BACKGROUND AND PURPOSE: Posterior fossa tumors are the most common pediatric brain tumors. MR imaging is key to tumor detection, diagnosis, and therapy guidance. We sought to develop an MR imaging-based deep learning model for posterior fossa tumor detection and tumor pathology classification. MATERIALS AND METHODS:The study cohort comprised 617 children (median age, 92 months; 56% males) from 5 pediatric institutions with posterior fossa tumors: diffuse midline glioma of the pons (n ¼ 122), medulloblastoma (n ¼ 272), pilocytic astrocytoma (n ¼ 135), and ependymoma (n ¼ 88). There were 199 controls. Tumor histology served as ground truth except for diffuse midline glioma of the pons, which was primarily diagnosed by MR imaging. A modified ResNeXt-50-32x4d architecture served as the backbone for a multitask classifier model, using T2-weighted MRIs as input to detect the presence of tumor and predict tumor class. Deep learning model performance was compared against that of 4 radiologists.
Hypertension is highly prevalent and morbid in the chronic kidney disease population, and blood pressure (BP) targets for this population are unclear. We aimed to compare all-cause mortality outcomes with intensively targeting systolic BP to <130 mm Hg versus a standard of <140 mm Hg. Individual patient data from 4983 chronic kidney disease patients with hypertension were pooled from 4 multicenter randomized control trials—AASK (African American Study of Kidney Disease and Hypertension), ACCORD (Action to Control Cardiovascular Risk in Diabetes), MDRD (Modification of Diet in Renal Disease), and the SPRINT (Systolic Blood Pressure Intervention Trial). Patients were assigned their trial-assigned randomized intervention group—standard (n=2474) versus intensive (n=2509) BP targets. Additional analyses included excluding patients with a glomerular filtration rate ≥60 mL/min per 1.73 m 2 along with those undergoing intensive glycemic control. The primary outcome was all-cause mortality. Average achieved BP was 125.0 mm Hg in the intensive group and 136.9 mm Hg in the standard group. In the primary analysis, the all-cause mortality rate trended towards improved outcomes with intensive treatment but was not statistically significant (hazard ratio: 0.87 [0.69–1.08]; P =0.21). One hundred seventy-three of 2474 patients (1.95% per year) in the standard group and 153 of 2509 patients (1.71% per year) in the intensive group died. After excluding patients with higher glomerular filtration rate values and those undergoing intensive glycemic control, there was a statistically significant decrease in all-cause mortality rate (hazard ratio: 0.79 [0.63–1.00]; P =0.048). An intensive BP target of <130 mm Hg decreases all-cause mortality when compared with a standard target of <140 mm Hg in patients with chronic kidney disease stage 3 or greater who are not undergoing intensive glycemic therapy.
Background Incidental radiographic findings, such as adrenal nodules, are commonly identified in imaging studies and documented in radiology reports. However, patients with such findings frequently do not receive appropriate follow-up, partially due to the lack of tools for the management of such findings and the time required to maintain up-to-date lists. Natural language processing (NLP) is capable of extracting information from free-text clinical documents and could provide the basis for software solutions that do not require changes to clinical workflows. Objectives In this manuscript we present (1) a machine learning algorithm we trained to identify radiology reports documenting the presence of a newly discovered adrenal incidentaloma, and (2) the web application and results database we developed to manage these clinical findings. Methods We manually annotated a training corpus of 4,090 radiology reports from across our institution with a binary label indicating whether or not a report contains a newly discovered adrenal incidentaloma. We trained a convolutional neural network to perform this text classification task. Over the NLP backbone we built a web application that allows users to coordinate clinical management of adrenal incidentalomas in real time. Results The annotated dataset included 404 positive (9.9%) and 3,686 (90.1%) negative reports. Our model achieved a sensitivity of 92.9% (95% confidence interval: 80.9–97.5%), a positive predictive value of 83.0% (69.9–91.1)%, a specificity of 97.8% (95.8–98.9)%, and an F1 score of 87.6%. We developed a front-end web application based on the model's output. Conclusion Developing an NLP-enabled custom web application for tracking and management of high-risk adrenal incidentalomas is feasible in a resource constrained, safety net hospital. Such applications can be used by an institution's quality department or its primary care providers and can easily be generalized to other types of clinical findings.
Background Clinicians spend large amounts of their workday using electronic medical records (EMRs). Poorly designed documentation systems contribute to the proliferation of out-of-date information, increased time spent on medical records, clinician burnout, and medical errors. Beyond software interfaces, examining the underlying paradigms and organizational structures for clinical information may provide insights into ways to improve documentation systems. In particular, our attachment to the note as the major organizational unit for storing unstructured medical data may be a cause of many of the problems with modern clinical documentation. Notes, as currently understood, systematically incentivize information duplication and information scattering, both within a single clinician’s notes over time and across multiple clinicians’ notes. Therefore, it is worthwhile to explore alternative paradigms for unstructured data organization. Objective The aim of this study is to demonstrate the feasibility of building an EMR that does not use notes as the core organizational unit for unstructured data and which is designed specifically to disincentivize information duplication and information scattering. Methods We used specific design principles to minimize the incentive for users to duplicate and scatter information. By default, the majority of a patient’s medical history remains the same over time, so users should not have to redocument that information. Clinicians on different teams or services mostly share the same medical information, so all data should be collaboratively shared across teams and services (while still allowing for disagreement and nuance). In all cases where a clinician must state that information has remained the same, they should be able to attest to the information without redocumenting it. We designed and built a web-based EMR based on these design principles. Results We built a medical documentation system that does not use notes and instead treats the chart as a single, dynamically updating, and fully collaborative workspace. All information is organized by clinical topic or problem. Version history functionality is used to enable granular tracking of changes over time. Our system is highly customizable to individual workflows and enables each individual user to decide which data should be structured and which should be unstructured, enabling individuals to leverage the advantages of structured templating and clinical decision support as desired without requiring programming knowledge. The system is designed to facilitate real-time, fully collaborative documentation and communication among multiple clinicians. Conclusions We demonstrated the feasibility of building a non–note-based, fully collaborative EMR system. Our attachment to the note as the only possible atomic unit of unstructured medical data should be reevaluated, and alternative models should be considered.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.