Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Monday, March 11, 2019 at 4:00 PM to 4:30 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?


Journal Description

JMIR Medical Informatics (JMI, ISSN 2291-9694) (Editor-in-chief: Christian Lovis MD MPH FACMI) is a Pubmed/SCIE-indexed, top-rated, tier A journal with impact factor expected in 2019, which focuses on clinical informatics, big data in health and health care, decision support for health professionals, electronic health records, ehealth infrastructures and implementation. It has a focus on applied, translational research, with a broad readership including clinicians, CIOs, engineers, industry and health informatics professionals.

Published by JMIR Publications, publisher of the Journal of Medical Internet Research (JMIR), the leading eHealth/mHealth journal (Impact Factor 2018: 4.671), JMIR Med Inform has a slightly different scope (emphasizing more on applications for clinicians and health professionals rather than consumers/citizens, which is the focus of JMIR), publishes even faster, and also allows papers which are more technical or more formative than what would be published in the Journal of Medical Internet Research.

JMIR Medical Informatics is indexed in PubMed Central/PubMed, and has also been accepted for SCIE, with an official Clarivate impact factor 2018 expected to be released in 2019 (see announcement).

JMIR Medical Informatics adheres to the same quality standards as JMIR and all articles published here are also cross-listed in the Table of Contents of JMIR, the worlds' leading medical journal in health sciences / health services research and health informatics (


Recent Articles:

  • Source: Image created by the Authors; Copyright: The Authors; URL:; License: Creative Commons Attribution (CC-BY).

    A Computer Application to Predict Adverse Events in the Short-Term Evolution of Patients With Exacerbation of Chronic Obstructive Pulmonary Disease


    Background: Chronic obstructive pulmonary disease (COPD) is a common chronic disease. Exacerbations of COPD (eCOPD) contribute to the worsening of the disease and the patient’s evolution. There are some clinical prediction rules that may help to stratify patients with eCOPD by their risk of poor evolution or adverse events. The translation of these clinical prediction rules into computer applications would allow their implementation in clinical practice. Objective: The goal of this study was to create a computer application to predict various outcomes related to adverse events of short-term evolution in eCOPD patients attending an emergency department (ED) based on valid and reliable clinical prediction rules. Methods: A computer application, Prediction of Evolution of patients with eCOPD (PrEveCOPD), was created to predict 2 outcomes related to adverse events: (1) mortality during hospital admission or within a week after an ED visit and (2) admission to an intensive care unit (ICU) or an intermediate respiratory care unit (IRCU) during the eCOPD episode. The algorithms included in the computer tool were based on clinical prediction rules previously developed and validated within the Investigación en Resultados y Servicios de Salud COPD study. The app was developed for Windows and Android systems, using Visual Studio 2008 and Eclipse, respectively. Results: The PrEveCOPD computer application implements the prediction models previously developed and validated for 2 relevant adverse events in the short-term evolution of patients with eCOPD. The application runs under Windows and Android systems and it can be used locally or remotely as a Web application. Full description of the clinical prediction rules as well as the original references is included on the screen. Input of the predictive variables is controlled for out-of-range and missing values. Language can be switched between English and Spanish. The application is available for downloading and installing on a computer, as a mobile app, or to be used remotely via internet. Conclusions: The PrEveCOPD app shows how clinical prediction rules can be summarized into simple and easy to use tools, which allow for the estimation of the risk of short-term mortality and ICU or IRCU admission for patients with eCOPD. The app can be used on any computer device, including mobile phones or tablets, and it can guide the clinicians to a valid stratification of patients attending the ED with eCOPD. Trial Registration: NCT00102401; (Archived by WebCite at International Registered Report Identifier (IRRID): RR2-10.1186/1472-6963-11-322

  • Source: Pexels; Copyright: Christina Morillo; URL:; License: Licensed by JMIR.

    Patient-Sharing Relations in the Treatment of Diabetes and Their Implications for Health Information Exchange: Claims-Based Analysis


    Background: Health information exchange (HIE) among care providers who cooperate in the treatment of patients with diabetes mellitus (DM) has been rated as an important aspect of successful care. Patient-sharing relations among care providers permit inferences about corresponding information-sharing relations. Objectives: This study aimed to obtain information for an effective HIE platform design to be used in DM care by analyzing patient-sharing relations among various types of care providers (ToCPs), such as hospitals, pharmacies, and different outpatient specialists, within a nationwide claims dataset of Austrian DM patients. We focus on 2 parameters derived from patient-sharing networks: (1) the principal HIE partners of the different ToCPs involved in the treatment of DM and (2) the required participation rate of ToCPs in HIE platforms for the purpose of effective communication. Methods: The claims data of 7.9 million Austrian patients from 2006 to 2007 served as our data source. DM patients were identified by their medication. We established metrics for the quantification of our 2 parameters of interest. The principal HIE partners were derived from the portions of a care provider’s patient-sharing relations with different ToCPs. For the required participation rate of ToCPs in an HIE platform, we determine the concentration of patient-sharing relations among ToCPs. Our corresponding metrics are derived in analogy from existing work for the quantification of the continuity of care. Results: We identified 324,703 DM patients treated by 12,226 care providers; the latter were members of 16 ToCPs. On the basis of their score for 2 of our parameters, we categorized the ToCPs into low, medium, and high. For the most important HIE partner parameter, pharmacies, general practitioners (GPs), and laboratories were the representatives of the top group, that is, our care providers shared the highest numbers of DM patients with these ToCPs. For the required participation rate of type of care provide (ToCP) in HIE platform parameter, the concentration of DM patient-sharing relations with a ToCP tended to be inversely related to the ToCPs member count. Conclusions: We conclude that GPs, pharmacies, and laboratories should be core members of any HIE platform that supports DM care, as they are the most important DM patient-sharing partners. We further conclude that, for implementing HIE with ToCPs who have many members (in Austria, particularly GPs and pharmacies), an HIE solution with high participation rates from these ToCPs (ideally a nationwide HIE platform with obligatory participation of the concerned ToCPs) seems essential. This will raise the probability of HIE being achieved with any care provider of these ToCPs. As chronic diseases are rising because of aging societies, we believe that our quantification of HIE requirements in the treatment of DM can provide valuable insights for many industrial countries.

  • Source: Wikimedia Commons; Copyright: Theihb; URL:; License: Creative Commons Attribution + ShareAlike (CC-BY-SA).

    Structure and Content of Drug Monitoring Advices Included in Discharge Letters at Interfaces of Care: Exploratory Analysis Preceding Database Development


    Background: Inadequate drug monitoring of drug therapy after hospital discharge facilitates adverse drug events and preventable hospital readmissions. Objective: This study aimed to analyze the structure and content of drug monitoring advices of a representative sample of discharge letters as a basis for future electronic information systems. Methods: On 2 days in November 2016, all discharge letters of 3 departments of a university hospital were extracted from the hospital information system. The frequency, content, and structure of drug monitoring advices in discharge letters were investigated and compared with the theoretical monitoring requirements expressed in the corresponding summaries of product characteristics (SmPC). The quality of the drug monitoring advices in the discharge letters was rated with the domains of an adapted systematic instructions for monitoring (SIM) score. Results: In total, 154 discharge letters were analyzed containing 1180 brands (240 active pharmaceutical substances), of which 50.42% (595/1180) could theoretically be amended with a monitoring advice according to the SmPC. In reality, 40 discharge letters (26.0%, 40/154) contained a total of 66 monitoring advices for 57 brands (4.83%, 57/1180), comprising 18 different monitoring parameters. Drug monitoring advices only addressed mean 1.9 (SD 0.8) of the 7 domains of the SIM score and frequently did not address reasons for monitoring (86%, 57/66), the timing of monitoring, that is, the start (76%, 50/66), the frequency (94%, 63/66), the stop (95%, 63/66), and how to react (83%, 55/66). Conclusions: Drug monitoring advices were mostly absent in discharge letters and a gold standard for appropriate drug monitoring advices was lacking. Hence, more effort should be put in the development of tools that facilitate easy presentation of clinically meaningful drug monitoring advices at the point of care.

  • Source: Image created by the Authors; Copyright: The Authors; URL:; License: Creative Commons Attribution (CC-BY).

    Medication Adherence Prediction Through Online Social Forums: A Case Study of Fibromyalgia


    Background: Medication nonadherence can compound into severe medical problems for patients. Identifying patients who are likely to become nonadherent may help reduce these problems. Data-driven machine learning models can predict medication adherence by using selected indicators from patients’ past health records. Sources of data for these models traditionally fall under two main categories: (1) proprietary data from insurance claims, pharmacy prescriptions, or electronic medical records and (2) survey data collected from representative groups of patients. Models developed using these data sources often are limited because they are proprietary, subject to high cost, have limited scalability, or lack timely accessibility. These limitations suggest that social health forums might be an alternate source of data for adherence prediction. Indeed, these data are accessible, affordable, timely, and available at scale. However, they can be inaccurate. Objective: This paper proposes a medication adherence machine learning model for fibromyalgia therapies that can mitigate the inaccuracy of social health forum data. Methods: Transfer learning is a machine learning technique that allows knowledge acquired from one dataset to be transferred to another dataset. In this study, predictive adherence models for the target disease were first developed by using accurate but limited survey data. These models were then used to predict medication adherence from health social forum data. Random forest, an ensemble machine learning technique, was used to develop the predictive models. This transfer learning methodology is demonstrated in this study by examining data from the Medical Expenditure Panel Survey and the PatientsLikeMe social health forum. Results: When the models are carefully designed, less than a 5% difference in accuracy is observed between the Medical Expenditure Panel Survey and the PatientsLikeMe medication adherence predictions for fibromyalgia treatments. This design must take into consideration the mapping between the predictors and the outcomes in the two datasets. Conclusions: This study exemplifies the potential and limitations of transfer learning in medication adherence–predictive models based on survey data and social health forum data. The proposed approach can make timely medication adherence monitoring cost-effective and widely accessible. Additional investigation is needed to improve the robustness of the approach and extend its applicability to other therapies and other sources of data.

  • Source: Image created by the Authors; Copyright: The Authors; URL:; License: Creative Commons Attribution (CC-BY).

    Physician Use of Electronic Health Records: Survey Study Assessing Factors Associated With Provider Reported Satisfaction and Perceived Patient Impact


    Background: The effect electronic health record (EHR) implementation has on physician satisfaction and patient care remains unclear. A better understanding of physician perceptions of EHRs and factors that influence those perceptions is needed to improve the physician and patient experience when using EHRs. Objective: The objective of this study was to determine provider and clinical practice factors associated with physician EHR satisfaction and perception of patient impact. Methods: We surveyed a random sample of physicians, including residents and fellows, at a US quaternary care academic hospital from February to March 2016. The survey assessed provider demographics, clinical practice factors (ie, attending, fellow, or resident), and overall EHR experience. The primary outcomes assessed were provider satisfaction and provider perceptions of impact to patient care. Responses on the satisfaction and patient impact questions were recorded on a continuous scale initially anchored at neutral (scale range 0 to 100: 0 defined as “extremely negatively” and 100 as “extremely positively”). Independent variables assessed included demographic and clinical practice factors, including perceived efficiency in using the EHR. One-way analysis of variance or the Kruskal-Wallis test was used for bivariate comparisons, and linear regression was used for multivariable modeling. Results: Of 157 physicians, 111 (70.7%) completed the survey; 51.4% (57/111) of the respondents were attending physicians, and of those, 71.9% (41/57) reported a >50% clinical full-time-equivalency and half reported supervising residents >50% of the time. A total of 50.5% (56/111) of the respondents were primary care practitioners, previous EHR experience was evenly distributed, and 12.6% (14/111) of the total sample were EHR super-users. Responses to how our current EHR affects satisfaction were rated above the neutral survey anchor point (mean 58 [SD 22]), as were their perceptions as to how the EHR impacts the patient (mean 61 [SD 18]). In bivariate comparisons, only physician age, clinical role (resident, fellow, or attending), and perceived efficiency were associated with EHR satisfaction. In the linear regression models, physicians with higher reported perceived efficiency reported higher overall satisfaction and patient impact after controlling for other variables in the model. Conclusions: Physician satisfaction with EHRs and their perception of its impact on clinical care were generally positive, but physician characteristics, greater age, and attending level were associated with worse EHR satisfaction. Perceived efficiency is the factor most associated with physician satisfaction with EHRs when controlling for other factors. Understanding physician perceptions of EHRs may allow targeting of technology resources to ensure efficiency and satisfaction with EHR system use during clinical care.

  • Source: Flickr; Copyright: ILO in Asia and the Pacific; URL:; License: Creative Commons Attribution + Noncommercial + NoDerivatives (CC-BY-NC-ND).

    Establishment of a New Platform for the Management of Patients After Cardiac Surgery: Descriptive Study


    Background: Medical care for the Chinese population has been focused on first-line treatment, but with little follow-up on treated patients. As an important part of clinical work, follow-up evaluations are of great significance for the long-term survival of patients and for clinical and scientific research. However, the overall follow-up rate of discharged patients after surgery has been low for many years because of the limitations of certain follow-up methods and the presence of objective, practical problems. Objective: This study aimed to construct a new two-way interactive telemedicine follow-up platform to improve the collection of clinical data after cardiac surgery and provide reliable and high-quality follow-up services. Methods: Computer and network technologies were employed in the context of “Internet +” to develop follow-up databases and software compatible with a mobile network. Postoperative follow-up quality data including the follow-up rate and important postoperative indices were used as standards to evaluate the new follow-up management model after cardiac surgery. Results: This system has been officially operated for more than 5 years. A total of 5347 patients undergoing cardiac surgery have been enrolled, and the total follow-up rate was 90.22%. In addition, 6349 echocardiographic images, 4717 electrocardiographic images, and 3504 chest radiographic images have been uploaded during follow-up assessments. The international standardized ratio was 20,696 person-times. Conclusions: This new management follow-up platform can be used to effectively collect clinical data, provide technical support for academic research, extend medical services, and provide more help to patients. It is of great significance for managing patients after cardiac surgery.

  • Source: Flickr; Copyright: Asian Development Bank; URL:; License: Creative Commons Attribution + Noncommercial (CC-BY-NC).

    An Assessment of the Interoperability of Electronic Health Record Exchanges Among Hospitals and Clinics in Taiwan


    Background: The rapid aging of the Taiwanese population in recent years has led to high medical needs for the elderly and increasing medical costs. Integrating patient information through electronic health records (EHRs) to reduce unnecessary medications and tests and enhance the quality of care has currently become an important issue. Although electronic data interchanges among hospitals and clinics have been implemented for many years in Taiwan, the interoperability of EHRs has not adequately been assessed. Objective: The study aimed to analyze the efficiency of data exchanges and provide suggestions for future improvements. Methods: We obtained 30 months of uploaded and downloaded data of EHRs among hospitals and clinics. The research objects of this study comprised 19 medical centers, 57 regional hospitals, 95 district hospitals, and 5520 clinics. We examined 4 exchange EHR forms: laboratory test reports, medical images, discharge summaries, and outpatient medical records. We used MySQL (Oracle Corporation) software (to save our data) and phpMyAdmin, which is a Personal Home Page program, to manage the database and then analyzed the data using SPSS 19.0 statistical software. Results: The quarterly mean uploaded volume of EHRs among hospitals was 52,790,721 (SD 580,643). The quarterly mean downloaded volume of EHRs among hospitals and clinics was 650,323 (SD 215,099). The ratio of uploaded to downloaded EHRs was about 81:1. The total volume of EHRs was mainly downloaded by medical centers and clinics, which accounted for 53.82% (mean 318,717.80) and 45.41% (mean 269,082.10), respectively, and the statistical test was significant among different hospital accreditation levels (F2=7.63; P<.001). A comparison of EHR download volumes among the 6 National Health Insurance (NHI) branches showed that the central NHI branch downloaded 11,366,431 records (21.53%), which was the highest, and the eastern branch downloaded 1,615,391 records (3.06%), which was the lowest. The statistical test among the 6 NHI branches was significant (F5=8.82; P<.001). The download volumes of laboratory tests reports and outpatient medical records were 26,980,425 (50.3%) and 21,747,588 records (40.9%), respectively, and were much higher than medical images and discharge summaries. The statistical test was also significant (F=17.72; P<.001). Finally, the download time showed that the average for x-rays was 32.05 seconds, which was the longest, and was 9.92 seconds for electrocardiogram, which was the shortest, but there was no statistically significant difference among download times for various medical images. Conclusions: After years of operation, the Electronic Medical Record Exchange Center has achieved the initial goal of EHR interoperability, and data exchanges are running quite stably in Taiwan. However, the meaningful use of EHRs among hospitals and clinics still needs further encouragement and promotion. We suggest that the government’s leading role and collective collaboration with health care organizations are important for providing effective health information exchanges.

  • Source: Pexels; Copyright:; URL:; License: Licensed by the authors.

    Extraction of Geriatric Syndromes From Electronic Health Record Clinical Notes: Assessment of Statistical Natural Language Processing Methods


    Background: Geriatric syndromes in older adults are associated with adverse outcomes. However, despite being reported in clinical notes, these syndromes are often poorly captured by diagnostic codes in the structured fields of electronic health records (EHRs) or administrative records. Objective: We aim to automatically determine if a patient has any geriatric syndromes by mining the free text of associated EHR clinical notes. We assessed which statistical natural language processing (NLP) techniques are most effective. Methods: We applied conditional random fields (CRFs), a widely used machine learning algorithm, to identify each of 10 geriatric syndrome constructs in a clinical note. We assessed three sets of features and attributes for CRF operations: a base set, enhanced token, and contextual features. We trained the CRF on 3901 manually annotated notes from 85 patients, tuned the CRF on a validation set of 50 patients, and evaluated it on 50 held-out test patients. These notes were from a group of US Medicare patients over 65 years of age enrolled in a Medicare Advantage Health Maintenance Organization and cared for by a large group practice in Massachusetts. Results: A final feature set was formed through comprehensive feature ablation experiments. The final CRF model performed well at patient-level determination (macroaverage F1=0.834, microaverage F1=0.851); however, performance varied by construct. For example, at phrase-partial evaluation, the CRF model worked well on constructs such as absence of fecal control (F1=0.857) and vision impairment (F1=0.798) but poorly on malnutrition (F1=0.155), weight loss (F1=0.394), and severe urinary control issues (F1=0.532). Errors were primarily due to previously unobserved words (ie, out-of-vocabulary) and a lack of context. Conclusions: This study shows that statistical NLP can be used to identify geriatric syndromes from EHR-extracted clinical notes. This creates new opportunities to identify patients with geriatric syndromes and study their health outcomes.

  • Source: Image created by the Authors; Copyright: The Authors; URL:; License: Creative Commons Attribution (CC-BY).

    Transcription Errors of Blood Glucose Values and Insulin Errors in an Intensive Care Unit: Secondary Data Analysis Toward Electronic Medical...


    Background: Critically ill patients require constant point-of-care blood glucose testing to guide insulin-related decisions. Transcribing these values from glucometers into a paper log and the electronic medical record is very common yet error-prone in intensive care units, given the lack of connectivity between glucometers and the electronic medical record in many US hospitals. Objective: We examined (1) transcription errors of glucometer blood glucose values documented in the paper log and in the electronic medical record vital signs flow sheet in a surgical trauma intensive care unit, (2) insulin errors resulting from transcription errors, (3) lack of documenting these values in the paper log and the electronic medical record vital signs flow sheet, and (4) average time for docking the glucometer. Methods: This secondary data analysis examined 5049 point-of-care blood glucose tests. We obtained values of blood glucose tests from bidirectional interface software that transfers the meters’ data to the electronic medical record, the paper log, and the vital signs flow sheet. We obtained patient demographic and clinical-related information from the electronic medical record. Results: Of the 5049 blood glucose tests, which were pertinent to 234 patients, the total numbers of undocumented or untranscribed tests were 608 (12.04%) in the paper log, 2064 (40.88%) in the flow sheet, and 239 (4.73%) in both. The numbers of transcription errors for the documented tests were 98 (2.21% of 4441 documented tests) in the paper log, 242 (8.11% of 2985 tests) in the flow sheet, and 43 (1.64% of 2616 tests) in both. The numbers of transcription errors per patient were 0.4 (98 errors/234 patients) in the paper log, 1 (242 errors/234 patients) in the flow sheet, and 0.2 in both (43 errors/234 patients). Transcription errors in the paper log, the flow sheet, and in both resulted in 8, 24, and 2 insulin errors, respectively. As a consequence, patients were given a lower or higher insulin dose than the dose they should have received had there been no errors. Discrepancies in insulin doses were 2 to 8 U lower doses in paper log transcription errors, 10 U lower to 3 U higher doses in flow sheet transcription errors, and 2 U lower in transcription errors in both. Overall, 30 unique insulin errors affected 25 of 234 patients (10.7%). The average time from point-of-care testing to meter docking was 8 hours (median 5.5 hours), with some taking 56 hours (2.3 days) to be uploaded. Conclusions: Given the high dependence on glucometers for point-of-care blood glucose testing in intensive care units, full electronic medical record-glucometer interoperability is required for complete, accurate, and timely documentation of blood glucose values and elimination of transcription errors and the subsequent insulin-related errors in intensive care units.

  • Source: iStock by Getty Images; Copyright: South_agency; URL:; License: Licensed by the authors.

    European Hospitals’ Transition Toward Fully Electronic-Based Systems: Do Information Technology Security and Privacy Practices Follow?


    Background: Traditionally, health information has been mainly kept in paper-based records. This has deeply changed throughout approximately the last three decades with the widespread use of multiple health information technologies. The digitization of health care systems contributes to improving health care delivery. However, it also exposes health records to security and privacy breaches inherently related to information technology (IT). Thus, health care organizations willing to leverage IT for improved health care delivery need to put in place IT security and privacy measures consistent with their use of IT resources. Objective: In this study, 2 main objectives are pursued: (1) to assess the state of the implementation of IT security and privacy practices in European hospitals and (2) to assess to what extent these hospitals enhance their IT security and privacy practices as they move from paper-based systems toward fully electronic-based systems. Methods: Drawing on data from the European Commission electronic health survey, we performed a cluster analysis based on IT security and privacy practices implemented in 1723 European hospitals. We also developed an IT security index, a compounded measure of implemented IT security and privacy practices, and compared it with the hospitals’ level in their transition from a paper-based system toward a fully electronic-based system. Results: A total of 3 clearly distinct patterns of health IT–related security and privacy practices were unveiled. These patterns, as well as the IT security index, indicate that most of the sampled hospitals (70.2%) failed to implement basic security and privacy measures consistent with their digitization level. Conclusions: Even though, on average, the most electronically advanced hospitals display a higher IT security index than hospitals where the paper system still dominates, surprisingly, it appears that the enhancement of IT security and privacy practices as the health information digitization advances in European hospitals is neither systematic nor strong enough regarding the IT-security requirements. This study will contribute to raising awareness among hospitals’ managers as to the importance of enhancing their IT security and privacy measures so that they can keep up with the security threats inherently related to the digitization of health care organizations.

  • Source: Flickr; Copyright: National Center for Advancing Translational Sciences; URL:; License: Creative Commons Attribution (CC-BY).

    OpenEHR and General Data Protection Regulation: Evaluation of Principles and Requirements


    Background: Concerns about privacy and personal data protection resulted in reforms of the existing legislation in the European Union (EU). The General Data Protection Regulation (GDPR) aims to reform the existing directive on the topic of personal data protection of EU citizens with a strong emphasis on more control of the citizens over their data and in the establishment of rules for the processing of personal data. OpenEHR is a standard that embodies many principles of interoperable and secure software for electronic health records (EHRs) and has been advocated as the best approach for the development of hospital information systems. Objective: This study aimed to understand to what extent the openEHR standard can help in the compliance of EHR systems to the GDPR requirements. Methods: A list of requirements for an EHR to support GDPR compliance and also a list of the openEHR design principles were made. The requirements were categorized and compared with the principles by experts on openEHR and GDPR. Results: A total of 50 GDPR requirements and 8 openEHR design principles were identified. The openEHR principles conformed to 30% (15/50) of GDPR requirements. All the openEHR principles were aligned with GDPR requirements. Conclusions: This study showed that the openEHR principles conform well to GDPR, underlining the common wisdom that truly realizing security and privacy requires it to be built in from the start. By using an openEHR-based EHR, the institutions are closer to becoming compliant with GDPR while safeguarding the medical data.

  • A rehabilitated child at the Children's Hospital of Shanghai. Source: Image created by the Authors; Copyright: Children's Hospital of Shanghai; URL:; License: Creative Commons Attribution (CC-BY).

    Medication Use for Childhood Pneumonia at a Children’s Hospital in Shanghai, China: Analysis of Pattern Mining Algorithms


    Background: Pattern mining utilizes multiple algorithms to explore objective and sometimes unexpected patterns in real-world data. This technique could be applied to electronic medical record data mining; however, it first requires a careful clinical assessment and validation. Objective: The aim of this study was to examine the use of pattern mining techniques on a large clinical dataset to detect treatment and medication use patterns for childhood pneumonia. Methods: We applied 3 pattern mining algorithms to 680,138 medication administration records from 30,512 childhood inpatients with diagnosis of pneumonia during a 6-year period at a children’s hospital in China. Patients’ ages ranged from 0 to 17 years, where 37.53% (11,453/30,512) were 0 to 3 months old, 86.55% (26,408/30,512) were under 5 years, 60.37% (18,419/30,512) were male, and 60.10% (18,338/30,512) had a hospital stay of 9 to 15 days. We used the FP-Growth, PrefixSpan, and USpan pattern mining algorithms. The first 2 are more traditional methods of pattern mining and mine a complete set of frequent medication use patterns. PrefixSpan also incorporates an administration sequence. The newer USpan method considers medication utility, defined by the dose, frequency, and timing of use of the 652 individual medications in the dataset. Together, these 3 methods identified the top 10 patterns from 6 age groups, forming a total of 180 distinct medication combinations. These medications encompassed the top 40 (73.66%, 500,982/680,138) most frequently used medications. These patterns were then evaluated by subject matter experts to summarize 5 medication use and 2 treatment patterns. Results: We identified 5 medication use patterns: (1) antiasthmatics and expectorants and corticosteroids, (2) antibiotics and (antiasthmatics or expectorants or corticosteroids), (3) third-generation cephalosporin antibiotics with (or followed by) traditional antibiotics, (4) antibiotics and (medications for enteritis or skin diseases), and (5) (antiasthmatics or expectorants or corticosteroids) and (medications for enteritis or skin diseases). We also identified 2 frequent treatment patterns: (1) 42.89% (291,701/680,138) of specific medication administration records were of intravenous therapy with antibiotics, diluents, and nutritional supplements and (2) 11.53% (78,390/680,138) were of various combinations of inhalation of antiasthmatics, expectorants, or corticosteroids. Fleiss kappa for the subject experts’ evaluation was 0.693, indicating moderate agreement. Conclusions: Utilizing a pattern mining approach, we summarized 5 medication use patterns and 2 treatment patterns. These warrant further investigation.

Citing this Article

Right click to copy or hit: ctrl+c (cmd+c on mac)

Latest Submissions Open for Peer-Review:

View All Open Peer Review Articles
  • Automated outlier Identification for Validation (AIV) of Stroke Outcomes Using Multiple Large Repositories

    Date Submitted: Apr 5, 2019

    Open Peer Review Period: Apr 8, 2019 - Jun 3, 2019

    Background: Introduction: Researchers commonly use the Modified Rankin Scale (mRS) and the Barthel Index (BI) to assess a patient’s clinical outcome after stroke. These are potential targets in mach...

    Background: Introduction: Researchers commonly use the Modified Rankin Scale (mRS) and the Barthel Index (BI) to assess a patient’s clinical outcome after stroke. These are potential targets in machine learning models for stroke outcome prediction. Objective: The objective of this study was to evaluate density-based outlier detection methods on mRS and BI assessments. Methods: We trained three density-based outlier detection methods including density-based spatial clustering of applications (DBSCAN), hierarchical DBSCAN (HDBSCAN) and local outlier factor (LOF) based on data obtained from a nationwide prospective stroke registry in Taiwan. The testing of each method was done by using four different NINDS stroke datasets. Results: The DBSCAN achieved a high performance across all mRS values (the highest average accuracy was 99.2±0.7 at mRS-4 and the lowest average accuracy was 92.0±4.6 at mRS-3). The LOF also achieved similar performance but the HDBSCAN needs further improvement. Conclusions: The evaluation results showed that promising density-based outlier detection methods are promising for validation of stroke outcome measures during the study data clean up processes to improve completion of clinical data for increased data quality and research analyses.

  • Common Data Elements of patients with acute coronary syndrome (ACS)

    Date Submitted: Mar 22, 2019

    Open Peer Review Period: Mar 26, 2019 - May 21, 2019

    Background: Standardization in clinical documentation can increase efficiency, save time and resources. Objective: The objectives of this work are to compare documentation forms for acute coronary syn...

    Background: Standardization in clinical documentation can increase efficiency, save time and resources. Objective: The objectives of this work are to compare documentation forms for acute coronary syndrome (ACS), check for standardization and generate a list of the most common data elements using semantic form annotation with Unified Medical Language System (UMLS). Methods: 86 sources from registries, studies, risk scores, quality assurance, official guidelines and routine documentation of four hospitals in Germany were semantically annotated. This allowed for automatic comparison of concept frequencies and generation of a list of most common concepts. 3.710 forms items were semantically annotated using 842 unique UMLS concepts. Results: Half of all medical concept occurrences are covered by 60 unique concepts, which suggests the existence of a core data set of relevant concepts. Overlap percentages between forms are relatively low, hinting at inconsistent documentation structures and lack of standardization. Conclusions: This analysis shows a lack of standardized and semantically enriched documentation for patients with ACS. Efforts made by official institutions like the ESC have not yet been fully implemented. Utilizing a standardized and annotated core data set of the most important data concepts could make export and automatic re-use of data easier. The generated list of common data elements is an exemplary implementation suggestion of the concepts to use in a standardized approach.