Deep Learning Applications for Acute Stroke Management.

Strategies for stroke management

We integrate literature- and data-driven task analysis methods to derive an initial task taxonomy for electronic health record (EHR) and electronic medical record (EMR) data analysis. An EHR (EMR) is a digital and longitudinal version of a patients health(medical) information and may include all key clinical events relevant to that persons health (medical) history, such as provider, demographics, progress notes, medicine, diagnosis, etc. Our goal is to arrive a task taxonomy for analyzing EHR (EMR) datasets because tasks play an important role in the design and evaluation of visualization techniques. Our method has three stages: data collection, task modelling, and task taxonomy summary. In data collection, we first survey related literature from the past two decades and extract typical tasks and corresponding data by extracting goals and scenarios of the particular work. We introduce multiple continuous relations to describe specific binary or multiple continuous relation-seeking tasks. Finally, we arrive an initial set of task types for EHR/EMR analysis that guide the design and evaluation of visualization techniques.

The incidence rate of stroke in hospitalized patients ranges between 2% and 17% of all strokes-a higher rate than in the community. Delays in recognition and management of stroke in hospitalized patients lead to worse outcomes. At our hospital, the existing in-hospital stroke (IHS) code showed low usage and effectiveness. In a quality improvement (QI) project, we aimed to improve the identification of and the quality of care for inpatient strokes. A nurse-driven IHS protocol was implemented, which alerted a specialized stroke team and cleared the computed tomography (CT) scanner. The protocol focused on prioritizing staff education, simplifying the process, empowering staff to activate an IHS code, ensuring adequate support and teamwork, identifying well-defined quality metrics (eg, time to CT and documentation tool use), and providing feedback communication. We analyzed 2 years of postimplementation IHS data for impact on stroke detection and outcomes. In the 2 years post QI, there was a more than 10-fold increase in IHS (pre-QI, n = 8; first year post QI, n = 94; second year post QI, n = 123). In the post-QI cohort, after excluding patients with missing information (n = 26), 69 cases had new stroke diagnoses (63 ischemic, 6 hemorrhagic), and 148 were stroke mimics. The mean (SD) time from IHS to CT was 18.7 (7.0) minutes. Of the 63 new ischemic stroke cases, 25 (39.7%) were treated with thrombolytic therapy and/or mechanical thrombectomy. The new IHS protocol has led to a marked increase in cases identified, rapid evaluation, and high utilization rate of acute stroke therapies.

Electronic Medical Record (EMR) data are an efficient method for constructing large type 1 diabetes (T1D) cohorts but are limited by availability of accurate diagnosis codes. We sought to derive an algorithm for identifying T1D using Electronic Medical Records - Primary Care (EMRPC) also known as EMRALD, a database including records from >300,000 primary care patients in Ontario, Canada. A 25% random sample of adults with diabetes and at least 1 year of records in EMRPC prior to September 30, 2015 formed the reference cohort. Charts of potential T1D cases (those on insulin±metformin) were abstracted; all others were classified as non-T1D. Algorithms were derived using variables from free-text searches (diagnosis terms in the Cumulative Patient Profile (CPP)) and standardized fields (medications, age, BMI) using 1)a priori combinations of variables; 2)classification and regression tree analysis (CART). Algorithms were evaluated for sensitivity, specificity, positive and negative predictive values (PPV and NPV) with adjustment for optimism for CART. Of 21,547 eligible patients with diabetes, the reference cohort was 5407: 4968 non-abstracted non-T1D, 240 abstracted non-T1D and 199 T1D. The prevalence of T1D was 3.7%. The optimal algorithm using a priori variable combinations was T1D diagnosis in the CPP + rapid-acting insulin (sensitivity 69.3% (95% CI 62.4-75.7%), specificity 99.8% (99.6-99.9), PPV 92.6% (87.2-96.3), NPV 98.8% (98.5-99.1)). CART partitioned on T1D diagnosis in the CPP, any insulin, and non-metformin oral hypoglycemic medications and improved performance with optimism-adjusted sensitivity of 77% (72.0-83.9), specificity 99.8% (99.8-100.0), PPV 96.8% (94.6-99.6), and NPV 99.1% (98.9-99.4). Simple algorithms using EMR variables yielded good diagnostic performance for identification of T1D and CART further improved performance. Pending additional validation these algorithms can be applied to study large T1D cohorts in EMR databases. Disclosure A. Weisman: None. J. Young: None. M. Kumar: None. P. Austin: None. K. Tu: None. L. Jaakkimainen: None. L. Lipscombe: None. G. Booth: None. Funding Diabetes Action Canada

Recent advancements in deep learning techniques tailored for autonomous driving in Indian road conditions are crucial for revolutionizing transportation systems. Indian roads present unique challenges, including unpredictable traffic patterns, diverse road infrastructures, and challenging weather conditions. Deep learning is pivotal in addressing these challenges, focusing on perception, decision-making, and control. Various architectures, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Deep Reinforcement Learning (DRL), are analyzed for their efficacy in autonomous driving. Sensor fusion techniques enhance perception capabilities, such as lidar, radar, and camera data integration. Additionally, advancements in semantic segmentation and object detection algorithms improve scene understanding and obstacle recognition. Collaborative efforts between academia, industry, and government agencies are essential for accelerating the deployment of autonomous driving technology in Indian road conditions. In conclusion, leveraging cutting-edge deep learning methodologies enables safe and efficient navigation of autonomous vehicles, ushering in a transformative future in transportation systems.

This study analyzed GE Centricity Electronic Medical Record (EMR) data to examine the effects of body mass index (BMI) and obesity, key risk factor components of metabolic syndrome, on the prevalence of 3 chronic diseases: type II diabetes mellitus, hyperlipidemia, and hypertension. These chronic diseases occur with high prevalence and impose high disease burdens. The rationale for using Centricity EMR data is 2-fold. First, EMRs may be a good source of BMI/obesity data, which are often underreported in surveys and administrative databases. Second, EMRs provide an ideal means to track variables over time and, thus, allow longitudinal analyses of relationships between risk factors and disease prevalence and progression. Analysis of Centricity EMR data showed associations of age, sex, race/ethnicity, and BMI with diagnosed prevalence of the 3 conditions. Results include uniform direct correlations between age and BMI and prevalence of each disease; uniformly greater disease prevalence for males than females; varying differences by race/ethnicity (ie, African Americans have the highest prevalence of diagnosed type II diabetes and hypertension, while whites have the highest prevalence of diagnosed hypertension); and adverse effects of comorbidities. The direct associations between BMI and disease prevalence are consistent for males and females and across all racial/ethnic groups. The results reported herein contribute to the growing literature about the adverse effects of obesity on chronic disease prevalence and about the potential value of EMR data to elucidate trends in disease prevalence and facilitate longitudinal analyses.

IntroductionFrailty is a combination of factors that increase vulnerability to functional decline, dependence and/or death. Frailty cannot easily be defined by comorbidities or medical treatment alone. Accurate detection of frailty in practice and at a population level is needed. This may be achieved using a combination of data sources.
 Objectives and ApproachWe construct algorithms that can identify frailty using electronic medical record (EMR) and administrative data. We linked EMR data from the Canadian Primary Care Sentinel Surveillance Network (CPCSSN) nodes and the administrative (e.g. billings, hospitalizations) from Population Data BC and the Manitoba Health Policy Centre. Frailty was defined as individuals 65+ who were receiving home services, had specific chronic conditions, received specific diagnoses, and/or had specific lab or other clinical indicators. We describe sociodemographic characteristics, risk factors, prescribed medications, use and costs of healthcare for those identified as frail.
 ResultsPeople were identified as frail in 2014 and all analysis was completed with 2015 data. Among those who were > 65 years, who had a record in both EMR and administrative data, 5\%-8\% (n=191 of 3,553, BC; n=2,396 of 29,382, MB) were identified as frail. There was a higher likelihood of being frail with increasing age and being a woman. In BC, those identified as frail have higher contacts with primary care (n=20 vs. n=10) and more days in hospital (n=7.4 vs. n=2.0) compared to those who are not frail. Twenty two percent of those identified as frail in 2014 died in 2015, compared to a mortality rate of 2\% among those who are not frail.
 Conclusion/ImplicationsIdentifying and reporting on those who are frail in primary care as well as in communities could enable targeted communications with patients and families and community based resources in order to improve patient care, patients’ and caregivers’ quality of life and better use of the healthcare system.

This study aims to identify stages of Duchenne muscular dystrophy (DMD) and assess the disease burden by progression stage using real-world administrative claims supplemented by relevant electronic medical record (EMR) data. Claims and EMR data from the Decision Resources Group's Real World Data Repository (2011-2020) were used to identify patients with DMD by diagnosis code and to stratify them into four disease stages by diagnosis and procedure markers reflective of DMD progression. Clinical and medical history data from the Cooperative International Neuromuscular Research Group (CINRG) were used to validate the developed claims-based staging algorithm. The distribution and drivers by disease stage, as well as disease burden, were examined. A total of 938 (94%) of patients with DMD identified in claims/EMR data had sufficient information for stage classification. Patients were classified by stage based on patient characteristics and the presence or absence of progression markers such as genetic testing, wheelchair usage, scoliosis treatment, or ventilation assistance. Average ages at stages1-4 are 7, 13, 18, and 23years, respectively. Using natural history data, the claims-based staging algorithm was validated with high sensitivity and specificity rates. Both healthcare resource utilization and medical charges increased by stage. For example, the average annualized total charges were $17,688 (stage1), $36,868 (stage2), $72,801 (stage3), and $167,285 (stage4). Large-scale claims data supplemented by EMR data can be used to characterize DMD progression and evaluate disease burden which may inform the design of future real-world studies about DMD.

Advancements in deep learning technologies and an increase in medical data have enhanced the accuracy of disease diagnosis and treatment strategies. Notably, significant progress has been made in the use of deep learning-based time-series prediction models for short-term disease onset prediction and analysis of important features. However, research on explainable deep learning for long-term disease progression, such as cancer and chronic diseases, still faces challenges. The difficulty in estimating explainable gradual disease progression from observable patient test data is a key factor. To address this issue, we propose a new approach called the “deep state-space analysis framework.” This framework utilizes sequentially obtained electronic health records (EHRs) to estimate and visualize temporal changes in the latent states of patients related to disease progression. It enables the clustering of latent patient states according to the severity of disease progression and identifies key factors leading to a poor prognosis with medication. To validate our framework, a detailed analysis of data from 12,695 patients with cancer was conducted. The estimated transitions of the latent states capture the clinical status of the patients and their continuous temporal changes. Furthermore, anemia was identified as a poor prognostic factor during state transitions in patients with cancer. Significant features were also confirmed, such as immune cell abnormalities, which are poor prognostic factors in patients treated with Nivolumab, Osimertinib, and Afatinib. This technological innovation deepens our understanding of disease progression and supports early treatment adjustments, prognostic evaluations, and the formulation of optimal long-term strategies. With the advancements in deep learning, its application in healthcare has even greater potential.

To examine whether data sourced from electronic medical records (EMR) and a large industrial spectacle lens manufacturing database can estimate refractive error distribution within large populations as an alternative to typical population surveys of refractive error. A total of 555,528 patient visits from 28 Irish primary care optometry practices between the years 1980 and 2019 and 141,547,436 spectacle lens sales records from an international European lens manufacturer between the years 1998 and 2016. Anonymized EMR data included demographic, refractive and visual acuity values. Anonymized spectacle lens data included refractive data. Spectacle lens data was separated into lenses containing an addition (ADD) and those without an addition (SV). The proportions of refractive errors from the EMR data and ADD lenses were compared to published results from the European Eye Epidemiology (E3) Consortium and the Gutenberg Health Study (GHS). Age and gender matched proportions of refractive error were comparable in the E3 data and the EMR data, with no significant difference in the overall refractive error distribution (χ2 = 527, p = 0.29, DoF = 510). EMR data provided a closer match to the E3 refractive error distribution by age than the ADD lens data. The ADD lens data, however, provided a closer approximation to the E3 data for total myopia prevalence than the GHS data, up to age 64. The prevalence of refractive error within a population can be estimated using EMR data in the absence of population surveys. Industry derived sales data can also provide insights on the epidemiology of refractive errors in a population over certain age ranges. EMR and industrial data may therefore provide a fast and cost-effective surrogate measure of refractive error distribution that can be used for future health service planning purposes.

Recent advancements in deep learning have significantly impacted ophthalmology, especially in glaucoma, a leading cause of irreversible blindness worldwide. In this study, we developed a reliable predictive model for glaucoma detection using deep learning models based on clinical data, social and behavior risk factor, and demographic data from 1652 participants, split evenly between 826 control subjects and 826 glaucoma patients. We extracted structural data from control and glaucoma patients' electronic health records (EHR). Three distinct machine learning classifiers, the Random Forest and Gradient Boosting algorithms, as well as the Sequential model from the Keras library of TensorFlow, were employed to conduct predictive analyses across our dataset. Key performance metrics such as accuracy, F1 score, precision, recall, and the area under the receiver operating characteristics curve (AUC) were computed to both train and optimize these models. The Random Forest model achieved an accuracy of 67.5%, with a ROC AUC of 0.67, outperforming the Gradient Boosting and Sequential models, which registered accuracies of 66.3% and 64.5%, respectively. Our results highlighted key predictive factors such as intraocular pressure, family history, and body mass index, substantiating their roles in glaucoma risk assessment. This study demonstrates the potential of utilizing readily available clinical, lifestyle, and demographic data from EHRs for glaucoma detection through deep learning models. While our model, using EHR data alone, has a lower accuracy compared to those incorporating imaging data, it still offers a promising avenue for early glaucoma risk assessment in primary care settings. The observed disparities in model performance and feature significance show the importance of tailoring detection strategies to individual patient characteristics, potentially leading to more effective and personalized glaucoma screening and intervention.

Posttraumatic stress disorder (PTSD) has significant morbidity and economic costs. This study describes the prevalence and characteristics of patients with PTSD using primary care electronic medical record (EMR) data. This retrospective cross-sectional study used EMR data from the Canadian Primary Care Sentinel Surveillance Network (CPCSSN). This study included 1,574 primary care providers located in 7 Canadian provinces. There were 689,301 patients that visited a CPCSSN provider between 1 January 2017 and 31 December 2019. We describe associations between PTSD and patient characteristics using descriptive statistics, chi-square, and multiple logistic regression models. Among the 689,301 patients included, 8,817 (1.3%, 95% CI 1.2-1.3) had a diagnosis of PTSD. On multiple logistic regression analysis, patients with depression (OR 4.4, 95% CI 4.2-4.7, P < 0.001), alcohol abuse/dependence (OR 1.7, 95% CI 1.6-1.9, P < 0.001), and/or drug abuse/dependence (OR 2.6, 95% CI 2.5-2.8, P < 0.001) had significantly higher odds of PTSD compared with patients without those conditions. Patients residing in community areas considered the most material deprived (OR 2.1, 95% CI 1.5-2.1, P < 0.001) or the most socially deprived (OR 2.8, 95% CI 2.7-5.3, P < 0.001) had higher odds of being diagnosed with PTSD compared with patients in the least deprived areas. The prevalence of PTSD in Canadian primary care is 1.3% (95% CI 1.25-1.31). Using EMR records we confirmed the co-occurrence of PTSD with other mental health conditions within primary care settings suggesting benefit for improved screening and evidence-based resources to manage PTSD.

In an ongoing effort to share heterogeneous electronic medical record (EMR) data in an i2b2 instance between the University Hospitals M&amp;uuml;nster and Erlangen for joint cancer research projects, an ontology based system for the mapping of EMR data to a set of common data elements has been developed. The system translates the mappings into local SQL scripts, which are then used to extract, transform and load the facts data from each EMR into the i2b2 database. By using Semantic Web standards, it is the authors' goal to reuse the laboriously compiled &amp;ldquo;mapping knowledge&amp;rdquo; in future projects, such as a comprehensive cancer ontology or even a hospital-wide clinical ontology.

We present an ontology based approach to identify hypertensive patients who show non-adherence to prescribed medication. Using the Web ontology language (OWL), we have developed an ontology that includes patient prescription details, medication possession ratios (MPRs) and blood pressure measurements (together with other patient related information) that has been populated with production electronic medical record (EMR) data from a general medical practice in New Zealand. We have written queries using the semantic query-enhanced Web rule language (SQWRL) to query this ontology to determine patients who have lapsed medication while having a low MPR. We also discuss some practical issues related to patient recall based on EMR data, as well as the suitability of the proposed scheme.

Severe sepsis and septic shock are global issues with high mortality rates. Early recognition and intervention are essential to optimize patient outcomes. Automated detection using electronic medical record (EMR) data can assist this process. This review describes automated sepsis detection using EMR data. PubMed retrieved publications between January 1, 2005 and January 31, 2015. Thirteen studies met study criteria: described an automated detection approach with the potential to detect sepsis or sepsis-related deterioration in real or near-real time; focused on emergency department and hospitalized neonatal, pediatric, or adult patients; and provided performance measures or results indicating the impact of automated sepsis detection. Detection algorithms incorporated systemic inflammatory response and organ dysfunction criteria. Systems in nine studies generated study or care team alerts. Care team alerts did not consistently lead to earlier interventions. Earlier interventions did not consistently translate to improved patient outcomes. Performance measures were inconsistent. Automated sepsis detection is potentially a means to enable early sepsis-related therapy but current performance variability highlights the need for further research.