Laboratory Data Timeliness and Completeness Improves Following Implementation of an Electronic Laboratory Information System in Côte d'Ivoire: Quasi-Experimental Study on 21 Clinical Laboratories From 2014 to 2020.

BACKGROUND: The Ministry of Health in Côte d'Ivoire and the International Training and Education Center for Health at the University of Washington, funded by the United States President's Emergency Plan for AIDS Relief, have been collaborating to develop and implement the Open-Source Enterprise-Level Laboratory Information System (OpenELIS). The system is designed to improve HIV-related laboratory data management and strengthen quality management and capacity at clinical laboratories across the nation. OBJECTIVE: This evaluation aimed to quantify the effects of implementing OpenELIS on data quality for laboratory tests related to HIV care and treatment. METHODS: This evaluation used a quasi-experimental design to perform an interrupted time-series analysis to estimate the changes in the level and slope of 3 data quality indicators (timeliness, completeness, and validity) after OpenELIS implementation. We collected paper and electronic records on clusters of differentiation 4 (CD4) testing for 48 weeks before OpenELIS adoption until 72 weeks after. Data collection took place at 21 laboratories in 13 health regions that started using OpenELIS between 2014 and 2020. We analyzed the data at the laboratory level. We estimated odds ratios (ORs) by comparing the observed outcomes with modeled counterfactual ones when the laboratories did not adopt OpenELIS. RESULTS: There was an immediate 5-fold increase in timeliness (OR 5.27, 95% CI 4.33-6.41; P<.001) and an immediate 3.6-fold increase in completeness (OR 3.59, 95% CI 2.40-5.37; P<.001). These immediate improvements were observed starting after OpenELIS installation and then maintained until 72 weeks after OpenELIS adoption. The weekly improvement in the postimplementation trend of completeness was significant (OR 1.03, 95% CI 1.02-1.05; P<.001). The improvement in validity was not statistically significant (OR 1.34, 95% CI 0.69-2.60; P=.38), but validity did not fall below pre-OpenELIS levels. CONCLUSIONS: These results demonstrate the value of electronic laboratory information systems in improving laboratory data quality and supporting evidence-based decision-making in health care. These findings highlight the importance of OpenELIS in Côte d'Ivoire and the potential for adoption in other low- and middle-income countries with similar health systems.

BBDash: An Electron-based tool for analyzing blood product utilization.

Blood transfusions can be associated with side effects ranging from occasional febrile reactions to extremely rare fatal reactions. Monitoring blood product orders and ensuring appropriate utilization is therefore an important strategy to ensure patient safety. However, data extracted from laboratory information systems can be difficult to interpret. We created BBDash, an Electron-based tool that reads Sunquest reports to create easy-to-interpret graphs related to blood product utilization.

Data-Exchange Between Electronic Medical Record and Viral-Load Laboratory Database Towards Improving HIV Care in Ethiopia.

Electronic viral load (VL) Test Ordering and Result Reporting System (ETORRS) was introduced to create data exchange between the existing VL database and the electronic medical record (EMR) system, with the aim of reducing laboratory test results turnaround time (TAT), improving data quality, and supporting timely clinical response for patients with high VL. This use case is an illustrative example of initiating and adopting the principles of health information exchange for a priority health program.

Drug supply measures and drug overdose mortality in the era of fentanyl and stimulants.

Background: Illicitly-manufactured fentanyl and stimulants have replaced prescription opioids as the primary contributors to fatal overdoses in the United States (US), yet the street supply of these substances is challenging to quantify. Building on the foundation of prior research on law enforcement drug reports, the present study compares publicly available forensic laboratory drug report measures to identify which measures account for the most variation in drug overdose mortality between states, within states over time, and in various demographic groups. Methods: Drug reports from the National Forensic Laboratory Information System and drug overdose mortality rates from the Centers for Disease Control and Prevention were examined for all US states and the District of Columbia, 2013-2021 (459 state-years). State- and year- fixed effects models regressed drug overdose mortality rates (in the overall population and subpopulations by sex, age, and race/ethnicity) on various drug report measures, including rates per population and proportional shares of drug reports positive for fentanyl/fentanyl-related compounds, heroin, cocaine, methamphetamine, and xylazine. Results: For drug overdose death rates in the overall population and nearly all subpopulations examined by sex, race/ethnicity, and age, the model including all drug report proportional measures represented the best-performing model (as identified via the lowest Akaike Information Criterion and highest within R-squared value), followed by the model including only the fentanyl/fentanyl-related compounds proportion. Conclusions: Findings support the utility of publicly available drug report composition measures, particularly the proportion of fentanyl/fentanyl-related compounds, as predictors of drug overdose mortality in the US and in various subpopulations.

Establishment of age- and -gender specific reference intervals for amino acids and acylcarnitines by tandem mass spectrometry in Turkish paediatric population.

Introduction: We determined age- and gender-specific reference intervals (RIs) for acylcarnitines and amino acids by tandem mass spectrometry (MS/MS) in the Turkish paediatric population by using laboratory information system (LIS) data. Materials and methods: A total of 9156 MS/MS results of children between 0-18 years of age, were downloaded from the LIS. Premature infants and newborns followed in the intensive care unit were excluded and only the first result of each patient attending outpatient clinics was included. Children with a known or suspected diagnosis of metabolic disease, malignancy, epilepsy, mental retardation, or genetic disorder were excluded. Laboratory results were evaluated and children with any pathological laboratory finding were excluded, resulting in a final sample size of 3357 (2029 boys and 1328 girls). Blood was collected by capillary puncture and spotted on Whatman 903 filter paper cards and analysed by MS/MS (Shimadzu LCMS-8050, Shimadzu Corporation, Kyoto, Japan). Data were evaluated for age and gender differences and age partitioning was performed according to the literature and visual evaluation of the data. Age subgroups were: ≤ 1 month, 2 months-1 year, 2-5 years, 6-10 years, and 11-18 years. Results: There were significant age-related differences for the majority of amino acids and acylcarnitines thus age dependent RIs were established. Gender-specific RIs were established for tyrosine, leucine-isoleucine, isovalerylcarnitine (C5) and hexadecanoylcarnitine (C16). Conclusions: Establishing age-related RIs can enhance the quality of medical care by facilitating early diagnosis and therapy, especially in certain metabolic disorders presenting with mild biochemical abnormalities and subtle clinical manifestations.

Understanding research methods, limitations, and applications of drug data collected by the National Forensic Laboratory Information System (NFLIS-Drug).

The National Forensic Laboratory Information System (NFLIS) is a drug surveillance program of the US Drug Enforcement Administration that systematically collects data on drugs that are seized by law enforcement and submitted to and analyzed by the Nation's forensic laboratories (NFLIS-Drug). NFLIS-Drug data are increasingly used in predictive modeling and drug surveillance to examine drug availability patterns. Given the complexity of the data and data collection, there are some common methodological pitfalls that we highlight with the aim of helping researchers avoid these concerns. The analysis done for this Technical Note is based on a review of the scientific literature that includes 428 unique, refereed article citations in 182 distinct journals published between January 1, 2005, and April 30, 2021. Each article was analyzed according to how NFLIS-Drug data were mentioned and whether NFLIS-Drug data were included. A sample of 37 articles was studied in-depth, and data issues were summarized. Using examples from the literature, this Technical Note highlights eight broad concerns that have important implications for the proper applications, interpretations, and limitations of NFLIS-Drug data with suggestions for improving research methods and accurate reporting of forensic drug data. NFLIS-Drug data are timely and provide key information to inform drug use trends across the United States; however, our present analysis shows that NFLIS-Drug data are misunderstood and represented in the literature. In addition to highlighting these issues, DEA has created several resources to assist NFLIS data users and researchers, which are summarized in the discussion.

Auto-validation of Routine Coagulation/Hemostasis Assays with Reflex Testing of Abnormal Test Results.

Hemostasis laboratories play a crucial role in the diagnosis and treatment of individuals with bleeding or thrombotic disorders. Routine coagulation assays, including the prothrombin time (PT)/international normalized ratio (INR), and activated partial thromboplastin time (APTT), are used for various purposes. These include as a screen of hemostasis function/dysfunction (e.g., possible factor deficiency) and for monitoring of anticoagulant therapy, such as vitamin K antagonists (PT/INR) and unfractionated heparin (APTT). Clinical laboratories are also under increasing pressure to improve services, especially response (test turnaround) times. There is also a need for laboratories to try to reduce error rates and for laboratory networks to standardize/harmonize processes and policies. Accordingly, we describe our experience with the development and implementation of automated processes for reflex testing and validation of routine coagulation test results. This has been implemented in a large pathology network compromising 27 laboratories and is under consideration for expansion to our larger network (of 60 laboratories). These rules have been custom-built within our laboratory information system (LIS), perform reflex testing of abnormal results, and fully automate the process of routine test validation for appropriate results. These rules also permit adherence to standardized pre-analytical (sample integrity) checks, automate reflex decisions, automate verification, and provide an overall alignment of network practices in a large network of 27 laboratories. In addition, the rules enable clinically significant results to be quickly referred to hematopathologists for review. We also documented an improvement in test turnaround times, with savings in operator time and thus operating costs. Finally, the process was generally well received and determined to be beneficial for most laboratories in our network, in part identified by improved test turnaround times.

Improving documentation of blood product administration using a standardized electronic health record-based system: a single-institution experience.

OBJECTIVES: To improve documentation of blood product administration by assessing the completion status of blood transfusions. In this way, we can ensure compliance with the Association for the Advancement of Blood & Biotherapies standards and facilitate investigation of potential blood transfusion reactions. METHODS: This before-and-after study includes the implementation of an electronic health record (EHR)-based, standardized protocol for documenting the completion of blood product administration. Twenty-four months of retrospective data (January-December 2021) and prospective data (January-December 2022) were collected. Meetings were held before the intervention. Ongoing daily, weekly, and monthly reports were prepared, and targeted education to deficient areas as well as spot in-person audits by the blood bank residents were conducted. RESULTS: During 2022, 8,342 blood products were transfused, of which 6,358 blood product administrations were documented. The overall percentage of completed transfusion order documentation improved from 35.54% (units/units) in 2021 to 76.22% (units/units) in 2022. CONCLUSIONS: Interdisciplinary collaborative efforts helped produce quality audits to improve the documentation of blood product transfusion through a standardized and customized EHR-based blood product administration module.

Long-term follow-up with a clinical decision support system based on laboratory reports to manage patients with biochemical recurrence after radical prostatectomy.

INTRODUCTION: Currently, prostate cancer (PCa) is the second most common cause of cancer death, and radical prostatectomy (RP) remains the primary treatment for localized PCa. Although there is no consensus on an optimal strategy, the determination of total serum prostate-specific antigen (tPSA) is the cornerstone for the detection of postoperative biochemical recurrence (BCR). The aim of this study was to evaluate the prognostic utility of serial tPSA levels together with other clinicopathological factors and to assess the impact of a commentary algorithm implemented in our laboratory information system. METHODS: A descriptive and retrospective study of patients with clinically localized PCa who underwent RP. BCR-free survival was calculated over time (Kaplan-Meier analysis), and the ability of different clinicopathological factors to predict BCR was studied (univariate and multivariate analyses) with Cox models. RESULTS: A total of 203 patients underwent RP, of whom 51 presented with BCR during follow-up. In the multivariate model, doubling of tPSA, the Gleason score, tumour stage and tPSA nadir were detected as independent predictors of BCR. CONCLUSION: A patient with undetectable tPSA after 1959 days of RP is unlikely to develop BCR, regardless of preoperative or pathologic risk factors. Furthermore, doubling of tPSA in the first 2 years of follow-up was the main prognostic factor for BCR in patients undergoing RP. Other prognostic factors included a tPSA nadir detectable after surgery, a Gleason score ≥ 7 and a tumour stage T ≥ 2c.

A web application to support the coordination of reflexive, interpretative toxicology testing.

Background: Reflexive laboratory testing workflows can improve the assessment of patients receiving pain medications chronically, but complex workflows requiring pathologist input and interpretation may not be well-supported by traditional laboratory information systems. In this work, we describe the development of a web application that improves the efficiency of pathologists and laboratory staff in delivering actionable toxicology results. Method: Before designing the application, we set out to understand the entire workflow including the laboratory workflow and pathologist review. Additionally, we gathered requirements and specifications from stakeholders. Finally, to assess the performance of the implementation of the application, we surveyed stakeholders and documented the approximate amount of time that is required in each step of the workflow. Results: A web-based application was chosen for the ease of access for users. Relevant clinical data was routinely received and displayed in the application. The workflows in the laboratory and during the interpretation process served as the basis of the user interface. With the addition of auto-filing software, the return on investment was significant. The laboratory saved the equivalent of one full-time employee in time by automating file management and result entry. Discussion: Implementation of a purpose-built application to support reflex and interpretation workflows in a clinical pathology practice has led to a significant improvement in laboratory efficiency. Custom- and purpose-built applications can help reduce staff burnout, reduce transcription errors, and allow staff to focus on more critical issues around quality.

Quality improvement of outpatient clinical chemistry tests through a novel middleware-laboratory information system solution.

OBJECTIVES: Rapid and accurate laboratory tests are essential to support clinical decision-making. Despite the various efforts to control quality in the laboratory, our outpatient chemistry turnaround time (TAT) has deteriorated since 2018. Moreover, these difficulties have accelerated further due to the COVID-19 pandemic. Therefore, we aimed to improve laboratory work efficiency by identifying and eliminating the causes of reduced laboratory work efficiency. DESIGN & METHODS: We surveyed to identify tasks that reduce work efficiency. Based on our survey, a new-concept of work assistance middleware linked to laboratory information system (LIS) was developed. The middleware supports test end-time prediction, automatic real-time TAT monitoring, and urgent test requests so that medical technologists can focus on their chemistry tests. The developed middleware was used for 6 months in laboratory and outpatient clinics, and its effectiveness was evaluated. RESULTS: The median TAT for outpatient chemistry tests was reduced by 6.6 min, from 72.4 min to 65.8 min. And not only did the maximum TAT for the sample decrease from 353 min to 214 min, but the proportion of samples exceeding the TAT target (120 min) also decreased by 77%; from 2.00% in 2010 (1,905 out of 94,989 samples) to 0.46% in 2021 (453 out of 98,117 samples). 2,199 samples were urgently requested through middleware, and they were processed about 15% faster than other samples, effectively performing urgent tests. The test end-time prediction showed an error of 8.6 min in the evaluation using the MAE (Mean Absolute Error) index. CONCLUSIONS: Through this study, the quality and efficiency of the laboratory were improved, and while reducing the workload of medical staff, it contributed to enhancing patient safety and satisfaction.

One health system supporting surveillance during COVID-19 epidemic in Abruzzo region, southern Italy.

The Istituti Zooprofilattici Sperimentali (IZSs) are public health institutes dealing with the aetiology and pathogenesis of infectious diseases of domestic and wild animals. During Coronavirus Disease 2019 epidemic, the Italian Ministry of Health appointed the IZSs to carry out diagnostic tests for the detection of SARS-CoV-2 in human samples. In particular, the IZS of Abruzzo and Molise (IZS-Teramo) was involved in the diagnosis of SARS-CoV-2 through testing nasopharyngeal swabs by Real Time RT-PCR. Activities and infrastructures were reorganised to the new priorities, in a "One Health" framework, based on interdisciplinary, laboratory promptness, accreditation of the test for the detection of the RNA of SARS-CoV-2 in human samples, and management of confidentiality of sensitive data. The laboratory information system - SILAB - was implemented with a One Health module for managing data of human origin, with tools for the automatic registration of information improving the quality of the data. Moreover, the "National Reference Centre for Whole Genome Sequencing of microbial pathogens - database and bioinformatics analysis" - GENPAT - formally established at the IZS-Teramo, developed bioinformatics workflows and IT dashboard with ad hoc surveillance tools to support the metagenomics-based SARS-CoV-2 surveillance, providing molecular sequencing analysis to quickly intercept the variants circulating in the area. This manuscript describes the One Health system developed by adapting and integrating both SILAB and GENPAT tools for supporting surveillance during COVID-19 epidemic in the Abruzzo region, southern Italy. The developed dashboard permits the health authorities to observe the SARS-CoV-2 spread in the region, and by combining spatio-temporal information with metagenomics provides early evidence for the identification of emerging space-time clusters of variants at the municipality level. The implementation of the One Health module was designed to be easily modelled and adapted for the management of other diseases and future hypothetical events of pandemic nature.

Natural Language Processing to extract SNOMED-CT codes from pathological reports.

Objective: The use of standardized structured reports (SSR) and suitable terminologies like SNOMED-CT can enhance data retrieval and analysis, fostering large-scale studies and collaboration. However, the still large prevalence of narrative reports in our laboratories warrants alternative and automated labeling approaches. In this project, natural language processing (NLP) methods were used to associate SNOMED-CT codes to structured and unstructured reports from an Italian Digital Pathology Department. Methods: Two NLP-based automatic coding systems (support vector machine, SVM, and long-short term memory, LSTM) were trained and applied to a series of narrative reports. Results: The 1163 cases were tested with both algorithms, showing good performances in terms of accuracy, precision, recall, and F1 score, with SVM showing slightly better performances as compared to LSTM (0.84, 0.87, 0.83, 0.82 vs 0.83, 0.85, 0.83, 0.82, respectively). The integration of an explainability allowed identification of terms and groups of words of importance, enabling fine-tuning, balancing semantic meaning and model performance. Conclusions: AI tools allow the automatic SNOMED-CT labeling of the pathology archives, providing a retrospective fix to the large lack of organization of narrative reports.

Machine learning models for predicting one-year survival in patients with metastatic gastric cancer who experienced upfront radical gastrectomy.

Tumor metastasis is a common event in patients with gastric cancer (GC) who previously underwent curative gastrectomy. It is meaningful to employ high-volume clinical data for predicting the survival of metastatic GC patients. We aim to establish an improved machine learning (ML) classifier for predicting if a patient with metastatic GC would die within 12 months. Eligible patients were enrolled from a Chinese GC cohort, and the complete detailed information from medical records was extracted to generate a high-dimensional dataset. Appropriate feature engineering and feature filter were conducted before modeling with eight algorithms. A 10-fold cross validation (CV) nested in a holdout CV (8:2) was employed for hyperparameter tuning and model evaluation. Model selection was based on the area under the receiver operating characteristic (AUROC) curve, recall, and precision. The selected model was globally explained using interpretable surrogate models. Of the total 399 cases (median survival of 8.2 months), 242 patients survived less than 12 months. The linear discriminant analysis (LDA), support vector machine (SVM), and random forest (RF) model had the highest AUROC (0.78 ± 0.021), recall (0.93 ± 0.031), and precision (0.80 ± 0.026), respectively. The LDA model created a new function that generally separated the two classes. The predicted probability of the SVM model was interpreted using a linear regression model visualized by a nomogram. The predicted class of the RF model was explained using a decision tree model. In summary, analyzing high-volume medical data by ML is helpful to produce an improved model for predicting the survival in patients with metastatic GC. The algorithm should be carefully selected in different practical scenarios.

Impact of the Implementation of Laboratory Information System (WWDISA) on Timely Provision of HIV-1 Viral Load Results in a Rural Area, Inhambane, Mozambique.

WWDISA is an optional module of the DISA Laboratory Information system (LIS) that offers a web portal that allows access to test results over the internet for patient clinical management. This study aims to assess the applicability of using the WWDISA web application, and the lessons learned from its implementation in six health facilities in Mabote district, Inhambane province. Data from 2463 and 665 samples for HIV-viral load (HIVVL) tests, extracted from paper-based and WWDISA systems, respectively, were included, from January to December 2020. Data were simultaneously collected on a quarterly basis from both systems to allow comparison. The WWDISA turnaround time (TAT) from sample collection to results becoming available was found to be 10 (IQR: 8−12) days and significantly lower than the health unit manual logbook (p value < 0.001). Regarding the system efficiency, it was found that among 1978 search results, only 642 (32.5%) were found, and the main challenges according to the users were lack of connectivity (77%) and the website going down (62%). The WWDISA module has been shown to be effective in reducing the TAT, although a stable internet connection and accurate data entry are essential to make the system functional.

Lightweight, open source, easy-use algorithm and web service for paraprotein screening using spatial frequency domain analysis of electrophoresis studies.

Introduction: Serum protein electrophoresis (SPEP) is commonly used to detect monoclonal paraproteins to meet laboratory diagnostic criteria for plasma cell neoplasms. We propose an automated screening method for paraprotein detection that uses minimal computational resources for training and deployment. Methods: A model screening for paraproteins based on the presence of high-frequency components in the spatial frequency spectrum of the SPEP densitometry curve was calibrated on a set of 330 samples, and evaluated on representative (n=110) and external (n=1,321) test sets. The model takes as input a patient's serum densitometry curve and a standardized control curve and outputs a prediction of whether a paraprotein is present. We built an interactive web application allowing users to easily perform paraprotein screening given inputs for densitometry curves, as well as a macro-enabled spreadsheet for easy automated screening. Results: When tuned to maximize likelihood ratio with minimum sensitivity 0.90, the model achieved AUC 0.90, sensitivity 0.90, positive-predictive value 0.64, specificity 0.55, and accuracy 0.72 in the representative test set. In the external test set, the model achieved AUC 0.90, sensitivity 0.97, positive-predictive value 0.42, specificity 0.29, and accuracy 0.52. A subset analysis showed sensitivities of 0.90, 0.96, and 1.0 in detecting low (0.1-0.5 g/dL), medium (0.5-3.0 g/dL), and high paraprotein levels (≥3.0 g/dL), respectively. We have released a web service allowing users to score their own SPEP data, and also released the algorithm and application programming interface in an open-source package allowing users to customize the model to their needs. Conclusions: We developed a proof of concept for an automated method for paraprotein screening using only the characteristics of the SPEP curve. Future work should focus on testing the method with other laboratory data including immunofixation gels, as well as incorporation of outside data sources including clinical data.

Anatomic pathology quality assurance: Developing an LIS based tracking and documentation module for intradepartmental consultations.

Background: An electronic intradepartmental consultation system for anatomic pathology (AP) was conceived and developed in the laboratory information system (LIS) in 2019. Previously, all surgical pathology intradepartmental consultative activities were initiated and documented with paper forms which circulated with the pertinent microscopic slides and were eventually filed. In this study, we discuss the implementation and utilization of an electronic intradepartmental AP consultation system. Methods: Workflows and procedures were developed to organize intradepartmental surgical pathology consultations from the beginning to the end point of the consultative activities entirely using a paperless system that resided in the LIS. Results: The electronic consult system allowed electronic documentation of all steps of intradepartmental consultative activities. The system provided tracking ability for consulted cases and improved access to consult discussion for all departmental personnel, staff, and trainees. Consultation work queue was created for each pathologist and a summary of individual consultative workload was possible. Documentation of anatomic pathology quality assurance for intradepartmental consultative activity was easily assessed. Conclusions: The electronic intradepartmental consult system has allowed our department to electronically track intradepartmental consult cases, store the consultative opinion text with the case, record the pathologists involved, and document the consultation for internal quality assurance review as well as for accrediting organizations. Summarization of pathologist workload related to consultative activity was quantifiable and optimization of the consultative process was maximized for education in an academic setting.

A survey on Artificial Intelligence and Big Data utilisation in Italian clinical laboratories.

OBJECTIVES: The Italian Society of Clinical Biochemistry and Clinical Molecular Biology (SIBioC) Big Data and Artificial Intelligence (BAI) Working Group promoted a survey to frame the knowledge, skills and technological predisposition in clinical laboratories. METHODS: A questionnaire, focussing on digitization, information technology (IT) infrastructures, data accessibility, and BAI projects underway was sent to 1,351 SIBioC participants. The responses were evaluated using SurveyMonkey software and Google Sheets. RESULTS: The 227 respondents (17%) from all over Italy (47% of 484 labs), mainly biologists, laboratory physicians and managers, mostly from laboratories of public hospitals, revealed lack of hardware, software and corporate Wi-Fi, and dearth of PCs. Only 25% work daily on clouds, while 65%-including Laboratory Directors-cannot acquire health data from sources other than laboratories. Only 50% of those with access can review a clinical patient's health record, while the other access only to laboratory information. The integration of laboratory data with other health data is mostly incomplete, which limits BAI-type analysis. Many are unaware of integration platforms. Over 90% report pulling data from the Laboratory Information System, with varying degrees of autonomy. Very few have already undertaken BAI projects, frequently relying on IT partnerships. The majority consider BAI as crucial in helping professional judgements, indicating a growing interest. CONCLUSIONS: The questionnaire received relevant feedback from SIBioC participants. It highlighted the level of expertise and interest in BAI applications. None of the obstacles stands out more than the others, emphasising the need to all-around work: IT infrastructures, data warehouses, BAI analysis software acquisition, data accessibility and training.

Implementing Epic Beaker Laboratory Information System for Diagnostics in Anatomic Pathology.

Medicine is expeditiously evolving, and the number of diagnostic opportunities has increased exponentially in the last decade. Electronic medical records (EMRs) have been welcomed in most institutions worldwide following an early period of suspicious behavior. Unfortunately, several cracks dictated the initial approach to hospital systems and leadership incompetency. However, the pathway for a successful decade of EMRs is paved. This narrative review illustrates some principles implementing Epic Beaker software for anatomic pathology in academic medical institutions. Implementing such software improves the diagnostic approach in the division of anatomic pathology because the pathologists can directly access an enormous amount of clinical and radiological information now at their front desk using extremely versatile windows.

Neural Network Assisted Pathology Case Identification.

BACKGROUND: Traditionally, cases for cohort selection and quality assurance purposes are identified through structured query language (SQL) searches matching specific keywords. Recently, several neural network-based natural language processing (NLP) pipelines have emerged as an accurate alternative/complementary method for case retrieval. METHODS: The diagnosis section of 1000 pathology reports with the terms "colon" and "carcinoma" were retrieved from our laboratory information system through a SQL query. Each of the reports were labeled as either positive or negative, where cases are considered positive if the case was a primary adenocarcinoma of the colon. Negative cases comprised adenocarcinoma from other sites, metastatic adenocarcinomas, benign conditions, rectal cancers, and other cases that do not fit in the primary colonic adenocarcinoma category. The 1000 cases were randomly separated into training, validation, and holdout sets. A convolutional neural network (CNN) model built using Keras (a neural network library) was trained to identify positive cases, and the model was applied to the holdout set to predict the category for each case. RESULTS: The CNN model classified 141 out of 149 primary colonic adenocarcinoma cases, and 43 out of 51 negative cases correctly, achieving an accuracy of 92% and area under the ROC curve (AUC) of 0.957. CONCLUSION: Trained convolutional neural network models by itself, or as an adjunct to keyword and pattern-based text extraction methods may be used to search for pathology cases of interest with high accuracy.