8:30 a.m. – 10:00 a.m.
S21: Systems Demonstrations – Tools for CRI
K. Wagholikar, Harvard Medical School, Partners Healthcare; V. Vernekar, A. Zagade, Y. Ostrovsky, Persistent Systems; S. Chan, A. Goodson, Partners Healthcare; A. Pathak, Persistent Systems; C. Glynn, C. Herrick, Partners Healthcare; S. Murphy, Partners Healthcare, Harvard Medical School
Although phenotyping methods have been used to characterize cohorts for epidemiological studies, these methods are largely unused for driving clinical care processes. The objective of this project is to stratify patient populations for targeted clinical care through use of advanced phenotyping techniques. The approach is to integrate libraries for machine learning (ML) with workflows for creating clinical annotations to train ML. The outcome is to identify patients for selected therapeutic pathways, which will facilitate efficiency and scalability of care delivery to the patient population. We have developed and implemented a patient stratification process and applied it in the domain of cardiovascular medicine at Partner Healthcare. Moreover, we are working on implement the stratification process by extending i2b2 into a patient stratification platform (i2b2-PSP). Specifically, we have developed extensions to implement key steps the patient stratification process including creation of ontologies to model and compute concepts needed at point-of-care, and a feedback process to allow the clinical staff to provide inputs for improving the phenotyping algorithms. In this session, we will demonstrate the process for a real world use case but using simulated patient data, and demonstrate the developed i2b2 plugins.
A. Tanchoco, O. Strachna, S. Gardos, J. Wills, S. Gardos, J. Lengfellner, A. Zarski, I. Sher, M. Piscitelli, P. Stetson, Memorial Sloan Kettering Cancer Center
Given the growing interest for translational research in the clinical community, it’s imperative to integrate disparate sources of patient data that are captured within the electronic health record system (EHR) and augmented by data captured during the research phase. In this system demonstration we provide an overview of a simple human friendly mechanism for integrating REDCap data with other data sources by developing a custom Tableau Data Web Connector (WDC). We go over the key technical system design, data architecture and security features of this new system. We will demonstrate the multitude of use cases and the ease of use by combining with the new Tableau Ask Data feature available on version 2019.1.
J. Philip, R. Belenkaya, P. Stetson, Memorial Sloan Kettering
As in many places, Memorial Sloan Kettering Cancer Center has a need to provide researchers high quality data extracted from the clinical notes. Despite the advances in natural language processing and other types of machine learning, many researchers at MSK manually abstract data from clinical notes and store them in spreadsheets. This presents many problems since spreadsheets are often lost. Data entered in those files often using no standard terminology, making it very hard to share and reuse them. MSK EXTRACT is a three-tiered approach to luring researchers away from spreadsheets into REDCap. The first tier is a custom web application that uses the “shopping cart” motif to help people create their own REDCap databases. The shopping cart presents a list of data elements and permissible values mapped to various ontologies. Users search for, select and add these standard terms to their project. This functionality enables REDcap projects built with this application to use standard terminology and ultimately allows data reuse and facilitate data sharing.
1:45 p.m. – 3:15 p.m.
S42: Systems Demonstrations – FHIR®
R. Bradshaw, K. Kawamoto, G. Del Fiol, University of Utah
Through a research grant funded by the Informatics Technology for Cancer Research (ITCR) program of the US National Cancer Institute, the University of Utah Health system has implemented a population health management platform using a FHIR and CDS Hooks-based clinical decision support (CDS) system to identify and manage patient populations at risk for familial breast, ovarian, or colorectal cancer. The system is currently being piloted and is providing valuable insights for full scale deployment. In this demonstration, the leaders of this effort will describe in detail how the system is being used to manage patients, how the system has been architected, including the standards used, lessons learned so far, and future plans to deploy to multiple sites.
C. Hsiao, AHRQ; J. Blumenthal, MedStar Health National Center for Human Factors in Healthcare; D. Wesley, MedStar Health Research Institute
The effective use of patient-reported outcomes (PROs) can play a critical role in improving health care delivery and patient experience with care. Some electronic health records (EHRs) are able to capture structured PRO data, but this information is not commonly collected at the point of care. In addition, standards have not previously existed to guide and inform the collection and integration of PRO data into health information technology systems, thereby limiting the ability to easily share these data across providers. This demonstration will highlight a SMART on FHIR® application that enables the collection and sharing of standardized PRO data across health systems. The audience will observe how the application is used to collect PROMIS® measures, connects to a secure FHIR® endpoint, and sends these data for storage and subsequent retrieval. The demonstration will also include a customized data visualization of the PROMIS® scores and survey responses that healthcare providers would see in the EHR. In addition, the demonstration will show the “data on demand” feature that expedites site-specific implementation by eliminating the need to create data feeds through proprietary interface engineering. The architecture of the PRO data collection and visualization ecosystem features a decentralized, scalable, and replicable model.
D. Vreeman, Indiana University, Regenstrief Institute, Inc; T. Briscoe, J. Hook, S. Wagers, Regenstrief Institute, Inc.; J. Agnew, Smile CDR; S. Abhyankar, Regenstrief Institute, Inc.
LOINC® is a freely available international standard for identifying measurements, observations, and documents that has become ubiquitous in health data systems worldwide. Previously, LOINC was distributed in custom table formats and with LOINC-specific software. We have deployed a standardized API for LOINC that implements FHIR terminology services specifications, including the CodeSystem, ValueSet, and ConceptMap resources. In September 2018, we deployed a publicly available server instance powered by the HAPI open source FHIR library on Amazon Web Services. Since its launch, this service has processed an average of more than 350,000 monthly requests with an average processing time of 163ms per request. We updated the terminology content with the December 2018 and June 2019 LOINC releases, and plan to do so with each upcoming LOINC release. By providing programmatic access to current LOINC release content in a uniform and well documented format using open source software we hope to accelerate the use of LOINC in diverse biomedical informatics applications.
10:30 a.m. – 12:00 p.m.
S73: System Demonstrations – CDS Systems
Authoring and Integrating Interoperable Clinical Decision Support: CDS Connect Open Source Tools
The Agency for Healthcare Research and Quality (AHRQ) launched CDS Connect to advance evidence into clinical practice through shareable, standards-based, and publicly available clinical decision support (CDS). The CDS authoring tool provides “non-technical” individuals with the ability to author interoperable CDS logic using HL7 standards (i.e., Clinical Quality Language [CQL]), facilitating interoperability and implementation among different health IT systems. The tool integrates with the National Library of Medicine’s Value Set Authority Center to leverage value sets comprised of standard terminology codes that define clinical concepts in the logic. Furthermore, the tool enables users to select which FHIR version is used in the CDS expression (i.e., DSTU2 or STU3). Output from the authoring tool can be used with the CQL Services prototype tool to integrate the code with a health IT system using the CDS Hooks standard. Attendees of this system demonstration will learn how the CDS Connect platform can be used to author and test CDS logic and make it publicly-available; how to integrate the CDS logic with a health IT system using CQL Services, leveraging the CDS Hooks standard; and how to provide feedback to the CDS Connect project.
K. Cyras, Imperial College London; J. Dominguez, King's College London; A. Karamlou, D. Prociuk, Imperial College London; V. Curcin, King's College London; B. Delaney, F. Toni, K. Chalkidou, A. Darzi, Imperial College London
We demonstrate an end-to-end argumentation-assisted decision support system (DSS) to provide patient-centric, non-conflicting clinical guideline recommendations in multimorbidity settings. Specifically, our system integrates an electronic health record (EHR) component to identify patient-tailored guideline recommendations represented in the computer-interpretable Transition-based Medical Recommendations (TMR) model, and an argumentation methodology (from Artificial Intelligence) to reason with conflicting patient-specific recommendations and various preferences. The system transparently yields individual clinical recommendations together with the underlying arguments considered, thus providing reasons for and against the suggested decisions. Following the Learning Health Systems (LHS) paradigm and aiming for our argumentation-assisted DSS to learn and adapt from the existing evidence, we intend for a design that allows our system to take into account the ‘expected’ outcomes of health policies and their changes.
C. Moesel, The MITRE Corporation; K. Valdes, b.well Connected Health, Inc.; G. Meadows, The MITRE Corporation; S. Al-Showk, E. Lomotan, Q. Ngo-Metzger, AHRQ; S. Sebastian, The MITRE Corporation
Clinical decision support (CDS) has traditionally focused on supporting health care providers at the point of care. However, most decision-making that impacts individual health, particularly overall health and wellness, occurs outside of the clinical encounter. Preventive care recommendations, such as those provided by the U.S. Preventive Services Task Force (USPSTF), may be amenable to CDS targeted directly to patients, especially if implemented through innovative platforms that combine comprehensive patient-specific data, user-friendly interfaces that consumers have come to expect, and methods to communicate back to health care providers so as to remain connected with the clinical team. The Agency for Healthcare Research and Quality and its partners, through a project called CDS Connect, developed interoperable, publicly-available CDS for preventive care based on USPSTF recommendations. Recommendations were transformed into HL7 Clinical Quality Language, used HL7 FHIR and HL7 CDS Hooks for data exchange, and were delivered to a patient-facing platform provided by b.well. To make the recommendations patient-specific, the CDS evaluated data from the patient’s electronic health record, insurance claims, and patient-generated sources. Consumers received reminders for preventive screenings tailored to their specific health history and could earn rewards by discussing screenings with their health care provider.
1:45 p.m. – 3:15 p.m.
S83: Systems Demonstrations – Clinical Evidence
K. Kawamoto, D. Martin, C. Nanjo, University of Utah
The U.S. healthcare system faces a crisis in care quality, with adults receiving only about half of recommended evidence-based care. At the same time, the U.S. healthcare system faces a crisis in clinician burnout, caused in no small part by frustration with the electronic health record (EHR). The University of Utah’s ReImagine EHR initiative has invested heavily in leveraging SMART on FHIR as a means of addressing these dual crises by improving patient care and enhancing the clinician experience with the EHR. As a culmination of this effort, the University of Utah has developed a SMART on FHIR platform for chronic disease management and health maintenance. Entering initial production deployment in an Epic EHR environment in Spring 2019, this platform provides a unified overview of the patient’s care needs related to common chronic diseases and health maintenance. Through seamless integration with the EHR, the platform pulls in and intuitively organizes relevant data; provides actionable, guideline-based recommendations; enables 1-click ordering of needed interventions; and facilitates required clinical documentation. In this session, the presenters will demonstrate how the Disease Management Platform makes it easy – and perhaps even enjoyable – to provide evidence-based care to every patient, every time.
Using trial2rev to Support Timely and Efficient Systematic Review Updates
Systematic reviews are time-consuming and often do not target the most relevant medical questions, leading to slow uptake of new trial results and many potentially redundant reviews. We developed the web-based platform trial2rev to connect published systematic reviews to emerging results in clinical trial registries, with the aim of supporting decisions around when a systematic review update should be undertaken. Major use cases include the proactive monitoring of emerging trial results to signal when a review update is warranted and using the database of systematic reviews linked to trial registrations to support evaluation of new methods for automatic screening of trials.
J. Dominguez, E. Fairweather, King's College London; B. Delaney, Imperial College; V. Curcin, King's College London
Randomized controlled trials are the most reliable approach to generate high quality medical evidence. However, protocol specifications and multiple configurations across clinical sites results in inefficient recruitment and execution of the trials. By using the latest advances in eSource and study design models, we aim to provide an integrated electronic clinical trial platform to support the workflow management of clinical studies across different environments found in clinical sites.
3:30 p.m. – 5:00 p.m.
S94: Systems Demonstrations – Disease Specific
A. Long, A. Glogowski, National Institute of Allergy and Infectious Diseases, Deloitte Consulting LLP
Tuberculosis is the leading cause of death due to infectious disease and the rate of mortality significantly increases in patients with drug-resistant tuberculosis (DR TB). The complex etiology of DR TB requires a multi-dimensional approach to analyses, spanning clinical, genomic, and radiological indicators. The National Institute of Allergy and Infectious Diseases (NIAID) developed the Tuberculosis Data Exploration Portal (TB DEPOT) to address the challenges that diagnosing and treating DR TB cases present. TB DEPOT fully supports the National Institute of Health’s (NIH) Findable, Accessible, Interoperable, and Reusable (FAIR) principles. TB DEPOT is a publicly available, web-based analytics tool that enables researchers and clinicians to create virtual cohorts of deidentified patient cases, save them, and conduct on-demand comparative statistical analyses to facilitate hypothesis generation and understanding of the disease. The solution leverages Amazon Web Services (AWS) for cloud-based infrastructure, .NET webforms, Highcharts, R, PLINK, custom-developed algorithms, and APIs. During this system demonstration, we will navigate the platform live and discuss the benefits and considerations of software innovation to support an increased understanding of complex diseases.
A. Sethi, Ciitizen Corporation
Reducing cancer-related morbidity and mortality requires greater access to data. But today this data is too often trapped in silos, strangled by regulatory and business concerns. Ciitizen believes patients can help solve cancer’s data problem. Leveraging the legal right of patients to access their health information under HIPAA and other privacy laws, Ciitizen is creating a cancer data ecosystem that is fully patient controlled (including with respect to de-identified data). Ciitizen’s platform (1) collects comprehensive health data from every place the patient has received care; (2) standardizes and normalizes that data into actionable information; and (3) enables patients to leverage their cancer information to pursue treatment options and to contribute to advancing cancer research.