Evaluating Large Language Models for Drafting Emergency Department Discharge Summaries.

Importance: Large language models (LLMs) possess a range of capabilities which may be applied to the clinical domain, including text summarization. As ambient artificial intelligence scribes and other LLM-based tools begin to be deployed within healthcare settings, rigorous evaluations of the accuracy of these technologies are urgently needed. Objective: To investigate the performance of GPT-4 and GPT-3.5-turbo in generating Emergency Department (ED) discharge summaries and evaluate the prevalence and type of errors across each section of the discharge summary. Design: Cross-sectional study. Setting: University of California, San Francisco ED. Participants: We identified all adult ED visits from 2012 to 2023 with an ED clinician note. We randomly selected a sample of 100 ED visits for GPT-summarization. Exposure: We investigate the potential of two state-of-the-art LLMs, GPT-4 and GPT-3.5-turbo, to summarize the full ED clinician note into a discharge summary. Main Outcomes and Measures: GPT-3.5-turbo and GPT-4-generated discharge summaries were evaluated by two independent Emergency Medicine physician reviewers across three evaluation criteria: 1) Inaccuracy of GPT-summarized information; 2) Hallucination of information; 3) Omission of relevant clinical information. On identifying each error, reviewers were additionally asked to provide a brief explanation for their reasoning, which was manually classified into subgroups of errors. Results: From 202,059 eligible ED visits, we randomly sampled 100 for GPT-generated summarization and then expert-driven evaluation. In total, 33% of summaries generated by GPT-4 and 10% of those generated by GPT-3.5-turbo were entirely error-free across all evaluated domains. Summaries generated by GPT-4 were mostly accurate, with inaccuracies found in only 10% of cases, however, 42% of the summaries exhibited hallucinations and 47% omitted clinically relevant information. Inaccuracies and hallucinations were most commonly found in the Plan sections of GPT-generated summaries, while clinical omissions were concentrated in text describing patients' Physical Examination findings or History of Presenting Complaint. Conclusions and Relevance: In this cross-sectional study of 100 ED encounters, we found that LLMs could generate accurate discharge summaries, but were liable to hallucination and omission of clinically relevant information. A comprehensive understanding of the location and type of errors found in GPT-generated clinical text is important to facilitate clinician review of such content and prevent patient harm.

BiRD: Using Bidirectional Rotation Gain Differences to Redirect Users during Back-and-forth Head Turns in Walking.

Redirected walking (RDW) facilitates user navigation within expansive virtual spaces despite the constraints of limited physical spaces. It employs discrepancies between human visual-proprioceptive sensations, known as gains, to enable the remapping of virtual and physical environments. In this paper, we explore how to apply rotation gain while the user is walking. We propose to apply a rotation gain to let the user rotate by a different angle when reciprocating from a previous head rotation, to achieve the aim of steering the user to a desired direction. To apply the gains imperceptibly based on such a Bidirectional Rotation gain Difference (BiRD), we conduct both measurement and verification experiments on the detection thresholds of the rotation gain for reciprocating head rotations during walking. Unlike previous rotation gains which are measured when users are turning around in place (standing or sitting), BiRD is measured during users' walking. Our study offers a critical assessment of the acceptable range of rotational mapping differences for different rotational orientations across the user's walking experience, contributing to an effective tool for redirecting users in virtual environments.

A Deep Learning Algorithm to Identify Anatomical Landmarks on Computed Tomography of the Temporal Bone.

BACKGROUND: Petrous temporal bone cone-beam computed tomography scans help aid diagnosis and accurate identification of key operative landmarks in temporal bone and mastoid surgery. Our primary objective was to determine the accuracy of using a deep learning convolutional neural network algorithm to augment identification of structures on petrous temporal bone cone-beam computed tomography. Our secondary objective was to compare the accuracy of convolutional neural network structure identification when trained by a senior versus junior clinician. METHODS: A total of 129 petrous temporal bone cone-beam computed tomography scans were obtained from an Australian public tertiary hospital. Key intraoperative landmarks were labeled in 68 scans using bounding boxes on axial and coronal slices at the level of the malleoincudal joint by an otolaryngology registrar and board-certified otolaryngologist. Automated structure identification was performed on axial and coronal slices of the remaining 61 scans using a convolutional neural network (Microsoft Custom Vision) trained using the labeled dataset. Convolutional neural network structure identification accuracy was manually verified by an otolaryngologist, and accuracy when trained by the registrar and otolaryngologist labeled datasets respectively was compared. RESULTS: The convolutional neural network was able to perform automated structure identification in petrous temporal bone cone-beam computed tomography scans with a high degree of accuracy in both axial (0.958) and coronal (0.924) slices (P < .001). Convolutional neural network accuracy was proportionate to the seniority of the training clinician in structures with features more difficult to distinguish on single slices such as the cochlea, vestibule, and carotid canal. CONCLUSION: Convolutional neural networks can perform automated structure identification in petrous temporal bone cone-beam computed tomography scans with a high degree of accuracy, with the performance being proportionate to the seniority of the training clinician. Training of the convolutional neural network by the most senior clinician is desirable to maximize the accuracy of the results.

Chemokine receptor 5 signaling in PFC mediates stress susceptibility in female mice.

Chronic stress induces changes in the periphery and the central nervous system (CNS) that contribute to neuropathology and behavioral abnormalities associated with psychiatric disorders. In this study, we examined the impact of peripheral and central inflammation during chronic social defeat stress (CSDS) in female mice. Compared to male mice, we found that female mice exhibited heightened peripheral inflammatory response and identified C-C motif chemokine ligand 5 (CCL5), as a stress-susceptibility marker in females. Blocking CCL5 signaling in the periphery promoted resilience to CSDS. In the brain, stress-susceptible mice displayed increased expression of C-C chemokine receptor 5 (CCR5), a receptor for CCL5, in microglia in the prefrontal cortex (PFC). This upregulation was associated with microglia morphological changes, their increased migration to the blood vessels, and enhanced phagocytosis of synaptic components and vascular material. These changes coincided with neurophysiological alterations and impaired blood-brain barrier (BBB) integrity. By blocking CCR5 signaling specifically in the PFC were able to prevent stress-induced physiological changes and rescue social avoidance behavior. Our findings are the first to demonstrate that stress-mediated dysregulation of the CCL5-CCR5 axis triggers excessive phagocytosis of synaptic materials and neurovascular components by microglia, resulting in disruptions in neurotransmission, reduced BBB integrity, and increased stress susceptibility. Our study provides new insights into the role of cortical microglia in female stress susceptibility and suggests that the CCL5-CCR5 axis may serve as a novel sex-specific therapeutic target for treating psychiatric disorders in females.

A Computer Vision Algorithm to Classify Pneumatization of the Mastoid Process on Temporal Bone Computed Tomography Scans.

BACKGROUND: Pneumatization of the mastoid process is variable and of significance to the operative surgeon. Surgical approaches to the temporal bone require an understanding of pneumatization and its implications for surgical access. This study aims to determine the feasibility of using deep learning convolutional neural network algorithms to classify pneumatization of the mastoid process. METHODS: De-identified petrous temporal bone images were acquired from a tertiary hospital radiology picture archiving and communication system. A binary classification mode in the pretrained convolutional neural network was used to investigate the utility of convolutional neural networks in temporal bone imaging. False positive and negative images were reanalyzed by the investigators and qualitatively assessed to consider reasons for inaccuracy. RESULTS: The overall accuracy of the model was 0.954. At a probability threshold of 65%, the sensitivity of the model was 0.860 (95% CI 0.783-0.934) and the specificity was 0.989 (95% CI 0.960-0.999). The positive predictive value was 0.973 (95% CI 0.904-0.993) and the negative predictive value was 0.935 (95% CI 0.901-0.965). The false positive rate was 0.006. The F1 number was 0.926 demonstrating a high accuracy for the model. CONCLUSION: The temporal bone is a complex anatomical region of interest to otolaryngologists. Surgical planning requires high-resolution computed tomography scans, the interpretation of which can be augmented with machine learning. This initial study demonstrates the feasibility of utilizing machine learning algorithms to discriminate anatomical variation with a high degree of accuracy. It is hoped this will lead to further investigation regarding more complex anatomical structures in the temporal bone.

Association between disability status and dental attendance patterns in Australian children: A national survey.

OBJECTIVES: This study investigated the dental attendance patterns of Australian children with and without disabilities using data from Growing up in Australia: The Longitudinal Study of Australian Children. METHODS: Data on 6470 participants within two groups (B cohort [aged 12-13]: n = 3381; K cohort [aged 16-17]: n = 3089) were used for the study. Binomial regression models were fitted to examine the association between disability status and dental attendance. The models were adjusted for gender, parent's country of birth, region of residence, highest parental education and household weekly income, and multiple imputations was used for handling missing data. RESULTS: Children with disabilities constituted 2.4% and 3.8% of the study sample in the B and K cohort, respectively. The unadjusted risk ratio of irregular (vs. regular) dental attendance between children with and without disabilities was 1.07 (95% CI 0.78-1.46) in the B cohort and 1.15 (95% CI 0.93-1.42) in the K cohort. After adjustment and imputation, the risk ratios were 1.03 (95% CI 0.76-1.41) and 1.10 (95% CI 0.89-1.36) in the B and K cohort, respectively. CONCLUSIONS: Dental attendance pattern was positively, but minimally, associated with disabilities in older children, and factors including region of residence, parental education and household income were related to disability status and dental attendance. Further studies are required to clarify the association and ascertain key factors that affect the health and wellbeing of children with disabilities.

Benefits of dietary polyphenols in Alzheimer's disease.

Alzheimer's disease (AD) is an irreversible progressive neurodegenerative disease affecting approximately 50 million people worldwide. It is estimated to reach 152 million by the year 2050. AD is the fifth leading cause of death among Americans age 65 and older. In spite of the significant burden the disease imposes upon patients, their families, our society, and our healthcare system, there is currently no cure for AD. The existing approved therapies only temporarily alleviate some of the disease's symptoms, but are unable to modulate the onset and/or progression of the disease. Our failure in developing a cure for AD is attributable, in part, to the multifactorial complexity underlying AD pathophysiology. Nonetheless, the lack of successful pharmacological approaches has led to the consideration of alternative strategies that may help delay the onset and progression of AD. There is increasing recognition that certain dietary and nutrition factors may play important roles in protecting against select key AD pathologies. Consistent with this, select nutraceuticals and phytochemical compounds have demonstrated anti-amyloidogenic, antioxidative, anti-inflammatory, and neurotrophic properties and as such, could serve as lead candidates for further novel AD therapeutic developments. Here we summarize some of the more promising dietary phytochemicals, particularly polyphenols that have been shown to positively modulate some of the important AD pathogenesis aspects, such as reducing ß-amyloid plaques and neurofibrillary tangles formation, AD-induced oxidative stress, neuroinflammation, and synapse loss. We also discuss the recent development of potential contribution of gut microbiome in dietary polyphenol function.

Hypotension in Posterior Retroperitoneoscopic Versus Transperitoneal Laparoscopic Adrenalectomy.

INTRODUCTION: Despite preoperative optimization, hemodynamic instability can be a major challenge during adrenalectomy. Even brief episodes of intraoperative hypotension can be associated with ischemia-reperfusion injury. This study aimed to compare intraoperative hemodynamic parameters between posterior retroperitoneoscopic adrenalectomy (PRA) and transperitoneal laparoscopic adrenalectomy (TPA). METHODS: This is a retrospective study of patients undergoing PRA and TPA without conversion or concomitant intraabdominal pathology from 2008 to 2019. The primary outcome was intraoperative hypotension defined by mean arterial pressure <60 mm Hg or the need for ≥1 intravenous vasopressors at least 30 min after anesthetic induction. RESULTS: Overall, 108 patients met the inclusion criteria; 33 (30.6%) had pheochromocytoma, 26 (24.1%) had aldosterone excess, 8 (7.4%) had corticosteroid excess, and 41 (38.0%) had nonfunctioning adrenal tumors. Of these, 68 (63.0%) underwent PRA and 40 (37.0%) underwent TPA. Age, sex, body mass index, preinduction blood pressure, number of preoperative antihypertensives, and histopathological diagnosis were similar in the two groups. Tumor size was greater in the TPA group. The presence of pheochromocytoma was an independent risk factor for hypotension. Multivariate analysis revealed that PRA was associated with a higher risk of experiencing a mean arterial pressure <60 mm Hg (odds ratio 4.44, 95% confidence interval 1.27-15.54, P = 0.02) and the need for ≥1 intravenous vasopressors (odds ratio 9.97, 95% confidence interval 3.34-29.78, P < 0.001) compared with TPA. CONCLUSIONS: Although PRA offers several advantages over TPA, it carries a greater risk of intraoperative hypotension. A prospective trial is required to validate these findings. Nevertheless, institution of risk reduction strategies is encouraged to be considered for individuals undergoing PRA.

DICOM Image ANalysis and Archive (DIANA): an Open-Source System for Clinical AI Applications.

In the era of data-driven medicine, rapid access and accurate interpretation of medical images are becoming increasingly important. The DICOM Image ANalysis and Archive (DIANA) system is an open-source, lightweight, and scalable Python interface that enables users to interact with hospital Picture Archiving and Communications Systems (PACS) to access such data. In this work, DIANA functionality was detailed and evaluated in the context of retrospective PACS data retrieval and two prospective clinical artificial intelligence (AI) pipelines: bone age (BA) estimation and intra-cranial hemorrhage (ICH) detection. DIANA orchestrates activity beginning with post-acquisition study discovery and ending with online notifications of findings. For AI applications, system latency (exam completion to system report time) was quantified and compared to that of clinicians (exam completion to initial report creation time). Mean DIANA latency was 9.04 ± 3.83 and 20.17 ± 10.16 min compared to clinician latency of 51.52 ± 58.9 and 65.62 ± 110.39 min for BA and ICH, respectively, with DIANA latencies being significantly lower (p < 0.001). DIANA's capabilities were also explored and found effective in retrieving and anonymizing protected health information for "big-data" medical imaging research and analysis. Mean per-image retrieval times were 1.12 ± 0.50 and 0.08 ± 0.01 s across x-ray and computed tomography studies, respectively. The data herein demonstrate that DIANA can flexibly integrate into existing hospital infrastructure and improve the process by which researchers/clinicians access imaging repository data. This results in a simplified workflow for large data retrieval and clinical integration of AI models.

Non-diagnostic sonography may reduce negative appendicectomy rate in women when combined with abbreviated Alvarado score.

BACKGROUND: The diagnosis of acute appendicitis (AA) remains a clinical one, with selective use of adjunct imaging. Patients with equivocal clinical presentation often undergo a diagnostic laparoscopy. To help reduce negative appendicectomy rates in women, we aimed to develop a simple scoring system based on the Alvarado score (AS) and ultrasound scan (US), as a diagnostic aid for AA in females. METHODS: All patients who underwent appendicectomy for AA at The Alfred Hospital Melbourne between 1 July 2012 and 30 June 2017 were included for this case-control study. Logistic regression was used to identify pre-operative parameters predictive of AA. Histopathological identification of AA was interpreted as the gold standard. Statistical analysis was performed using IBM SPSS Statistics V26. RESULTS: A total of 1194 patients were included, with 26% negative appendicectomy rate in women. Of the 8 parameters in the AS, logistic regression identified migratory pain, leukocytosis and leukocyte left shift as most significant predictors for AA. These three parameters were used in a 3-point test which carried a sensitivity of 92.1% and specificity of 28.7%. In women, a negative or non-diagnostic US improved the negative predictive value of the 3-point test from 57% to 82%. CONCLUSION: The 3-point abbreviated AS in combination with US may be clinically useful in women to exclude appendicitis without diagnostic laparoscopy. Further large-scale prospective studies are required to validate the utility across different subgroups.

Caseload distribution of general surgeons in regional Australia: is there a role for a rural surgery sub-specialization?

BACKGROUND: Rural and regional Australia presents a unique challenge in the delivery of surgical services by virtue of its geographical vastness and low population density. While up to 33% of Australians live in rural or regional areas, only 14.8% of surgeons work in a rural or regional area. Data regarding the caseload distribution of general surgeons working in a regional setting in Australia remain scarce. In order to better examine the training needs of rural general surgeons, this study aims to examine the caseload distribution of general surgeons working in regional Australia. METHODS: A retrospective review of surgical procedures carried out by general surgeons at Albury Base Hospital between September 2006 and December 2014 was carried out. Surgical procedures were grouped according to the classifications of the Royal Australasian College of Surgeons Morbidity and Audit Logbook Tool. RESULTS: During the study period, 21 652 procedures were carried out by general surgeons. A total of 58.7% (12711) of these procedures consisted of general surgical procedures and 35.9% (7763) were endoscopic procedures. A total of 5.4% of procedures carried out by general surgeons fell outside the scope of traditional general surgery, including cardiothoracic, orthopaedic, ear, nose and throat, neurosurgical, vascular and urological procedures. All general surgeons performed operations in surgical specialities outside of general surgery. CONCLUSION: This study adds weight to the value of a broad skill set in provision of surgical services in a rural setting.

Evaluation of novel Griess-reagent candidates for nitrite sensing in aqueous media identified via molecular fingerprint searching.

The Griess reaction is the most often exploited colorimetric method for the quantitative analysis of nitrite in aqueous media. The application of the currently used reagents are associated with limitations (e.g. linear response range). Herein, molecular fingerprint searching on well-known Griess-reagents was used as a tool for the identification of structurally similar, new reagent candidate molecules. Rapid and high-throughput experimental evaluation of the newly identified Griess-reagent candidates revealed that 14 of the 18 tested reagent candidates had equal or superior response displaying broader linear ranges and/or increased response gradient against various nitrite concentrations in aqueous media when compared to the parent compounds at room temperature.

Tuning microtubule dynamics to enhance cancer therapy by modulating FER-mediated CRMP2 phosphorylation.

Though used widely in cancer therapy, paclitaxel only elicits a response in a fraction of patients. A strong determinant of paclitaxel tumor response is the state of microtubule dynamic instability. However, whether the manipulation of this physiological process can be controlled to enhance paclitaxel response has not been tested. Here, we show a previously unrecognized role of the microtubule-associated protein CRMP2 in inducing microtubule bundling through its carboxy terminus. This activity is significantly decreased when the FER tyrosine kinase phosphorylates CRMP2 at Y479 and Y499. The crystal structures of wild-type CRMP2 and CRMP2-Y479E reveal how mimicking phosphorylation prevents tetramerization of CRMP2. Depletion of FER or reducing its catalytic activity using sub-therapeutic doses of inhibitors increases paclitaxel-induced microtubule stability and cytotoxicity in ovarian cancer cells and in vivo. This work provides a rationale for inhibiting FER-mediated CRMP2 phosphorylation to enhance paclitaxel on-target activity for cancer therapy.

Qualitative Evaluation of a High-Resolution 3D Multi-Sequence Intracranial Vessel Wall Protocol at 3 Tesla MRI.

BACKGROUND AND PURPOSE: Intracranial vessel wall imaging using MRI has great potential as a clinical method for assessing intracranial atherosclerosis. The purpose of the current study was to compare three 3T MRI vessel wall sequences with different contrast weightings (T1w, PD, T2w) and dedicated sagittal orientation perpendicular to the middle cerebral artery, to the reconstructed sagittal image from a transverse 3D T1w volumetric isotropically reconstructed turbo spin-echo acquisition (VIRTA), and provide a clinical recommendation. MATERIALS AND METHODS: The above-mentioned sequences were acquired in 10 consecutive Chinese ischemic stroke or TIA patients (age: 68 years, sex: 4 females) with angiographic-confirmed MCA stenosis at 3T. Institutional review board approval was obtained. Two raters qualitatively scored all images on overall image quality, presence of artifacts, and visibility of plaques. Data were compared using Repeated measures ANOVA and Sidak's adjusted post hoc tests. RESULTS: All sequences except the T2w sequence were able to depict the walls of the large vessels of the Circle of Willis (p<0.05). T1w sagittal oblique VIRTA showed significantly more artifacts (p<0.01). Peripherally located plaques were sometimes missed on the sagittal sequences, but could be appreciated on the transverse T1w VIRTA. CONCLUSION: With the 3T multi-sequence vessel wall protocol we were able to assess the intracranial plaque with two different image contrast weightings. The sequence of preference to include in a clinical protocol would be the transverse 3D T1w VIRTA based on absence of artifacts, larger coverage including the whole Circle of Willis, and excellent lesion depiction.

Advanced arterial blood gas analysis in septic shock: a Singaporean nursing case review.

INTRODUCTION: The admission to the intensive care unit with a diagnosis of sepsis and/or septic shock is not uncommon. The aim of this article is to present a nursing case review of a patient admitted to the intensive care unit with a diagnosis of septic shock and the use of bedside acid-base formulae to inform clinical decision making. METHOD: We chose to use a case review. This method is useful in reporting unusual or rare cases and is typically seen more in medicine than in nursing. DISCUSSION: The gentleman in question was a self-presentation with a short history of fever and worsening shortness of breath. His condition worsened where he required admission to the intensive care unit. The use of 'advanced' acid-base interpretation to guide his nursing care provided a platform from which to advance a deeper understanding of the intricacies the critically ill patient often presents. CONCLUSION: The use of case review is enlightening in understanding the disease process and the decision-making that accompanies this. The lessons learnt are applicable to a wider nursing audience because understanding acid-base physiology is beneficial in supporting and advancing critical care nursing practice.