-stage TNBC and is now the standard of care for high-risk disease. However, patients with end-stage renal disease (ESRD) requiring dialysis were excluded from pivotal clinical trials, leaving limited evidence to guide treatment in this population. We report the case of a 41-year-old woman with clinical stage IIB (cT2N1M0) TNBC and ESRD on chronic peritoneal dialysis who received modified neoadjuvant chemoimmunotherapy based on the KEYNOTE-522 regimen. Chemotherapy dosing was individualized for renal failure, including flat-dose carboplatin and dose-reduced anthracycline and cyclophosphamide, while pembrolizumab was administered at standard dosing. She completed neoadjuvant therapy and 16 of 17 planned pembrolizumab cycles (final cycle omitted due to toxicity) with overall manageable adverse effects. Surgical pathology following lumpectomy and sentinel lymph node biopsy demonstrated pCR (residual cancer burden score of 0). At one-year follow-up, she remains without evidence of recurrence. This case demonstrates the feasibility of delivering curative-intent TNBC chemoimmunotherapy in a patient undergoing peritoneal dialysis and highlights the importance of multidisciplinary coordination in managing malignancy in patients with advanced renal disease.
Secondary hemophagocytic lymphohistiocytosis (HLH) is a rare and life-threatening hyperinflammatory syndrome that can occur after acute infection, predominantly caused by viruses. Due to its nonspecific presentation and clinical overlap with conditions like sepsis, HLH is frequently underrecognized in adults. We present the case of an immunocompetent adult admitted with undifferentiated shock, which rapidly progressed to multi-organ failure. Upon further investigation, the patient was diagnosed with HLH, and a comprehensive infectious workup revealed streptococcal pharyngitis as the precipitating cause. This case emphasizes the importance of recognizing HLH in the differential diagnosis in immunocompetent adults with pharyngitis presenting with shock, as early recognition is crucial for timely intervention and improved outcomes. Additionally, we review the current literature to characterize the reported infectious causes of HLH in immunocompetent adults in order to highlight the importance of having a low index of suspicion and conducting a comprehensive infectious workup.
Rx Kids, launched in Flint, Michigan, in January 2024, is the United States' first community-wide unconditional cash transfer program for expectant mothers and infants. The program offers all expectant mothers in the City of Flint a lump sum $1,500 during mid-pregnancy and $500 monthly for 12 months postnatally. The present study examined whether the Rx Kids program was associated with differences in parenting stress during the postpartum months. Using structural equation modeling in a sample of N = 954 mothers (Mage: 28.9 years; race/ethnicity: 37% Black, 52% White, 11% other), Flint mothers who were eligible to participate in Rx Kids (newborns born in 2024) reported lower parenting stress (0.36 of a SD lower) than noneligible Flint mothers (newborns born in 2023). This birth-timing difference (2024 vs. 2023) was not observed among non-Flint mothers from the surrounding area, providing quasi-experimental evidence that Rx Kids may reduce postpartum parenting stress. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
Large language models (LLMs) are increasingly used for clinical information retrieval and decision support, yet comparative performance on pharmacy board examination-style content across specialties remains incompletely characterized. We evaluated 15 LLMs using 145 publicly available Board of Pharmacy Specialties (BPS) certification practice questions spanning 14 specialty domains. Questions were entered using a standardized prompt without additional prompt engineering. Model responses were scored against BPS-posted answer keys. Overall and specialty-level accuracy were summarized descriptively. Differences among LLMs were tested using Cochran's Q with Bonferroni-adjusted McNemar pairwise comparisons when appropriate, and LLMs were assessed using their default user-facing settings. Across all LLMs, mean accuracy was 86.2% (standard deviation [SD], 3.5%), corresponding to an average of 125/145 items answered correctly. Accuracy ranged from 79.3% (95% confidence interval [CI], 72.6%-86%) for Perplexity AI to 91.7% (95% CI, 87.2%-96.3%) for Microsoft Copilot (GPT-5). Overall performance differed significantly across LLMs (Cochran's Q = 46.262; df = 14; p < 0.001). After Bonferroni adjustment, Microsoft Copilot (GPT-5), Google Gemini 2.5 Flash, and OpenAI o3 (Reasoning) outperformed Perplexity AI (p < 0.001). Microsoft Copilot (GPT-5) also outperformed an earlier version of Microsoft Copilot (GPT-4.1) (p < 0.001). Specialty-level heterogeneity was generally limited, with significant model differences observed in Solid Organ Transplantation Pharmacy and Nuclear Pharmacy. LLMs demonstrated high accuracy on BPS certification practice questions, with limited variability across LLMs and select specialty domains. These findings support continued evaluation of LLMs for potential use in pharmacy practice and clinical decision support, emphasizing the need for domain-specific validation and ongoing monitoring as LLMs evolve.
Body mass index (BMI) is widely used to guide surgical candidacy in total joint arthroplasty (TJA), with many institutions and payors applying uniform BMI thresholds, commonly ≥ 40, to both total hip arthroplasty (THA) and total knee arthroplasty (TKA). However, several differences between these procedures may influence how BMI affects postoperative infection risk. Emerging evidence suggests that BMI-related risk may differ between THA and TKA. This study assessed whether BMI confers differential 90-day postoperative deep infection risk between THA and TKA using a large statewide arthroplasty registry. A large United States statewide registry was queried for all primary THA and TKA procedures performed from 2019 to 2022. A total of 173,834 procedures were included (105,963 TKA; 67,871 THA). Participant BMI was analyzed categorically (less than 20, 20 to 24, 25 to 29, 30 to 34, 35 to 40, 40 to 45, and greater than 45) and continuously. The primary outcome was 90-day deep infection. Multivariable logistic regressions adjusted for demographic, clinical, and surgical factors. Non-parametric spline-smoothed logistic regressions were used to identify BMI thresholds at which infection risk significantly increased. Overall, 90-day infection rates were 0.40% for TKA and 0.62% for THA. Infection risk increased progressively across BMI categories for both procedures. Spline analysis demonstrated distinct inflection points: for THA, risk began rising near BMI 30 and increased sharply at higher BMI values. For TKA, infection risk remained relatively stable until approximately BMI 40, after which the slope of risk steepened. Patient BMI influences postoperative infection risk differently in THA and TKA. The THA patients experience rising infection risk beginning at a BMI greater than 30, whereas TKA risk increases modestly until a BMI greater than 40. These findings challenge uniform BMI cutoffs and support procedure-specific thresholds to improve patient selection, counseling, and perioperative optimization.
A substantial proportion of persons with inflammatory bowel disease (IBD) in remission continue to experience abdominal pain, altered bowel habits, and bloating that resemble irritable bowel syndrome (IBS). Lack of standardized definitions and evidence-based management strategies leads to diagnostic ambiguity and potentially unnecessary escalation of IBD therapy. A joint Rome Foundation/International Organization for the Study of Inflammatory Bowel Disease Working Team developed consensus recommendations on nomenclature, evaluation, and treatment of IBD with IBS-like symptoms. A multidisciplinary international panel applied a modified RAND/UCLA Appropriateness Method. Systematic literature reviews informed statement generation across multiple domains: nomenclature, diagnostic and symptom assessment, dietary therapies, drugs, and brain-gut behavioral therapies. Panelists rated the appropriateness of candidate statements independently on a 9-point Likert scale, followed by anonymized feedback, discussion, and re-voting across two iterative rounds. Thirteen panelists reviewed 133 initial statements; 105 proceeded to final scoring. Of these, 86 were rated appropriate, 16 uncertain, and 3 inappropriate. The preferred term was "IBD with IBS-like symptoms," defined as abdominal pain, bowel habit change, and/or bloating not explained by active inflammation or structural disease. For clinical care, diagnosis should combine Rome clinical criteria with objective exclusion of inflammation. For research, candidate thresholds for endoscopic, histologic, biomarker, and imaging remission were endorsed. Appropriate therapies included psyllium (if no stricture), a short-term low FODMAP diet, targeted drugs, and brain-gut behavioral therapies. This first joint consensus provides standardized terminology, evaluation strategies, and treatment recommendations for IBD with IBS-like symptoms, supporting improved clinical management and guiding future mechanistic and therapeutic research.
Targeted covalent inhibition of protein function is increasingly used as a therapeutic mode of action; however, there is a need to characterize off-target binding interactions and to understand whether this represents an immunological risk. Given that the proton-pump inhibitor omeprazole exerts its mechanism of action through covalent inhibition, it serves as an ideal model to investigate the relationship between off-target protein binding and T-cell activation. Binding of omeprazole, omeprazole metabolites and alternative proton-pump inhibitors to antigen presenting cells and GST-pi was characterised by mass spectrometry. Omeprazole-responsive clones were generated and assessed in terms of cytokine secretion, pathways of T-cell activation and crossreactivity with omeprazole metabolites, alternative proton-pump inhibitors and unrelated drugs. Omeprazole stimulated CD4+ and CD8+ T-cell clones to proliferate and secrete cytokines and cytolytic molecules. HLA-restricted T-cell activation was dependent on processing of omeprazole protein adducts by antigen presenting cells. Omeprazole-modified CYS-containing peptides derived from 36 off-target proteins were detected within antigen presenting cells. Omeprazole metabolites and alternative protein pump inhibitors that form protein adducts also activated omeprazole-responsive T-cells. In conclude, T-cells were activated with omeprazole via a hapten mechanism and exhibited considerable promiscuity to metabolites and structurally-related drugs of the same pharmacological class. Similar off-target binding interactions may be a relevant concern for the increasing number of covalent inhibitor drugs receiving regulatory approval.
Antigen-specific B cell profiling uncovers key determinants of SARS-CoV-2 antibody breadth after infection and vaccination.
Light a candle for human animals the people of Gaza;Light a candle because promises were made-"I have ordered a complete siege on the Gaza Strip. There will be no electricity, no food, no fuel, everything is closed."-and kept.Hold a candle in the crevice where a crushed child hopes until her last gasp evaporates.Light a candle to find her fingers, which still wiggle though the arm is feet from the rest of the body.Light a candle and scream.Weep.Then light a candle and stand.Your outrage, your righteous indignation, they will be used against you.So, light a candle to express your compassion meditatively.Light a candle in perpetual vigil.Light a candle while you fast.Light a candle for reason. For the triumph of science over faith, except when it matters.Light a candle if you are afraid to speak.Hold the candle firmly and speak, nevertheless. Genocide.Light a candle for praises on the lips of the dying.Light a candle for unencumbered transitions to the place where souls rest.Light a candle for Mother Earth, fat lady of the Global South.She has yet to sing.To view the original version of this poem, see the supplemental material section of this article online.
Background: Artificial intelligence (AI) is rapidly transforming nursing education and clinical practice. As AI becomes increasingly embedded in health care delivery, integrating AI competencies into Doctor of Nursing Practice (DNP) education is essential to prepare advanced practice registered nurses (APRNs) to utilize these tools effectively and ethically. Objective: This manuscript examines the integration of AI into DNP education, addressing policy implications, best practices, and strategies to prepare APRNs for leadership in AI-enhanced environments. Methods: A review of institutional innovations and faculty strategies demonstrates the application of AI in nursing education through adaptive learning platforms, virtual simulations, predictive analytics, and AI-driven clinical decision support systems. Case exemplars highlight implementation approaches and educational outcomes. Results: AI-enhanced tools have demonstrated several benefits, such as improved student engagement, individualized learning, and enhanced clinical reasoning. Case-based reflections revealed enhanced decision-making, mentorship, and student competency tracking. Limitations and potential risks of AI are also identified. Key guiding principles include evaluating existing competencies within the context of AI capabilities, defining emerging AI needs, supporting faculty development through AI training, and advancing policies for responsible and ethical AI use. Conclusions: The nursing profession is well recognized for its innovative approach to adopting new technologies. Embedding AI into DNP education requires intentional curricular reform, strong leadership support, and ethical oversight to ensure sustainable adoption. Nursing faculty must champion the strategic and responsible use of AI to prepare APRNs for evidence-based, technology-driven practice. DNP-prepared nurses, with their expertise in quality improvement and ethical practice, are uniquely positioned to shape the development and implementation of AI tools.
The Dallas consensus conference on liver transplantation for alcohol associated hepatitis (2020) provided a framework for selection of liver transplant candidates with alcohol associated liver disease (ALD) regardless of sobriety period. The primary aim of this study is to describe our experience with the implementation of this approach to a broader population with ALD. We established a new interdisciplinary ALD clinic in a large integrated health care system in Texas for expedited assessment for liver transplant (LT) candidacy. Candidates that had short sobriety were mandated to follow this pathway. Selected patients were seen through a structured interdisciplinary clinic pathway before and after transplantation with an ALD hepatologist, surgeon, coordinator and transplant psychologist. Between 2021 and 2024, 114 patients (MELD 25 (21-28), 47% female) were referred to the ALD clinic. Alcohol use disorder (AUD) and mental health related comorbid conditions were high. Overall, 97 (85%) were willing to participate in AUD counseling and mental health evaluation and attendance with ALD clinic was high (78%). The most common form of treatment preference was outpatient peer support groups (81%) and individual therapy (50%). Amongst the referrals, 23% were denied LT candidacy; 12% died during their evaluation and 50% were approved to proceed to LT evaluation. Among LT candidates, 13% clinically improved to the point where they no longer required LT and 32% patients received a LT. In intent-to treat analysis, survival at 1 year was similar for patients that were referred to ALD clinic as compared to ALD patients referred to the overall transplant program through the traditional mechanisms (89.5 vs 82.9, p=0.22). There were no deaths related to alcohol use. Relapse to alcohol use occurred in 6 patients (16%); 4 were defined as slips, and 2 had sustained alcohol use; relapse rates were similar to non-ALD clinic patients (p=0.258). Successful early LT assessment can be achieved in the setting of a structured pathway with excellent survival and low rates of relapse. Patients are adherent to recommendations and actively engage before and after LT once the appropriate programmatic structure is created.
Pulsed field ablation (PFA) using a variable loop circular catheter (VLCC) is increasingly adopted for atrial fibrillation (AF) ablation. Early experience suggests a favorable safety profile, yet outcomes from complex procedures remain limited. To evaluate the safety profile of the VLCC from a large, multicenter, real-world registry. Consecutive ablations performed with the VLCC in the prospective, multicenter REAL AF registry were evaluated for safety events through 3 months follow-up. Events were categorized by investigator-assessed relationship to procedure and/or VLCC. In 1,014 index ablation cases, the procedure-attributed complication rate was 0.6% (6/1,014). Complications included: cardiac tamponade/pericardial effusion attributed to a concomitant left atrial appendage closure device, sinoatrial block following accessory pathway ablation, iliac artery dissection, fluid overload, blurred vision with normal magnetic resonance imaging (MRI), and transient ischemic attack. Half of events occurred in cases involving adjunctive RFA. No strokes, coronary spasm, or deaths were reported. Complication rates were low in all AF types and ablation strategies. Same-day discharge occurred in 86.5% of cases, and acute pulmonary vein isolation was achieved in 99.6%. In a multicenter real-world cohort that included persistent AF and more complex strategies, the VLCC demonstrated a favorable early safety profile with no catheter-attributed adverse events and very low procedure-attributed complications. These findings support the feasibility of efficient, outpatient-focused PFA workflows while ongoing accrual will enable more granular assessment of rare events.
Psychogenic non-epileptic seizures (PNES) are functional episodes that mimic epileptic seizures without epileptiform activity. We describe a 16-year-old female with trauma history, mood disorder, and a recent suicide attempt who developed recurrent PNES during hospitalization. Episodes featured unresponsiveness, irregular shaking, and heightened sympathetic arousal with tachycardia, hypertension, diaphoresis, and pupillary dilation up to 8 mm. Events were often prolonged or clustered and triggered by environmental stressors. Given concerns for epileptic seizures, the patient underwent medical evaluation, which revealed no epileptiform activity captured on video-electroencephalogram during events and was consistent with a diagnosis of PNES. Management included monitoring, sertraline titration, and behavioral strategies such as mindfulness techniques and minimizing unnecessary staff interventions during episodes. By discharge, the patient no longer endorsed suicidal ideation; had shown a reduction in frequency and severity of PNES episodes; and was motivated to continue care. This case highlights pupillary dilation as an underrecognized finding in PNES and underscores the need for further study to clarify its prevalence, mechanisms, and diagnostic implications.
TikTok has surged in popularity as a primary source of entertainment for Americans, with many physicians leveraging the platform to disseminate up-to-date medical information. Within this landscape, dermatology has emerged as one of the most sought-after medical subjects, leading to the prominence of numerous board-certified dermatologists as influential figures. However, despite this visibility, dermatology remains among the least diverse medical specialties. This cross-sectional study investigates patterns of following and content reliability on TikTok, particularly concerning dermatological information, based on the racial, gender, and sexual orientation diversity of top influencers. Through qualitative and quantitative analysis of the top 55 dermatologist influencers on TikTok, based on data collected on October 10, 2022, we assess the demographics and video characteristics, with a focus on follower count. Our study found that content produced by Latinx and African American dermatologists demonstrated relatively higher DISCERN scores, although overall content reliability across all groups was low. Additionally, our findings underscore a significant lack of minority representation among dermatology influencers on TikTok, particularly among Latinx, African American, and LGBTQIA+ physicians. This lack of diversity may limit the availability of culturally representative dermatologic information, highlighting a potential gap that has been associated in prior literature with inequities in care and health outcomes. By addressing these diversity gaps, we can work towards fostering more inclusive and equitable healthcare environments on social media platforms.
To characterize the frequency, nature, timing, and transparency of prespecified outcome modifications in interventional trials for inflammatory sinonasal disease registered on ClinicalTrials.gov. We identified interventional trials for inflammatory sinonasal disease registered on ClinicalTrials.gov with published primary results. Earliest and most recent registry versions were compared to identify substantive modifications to primary, secondary, and other prespecified outcomes. Modifications were categorized by anticipated interpretive impact (high or moderate), and timing relative to trial completion and publication was recorded. Disclosure of outcome modifications was assessed in registry records, including posted protocol documents when available, and in associated peer-reviewed publications. Descriptive analyses summarized modification patterns, and unadjusted regression analyses explored factors associated with high-impact modifications. Seventy-two trials met inclusion criteria. Substantive outcome modifications occurred in 68 trials (94.4%), with primary outcomes affected in 59 trials (81.9%). The median number of substantive modifications per trial was seven (interquartile range, 3-10). Common modification types included clarification or increased specification, changes in assessment timing, outcome additions, and complete redefinitions involving changes in measurement instruments or methodologies. All recorded modifications were entered after primary completion or publication. Disclosure was uncommon: 3 of 68 modified trials (4.4%) acknowledged changes in the registry, 1 (1.5%) in the publication, and 4 (5.9%) in either source. Trials prespecifying more than three outcomes and trials completed after implementation of the FDAAA Final Rule had lower odds of high-impact modification. Outcome modifications are frequent in sinonasal trials but are rarely transparently documented. Given the central role of symptom-based, quality-of-life, and endoscopic outcomes in rhinology, clearer documentation of outcome changes is essential to support interpretability and evidence synthesis. 4.
Ambulatory hypercalcemia is a proxy for primary hyperparathyroidism. Reports of an increased incidence of hypercalcemia and undiagnosed primary hyperparathyroidism in several electronic medical record studies have prompted a population-based trend analysis of serum calcium. Data from the 2000-2020 U.S. National Health and Nutritional Examination Survey were used to study the trend of serum calcium and related factors. The NHANES has contemporary insight into the ambulatory state of health in the large and diverse U.S. population. Joinpoint regression estimated yearly changes of serum calcium and related factors using annual percentage changes. Serum calcium levels increased by an average of 0.65 mg/dL/y from 2000 to 2004 and then decreased on average by 0.12 mg/dL annually from 2004 to 2020. Among women, serum calcium levels increased by an average of 0.69 mg/dL/y from 2000 to 2004 but then decreased on average by 0.13 mg/dL annually from 2004 to 2020. Among men, serum calcium levels increased by an average of 0.61 mg/dL/y from 2000 to 2004 and then remained stable. Trends of body mass index increased by an average of 0.49/y from 2014 to 2020. Ambulatory hypercalcemia is a proxy for primary hyperparathyroidism. Over 20 years in the U.S. National Health and Nutritional Examination Survey (2000-2020), calcium levels have been decreasing slightly since 2004 after an increase while body mass index has been increasing since 2014. These data conflict with reported observations of the undiagnosed and increased incidence of primary hyperparathyroidism. These data may ultimately serve to refine primary hyperparathyroidism data phenotype for machine learning deployed within an electronic medical record.