Physicians make critical time-constrained decisions every day. Clinical predictive models can help physicians and administrators make decisions by forecasting clinical and operational events. Existing structured data-based clinical predictive models have limited use in everyday practice owing to complexity in data processing, as well as model development and deployment1–3. Here we show that unstructured clinical notes from the electronic health record can enable the training of clinical language models, which can be used as all-purpose clinical predictive engines with low-resistance development and deployment. Our approach leverages recent advances in natural language processing4,5 to train a large language model for medical language (NYUTron) and subsequently fine-tune it across a wide range of clinical and operational predictive tasks. We evaluated our approach within our health system for five such tasks: 30-day all-cause readmission prediction, in-hospital mortality prediction, comorbidity index prediction, length of stay prediction, and insurance denial prediction. We show that NYUTron has an area under the curve (AUC) of 78.7–94.9%, with an improvement of 5.36–14.7% in the AUC compared with traditional models. We additionally demonstrate the benefits of pretraining with clinical text, the potential for increasing generalizability to different sites through fine-tuning and the full deployment of our system in a prospective, single-arm trial. These results show the potential for using clinical language models in medicine to read alongside physicians and provide guidance at the point of care.

Background The current study aims to characterize the natural history of sporadic vestibular schwannoma volumetric tumor growth, including long-term growth patterns following initial detection of growth. Methods Volumetric tumor measurements from 3,505 serial MRI studies were analyzed from unselected consecutive patients undergoing wait-and-scan management at three tertiary referral centers between 1998 and 2018. Volumetric tumor growth was defined as a change in volume ≥20%. Results Among 952 patients undergoing observation, 622 experienced tumor growth with initial growth-free survival rates (95% CI) at 1, 3, and 5 years following diagnosis of 66% (63-69), 30% (27-34), and 20% (17-24). Among 405 patients who continued to be observed despite demonstrating initial growth, 210 experienced subsequent tumor growth with subsequent growth-free survival rates at 1, 3, and 5 years following initial growth of 77% (72-81), 37% (31-43), and 24% (18-31). Larger tumor volume at initial growth (HR 1.13, p=0.02) and increasing tumor growth rate (HR 1.31; p<0.001) were significantly associated with an increased likelihood of subsequent growth, whereas a longer duration of time between diagnosis and detection of initial growth was protective (HR 0.69; p<0.001). Conclusions While most vestibular schwannomas exhibit an overall propensity for volumetric growth following diagnosis, prior tumor growth does not perfectly predict future growth. Tumors can subsequently grow faster, slower, or demonstrate quiescence and stability. Larger tumor size and increasing tumor growth rate portend a higher likelihood of continued growth. These findings can inform timing of intervention: whether upfront at initial diagnosis, after detection of initial growth, or only after continued growth is observed.

OBJECTIVEThere remains a large discrepancy among surgeons in expectations of vestibular schwannoma (VS) growth. The anticipated growth rate of a VS and its potential clinical impact are important factors when deciding whether to observe the lesion over time or to intervene. Previous studies of VS natural growth remain limited, mostly confined to linear measurements, often without high-resolution, thin-sequence imaging. The present study comprehensively assessed natural tumor growth rates using volumetric measurements.METHODSBetween 2012 and 2018, 212 treatment-naïve patients diagnosed with a unilateral VS were evaluated. A total of 699 MR images were assessed, with a range of 2–11 MR images per patient. All MR images preceded any intervention, with patients subsequently being observed through completion of data analysis (36%) or treated with stereotactic radiosurgery (32%) or microsurgical resection (32%). To determine precise tumor volumes, the tumor area was outlined on every slice, and the products of the area and slice thickness were summed (99% of scans were ≤ 1-mm slice thickness). A multilevel model with random effects was used to assess the mean volume change over time. Each tumor was categorized as one of the following: growing (volume increase by more than 20% per year), fast growing (volume increase by more than 100% per year), stable (volume change between 20% decrease and 20% increase per year), and shrinking (volume decrease by more than 20% per year).RESULTSThe mean VS volumetric growth rate was 33.5% per year (95% CI 26.9%–40.5%, p < 0.001). When assessing the frequencies of individual tumor annual growth rates, 66% demonstrated growth (30% fast growing), 33% were stable, and 1% exhibited shrinking over an average interval of 25 months. Larger tumors were associated with increased absolute growth, but there was no relationship between tumor size and proportional growth rate. There was also no relationship between patient age and tumor growth rate.CONCLUSIONSThis study comprehensively assessed VS volumetric growth rates using high-resolution images and was conducted in a large and diverse patient sample. The majority of the tumors exhibited growth, with about one-third growing at a rate of 100% per year. These findings may contribute to a consensus understanding of tumor behavior and inform clinical decisions regarding whether to intervene or observe.

Understanding how the relevant medical community accepts new therapies is vital to patients, physicians, and society. Increasingly, focus is placed on how medical innovations are evaluated. But recognizing when a treatment has become accepted practice-essentially, acceptance by the scientific community-remains a challenge and a barrier to investigating treatment development. This report aims to demonstrate the theory, method, and limitations of a model for measuring a new metric that the authors term "progressive scholarly acceptance." A model was developed to identify when the scientific community has accepted an innovation, by observing when researchers have moved beyond the initial study of efficacy. This model could enable further investigations into the methods and influences of treatment development.

______________________________________________________________________________________________________________________________________ -------------------------INDICATIONS AND USAGE-------------------------------G GL LI IA AD DE EL L W Wa af fe er r i is s a an n a al lk ky yl la at ti in ng g d dr ru ug g i in nd di ic ca at te ed d f fo or r t th he e t tr re ea at tm me en nt t o of f: :   n ne ew wl ly y--d di ia ag gn no os se ed d h hi ig gh h--g gr ra ad de e--m ma al li ig gn na an nt t g gl li io om ma a a as s a an n a ad dj ju un nc ct t t to o s su ur rg ge er ry y a an nd d r ra ad di ia at ti io on n ( (1 1) ) a an nd d   r re ec cu ur rr re en nt t g gl li io ob bl la as st to om ma a m mu ul lt ti if fo or rm me e a as s a an n a ad dj ju un nc ct t t to o s su ur rg ge er ry y ( (1 1) )

Objective: To describe a novel neurophysiologic signature of the retinal ganglion cell and to elucidate its relationship to abnormalities in validated structural and functional measures of the visual system. Methods:We used multifocal electroretinogram-generated optic nerve head component (ONHC) responses from normal subjects (n 5 18), patients with multiple sclerosis (MS) (n 5 18), and those with glaucoma (n 5 3). We then characterized the relationship between ONHC response abnormalities and performance on low-contrast visual acuity, multifocal visual-evoked potentialinduced cortical responses, and average and quadrant retinal nerve fiber layer (RNFL) thicknesses, as measured by spectral-domain optical coherence tomography.Results: Compared with the eyes of normal subjects, the eyes of patients with MS exhibited an increased number of abnormal or absent ONHC responses (p , 0.0001). For every 7-letter reduction in low-contrast letter acuity, there were corresponding 4.6 abnormal ONHC responses at 2.5% contrast (p , 0.0001) and 6.6 abnormalities at the 1.25% contrast level (p , 0.0001). Regarding average RNFL thickness, for each 10-mm thickness reduction, we correspondingly observed 6.8 abnormal ONHC responses (p 5 0.0002). The most robust association was between RNFL thinning in the temporal quadrant and ONHC response abnormalities (p , 0.0001). Conclusion:Further characterization of ONHC abnormalities (those that are reversible and irreversible) may contribute to the development of novel neurotherapeutic strategies aimed at achieving neuroprotective, and perhaps even neurorestorative, effects in disorders that target the CNS in general, and MS in particular. Neurology ® 2014;82:1888-1896 GLOSSARY AON 5 acute optic neuritis; CI 5 confidence interval; GEE 5 generalized estimating equation; ICC 5 intraclass correlation coefficient; mfERG 5 multifocal electroretinography; mfVEP 5 multifocal visual-evoked potential; MS 5 multiple sclerosis; OCT 5 optical coherence tomography; ONHC 5 optic nerve head component; RGC 5 retinal ganglion cell; RNFL 5 retinal nerve fiber layer.The eye has been proposed as a "window" into the CNS. High-precision techniques have evolved and are accelerating scientific discovery, particularly regarding protective and restorative neurotherapeutic effects. [1][2][3][4] Whereas spectral-domain optical coherence tomography (OCT) has been of great utility in precisely characterizing the topography of retinal architecture, multifocal electroretinography (mfERG) is a technique that can yield information on discrete functional responses within the retina, following a stimulus with precisely defined characteristics. [5][6][7][8][9][10][11][12][13][14][15] The retinal ganglion cell (RGC) contribution to the global retinal response is small, thereby limiting the utility of this technique for the purpose of examining retinal mechanisms of axonal and neuronal degeneration. 16*These authors contributed equally to these investigations as senior authors.

Studying the relationship between ONHC abnormalities and alterations in validated structural and functional measures of the visual system may facilitate the ability to dissect and characterize the pathobiological mechanisms that contribute to tissue damage in MS, and may have utility to detect and monitor neuroprotective or restorative effects of novel therapies.

BACKGROUND:The most significant paradigm shift in surgical management of skull base chordomas has been the adoption of the endoscopic endonasal approach, but the impact on patient outcomes compared with open skull base approaches remains unclear. OBJECTIVE: To compare a large series of patients treated by a single surgeon using primarily endoscopic endonasal approaches with previously published outcomes by the same surgeon using open skull base approaches. METHODS: Between 2006 and 2020, 68 patients with skull base chordoma underwent resection using primarily endoscopic endonasal approaches. Outcomes and complications were compared with previously published results of resection of chordomas from 1991 to 2005 using open skull base approaches. RESULTS: Compared with the prior cohort, the current principally endoscopic cohort demonstrated similar rates of OS (P = .86) and progression-free survival (P = .56), but patients undergoing first-time resection had significantly higher rates of radical resection (82.9% compared with 64.3%, P = .05) and required fewer staged surgeries (9.8% compared with 33.3%, P = .01). CONCLUSION: There was no difference in survival rates for patients treated in the current era, primarily using endoscopic endonasal techniques, compared with previously published results using open skull-base approaches by the same surgeon. Although use of endoscopic endonasal approach resulted in higher rates of radical resection, patients undergoing first-time resection and fewer staged surgeries were required.