AcademyHealth

Creation and curation of knowledge graphs at scale can be used to exponentially accelerate the discovery, matching, and analysis of diseases in real-world data. While disease ontologies are useful for annotation, integration, and analysis of biological data, codified disease and procedure categories e.g. SNOMED-CT, ICD10, CPT, etc. rarely capture all of the nuances in a patient condition or, in the case of rare disease, may not even exist. Furthermore, there are multiple disease definitions used in data sources and publications, each having its own structure and hierarchy. Mapping between ontologies, finding disease clusters, and building a representation of the chosen disease area are resource-intensive, often requiring significant human capital. We propose the creation and curation of a patient knowledge graph utilizing large language model extraction techniques. In order to expand in volume and scale, knowledge graphs with generalized language capability allow for data to be extracted using natural language rather than being constrained by the exact terminology or hierarchy of existing ontologies. We develop a method of mapping back to existing ontologies such as MeSH, SNOMED-CT, RxNORM, HPO, etc. to ground the extracted entities to known entities in the medical community. We have access to one of the largest ambulatory care EHR databases in the country. To demonstrate the effectiveness of our method, we benchmark our extraction in a test set with over 33.6M unique patients, in the area of patient search. In this case study, we perform a patient search for a rare disease: Dravet syndrome. Dravet syndrome was codified as an ICD10 recognizable disease in October 2020. In the following research, we describe our method of the construction of patient-specific knowledge graphs and subsequent searches for patients who exhibit symptoms of a particular disease. Using patients with confirmed ICD10 codes for Dravet syndrome as our ground truth, we utilize our LLM-based entity extraction techniques and formalize an algorithmic way of characterizing patients in a grounded ontology to assist in mapping patients to specific diseases. Finally, we present the results of a real-world discovery method on Beta-propeller protein-associated neurodegeneration (BPAN), identifying patients with a rare disease, where no ground truth currently exists.

Health organizations and systems rely on increasingly sophisticated informatics infrastructure. Without anti-racist expertise, the field risks reifying and entrenching racism in information systems. We consider ways the informatics field can recognize institutional, systemic, and structural racism and propose the use of the Public Health Critical Race Praxis (PHCRP) to mitigate and dismantle racism in digital forms. We enumerate guiding questions for stakeholders along with a PHCRP-Informatics framework. By focusing on (1) critical self-reflection, (2) following the expertise of well-established scholars of racism, (3) centering the voices of affected individuals and communities, and (4) critically evaluating practice resulting from informatics systems, stakeholders can work to minimize the impacts of racism. Informatics, informed and guided by this proposed framework, will help realize the vision of health systems that are more fair, just, and equitable.

The purpose of this study was to measure sexually transmitted infection (STI) testing among Medicaid enrollees initiating preexposure prophylaxis (PrEP) to prevent human immunodeficiency virus. Secondary data are in the form of Medicaid enrollment and claims data in six states in the US South. Research partnerships in six states in the US South developed a distributed research network to accomplish study aims. Each state identified all first-time PrEP users in fiscal year 2017-2018 (combined N = 990) and measured the presence of STI testing for chlamydia, syphilis, and gonorrhea through 2019. Each state calculated the percentage of individuals with at least one STI test during 3-, 6-, and 12-month follow-up periods. The proportion of first-time PrEP users that received an STI test varied by state: 37% to 67% of all of the individuals in each state who initiated PrEP received a test within the first 6 months of PrEP treatment and 50% to 77% received a test within the first 12 months. Although the Centers for Disease Control and Prevention recommends STI testing at least every 6 months for PrEP users, our analysis of Medicaid data suggests that STI testing occurs less frequently than recommended in populations at elevated risk of syphilis, gonorrhea, and chlamydia.

This Viewpoint discusses the urgent need to build trust between physicians and the organizations in which they work.

WHO Director-General Tedros Adan Ghebreyesus announced on May 21, 2021, at the 74th World Health Assembly, that at least 115,000 health workers had died during the novel coronavirus pandemic. Analysis of literary sources showed that significant losses are borne by the doctors of the “Red Zones” of hospitals: doctors, nurses, as well as doctors, paramedics, ambulance drivers. The most likely cause of such losses is considered by health experts from different countries: an increased load on staff, which leads to early “Professional burnout” and “Emotional fatigue”, a higher probability of infection with the virus. This does not take into account other, probable causes, for example, distant interactions or the “Kaznacheev effect”.

Clinical decision support (CDS) is a process for enhancing health-related decisions and actions with pertinent, organized clinical knowledge and patient information that can significantly improve health outcomesand healthcare delivery. However, their impact on clinical outcomes has been inconsistent. Rigorous and continuous evaluation of CDS is necessary for improving CDS. An interface prototype designed based on User and Task analysis wasevaluated to measure its usability and effectiveness in evaluating CDSeffectiveness for improving quality outcomes based on the analysis. The results show that task of evaluating CDS effectiveness is relatively hard, and moreworkmay be needed to guide usersreach correct conclusions.

The Veterans Health Administration piloted a nationwide Lean Enterprise Transformation program to optimize delivery of services to patients for high value care. Barriers and facilitators to Lean implementation were evaluated. Guided by the Lean Enterprise Transformation evaluation model, 268 interviews were conducted, with stakeholders across 10 Veterans Health Administration medical centers. Interview transcripts were analyzed using thematic analysis techniques. Supporting the utility of the model, facilitators and barriers to Lean implementation were found in each of the Lean Enterprise Transformation evaluation model domains: (a) impetus to transform, (b) leadership commitment to quality, (c) improvement initiatives, (d) alignment across the organization, (e) integration across internal boundaries, (f) communication, (g) capability development, (h) informed decision making, (i) patient engagement, and (j) organization culture. In addition, three emergent themes were identified: staff engagement, sufficient staffing, and use of Lean experts (senseis). Effective implementation required staff engagement, strategic planning, proper scoping and pacing, deliberate coaching, and accountability structures. Visible, stable leadership drove Lean when leaders articulated a clear impetus to change, aligned goals within the facility, and supported middle management. Reliable data and metrics provided support for and evidence of successful change. Strategic early planning with continual reassessment translated into focused and sustained Lean implementation. Prominent best practices identified include (a) reward participants by broadcasting Lean successes; (b) provide time and resources for participation in Lean activities; (c) avoid overscoping projects; (d) select metrics that closely align with improvement processes; and (e) invest in coaches, informal champions, process improvement staff, and senior leadership to promote staff engagement and minimize turnover.

As states have embraced additional flexibility to change coverage of and payment for Medicaid services, they have also faced heightened expectations for delivering high-value care. Efforts to meet these new expectations have increased the need for rigorous, evidence-based policy, but states may face challenges finding the resources, capacity, and expertise to meet this need. By describing state-university partnerships in more than 20 states, this commentary describes innovative solutions for states that want to leverage their own data, build their analytic capacity, and create evidence-based policy. From an integrated web-based system to improve long-term care to evaluating the impact of permanent supportive housing placements on Medicaid utilization and spending, these state partnerships provide significant support to their state Medicaid programs. In 2017, these partnerships came together to create a distributed research network that supports multi-state analyses. The Medicaid Outcomes Distributed Research Network (MODRN) uses a common data model to examine Medicaid data across states, thereby increasing the analytic rigor of policy evaluations in Medicaid, and contributing to the development of a fully functioning Medicaid innovation laboratory.

This Viewpoint discusses the importance of trust in partnerships between health care and community-based organizations for achieving better health outcomes and proposes principles and strategies for building and nurturing trustworthy collaborations.