Search

* Context.--Machine learning (ML) allows for the analysis of massive quantities of high-dimensional clinical laboratory data, thereby revealing complex patterns and trends. Thus, ML can potentially improve the efficiency of clinical data interpretation and the practice of laboratory medicine. However, the risks of generating biased or unrepresentative models, which can lead to misleading clinical conclusions or overestimation of the model performance, should be recognized. Objectives.--To discuss the major components for creating ML models, including data collection, data preprocessing, model development, and model evaluation. We also highlight many of the challenges and pitfalls in developing ML models, which could result in misleading clinical impressions or inaccurate model performance, and provide suggestions and guidance on how to circumvent these challenges. Data Sources.--The references for this review were identified through searches of the PubMed database, US Food and Drug Administration white papers and guidelines, conference abstracts, and online preprints. Conclusions.--With the growing interest in developing and implementing ML models in clinical practice, laboratorians and clinicians need to be educated in order to collect sufficiently large and high-quality data, properly report the data set characteristics, and combine data from multiple institutions with proper normalization. They will also need to assess the reasons for missing values, determine the inclusion or exclusion of outliers, and evaluate the completeness of a data set. In addition, they require the necessary knowledge to select a suitable ML model for a specific clinical question and accurately evaluate the performance of the ML model, based on objective criteria. Domain-specific knowledge is critical in the entire workflow of developing ML models. (Arch Pathol Lab Med. 2023;147:826-836; doi: 10.5858/ arpa.2021-0635-RA), Machine learning (ML) (1) has emerged as a powerful tool for analyzing and interpreting laboratory test results as well as integrating clinical findings with laboratory data. (2) In recent years, [...]

Introduction: Integrating behavioral health (BH) and primary care is an important strategy to improve health behaviors, mental health, and substance misuse, particularly at community health centers (CHCs) where disease burden is high and access to mental health services is low. Components of different integrated BH models are often combined in practice. It is unknown which components distinguish developing versus established integrated BH programs. Method: A survey was mailed to 128 CHCs in 10 Midwestern states in 2016. Generalized estimating equation models were used to assess associations between program characteristics and stage of integration implementation (precontemplation, contemplation, preparation, action, or maintenance). Content analysis of open-ended responses identified integration barriers. Results: Response rate was 60% (N = 77). Most CHCs had colocated BH and primary care services, warm hand-offs from primary care to BH clinicians, shared scheduling and electronic health record (EHR) systems, and depression and substance use disorder screening. Thirty-two CHCs (42%) indicated they had completed integration and were focused on quality improvement (maintenance). Being in the maintenance stage was associated with having a psychologist on staff (odds ratio [OR] = 7.16, 95% confidence interval [CI] [2.76, 18.55]), a system for tracking referrals (OR = 3.42, 95% CI [1.03, 11.36]), a registry (OR = 2.71, 95% CI [1.86, 3.94]), PCMH designation (OR = 2.82, 95% CI [1.48, 5.37]), and a lower proportion of Black/African American patients (OR = .82, 95% CI [.75, .89]). The most common barriers to integration were difficulty recruiting and retaining BH clinicians and inadequate reimbursement. Discussion: CHCs have implemented many foundational components of integrated BH. Future work should address barriers to integration and racial disparities in access to integrated BH. Keywords: behavioral health, primary care, integration, health centers, Nearly one in five adults in the United States experiences a mental health or substance use disorder each year (National Institute of Mental Health, 2019). An important strategy to address [...]

Materials related to Japanese military 'comfort women' in Korea are managed by several institutions. Each digital archive has their own metadata schema and management policies. So far, a standard or a common guideline for describing digital records is not formalized. We propose a Japanese military 'comfort women' knowledge graph to semantically interlink the digital records from distributed digital archives. To build a Japanese military 'comfort women' knowledge graph, digital records and descriptive metadata were collected from existing digital archives. A list of metadata was defined by analyzing commonly used properties and a knowledge model designed by reusing standard vocabularies. Knowledge was constructed by interlinking the collected records, external data sources, and enriching data. The knowledge graph was evaluated using the FAIR data maturity model., INTRODUCTION In December 1991, Kim Hak-Sun (a Korean) became the first woman to disclose and identify as a former 'comfort woman.' (1) In February 1992, Ms. Itoh Hideko discovered three [...]

Context: Primary care has an important role in identifying individuals that use tobacco, providing advice to quit and assisting cessation by referring to evidence-based resources such as state tobacco quitlines (QL). Objective: To evaluate the degree to which provision of tobacco cessation care is sustained 24 months after implementation of an Ask-Advise-Refer strategy with an electronic referral to the QL for patients ready to quit tobacco. Setting: Eight community primary care clinics in a Midwest safety-net health system. Methods: A non-randomized stepped wedge study design was used to implement an intervention consisting of 1) changes to the electronic health record (EHR) referral functionality and 2) expansion of rooming staff roles to include performing tobacco cessation care tasks. Outcomes assessed from the EHR included asking about tobacco use, providing brief advice to quit and assessing readiness to quit. Generalized estimating equations (GEE) methods were used to compute odds ratios contrasting the pre-implementation (Pre-I) vs. 12, 18 and 24-month post-implementation (Post-I). Results: A total of 305,665 patient visits were evaluated. Significantly higher odds of asking and advising were observed for all three Post-I periods compared to Pre-I rates, with rates either remaining steady or increasing over time. Asking about tobacco use Pre-I was 27.5% vs. 63.3% 24-months, OR 5.1 (95% CI 4.8-5.4). Advising tobacco users to quit at Pre-I was 45.0% vs. 79.0% 24- months, OR 3.6 (95% CI 3.1-4.1). Odds of assessing readiness to quit were higher across all three Post-I periods compared with Pre-I but declined after 12 months. Assessing readiness to quit Pre-I was 15.9% vs 35.4% at 24 months, OR 2.7 (95% CI 2.4-3.1). There was substantial variability across each outcome by clinical site with the greatest variability for assessing readiness; site-level performance at 24 months range was 9%-74%. Conclusions: Embedding a new process in the rooming workflow can result in substantial and sustained impact on tobacco assessment and provision of brief advice. Improvement in assessing patient readiness to quit was less successfully sustained and highly variable across clinical sites. EHR upgrade changes and staff and clinician turnover are disruptions that may negatively impact performance and additional implementation support strategies to reinforce positive performance and mitigate disruptions should be deployed and tested., Presenters Susan Flocke, PhD, Steffani Bailey, PhD, Eileen Seeholzer, Elizabeth Albert, Steven [...]

Context: When diabetes is not well-controlled, it can lead to numerous negative sequelae. At present, interventions used to prevent these negative sequelae are tracked, measured, and reported, both in our EHR and nationally, but the outcomes of these interventions are not. Objective: To determine the outcomes of care for our patients with diabetes. Study Design and Analysis: Cross-sectional study calculating incidence and prevalence of diabetes and its associated comorbidities. Setting or Dataset: All JPS patients from January 1, 2010, through February 4, 2022. Population Studied: All JPS patients with a diagnosis of 'Diabetes'. Intervention/Instrument: SlicerDicer, a software tool accessible in Epic, was used to find all unique medical record numbers (MRNs) associated with a diagnosis of 'Diabetes' from January 1, 2010, through February 24 (first batch of data), 2021, and then again from January 1, 2010 through February 4, 2022 (second batch of data). Outcome Measures: Prevalence and incidence of the various negative sequelae of diabetes in comparison to published averages. Results: From the first batch of data, 93,496 unique MRNs had the diagnosis of 'Diabetes'. The prevalence of each comorbidity was as follows: peripheral neuropathy of 33.93%, chronic kidney disease of 20.09%, retinopathy due to diabetes of 10.32%, myocardial infarction of 10.21%, hypoglycemia of 8.07%, stroke of 6.64%, ketoacidosis of 3.28%, amputations of 2.17%, and hyperosmotic hyperketotic syndrome of 0.40%. From the second batch of data, 100,266 unique MRNs had the diagnosis of 'Diabetes', accounting for 6,770 new cases of diabetes to JPS. The incidence and prevalence, respectively, of each comorbidity for that subset of patients was as follows: peripheral neuropathy of 5.11% and 35.13%, chronic kidney disease of 2.24% and 20.55%, retinopathy due to diabetes of 1.04% and 10.56%, myocardial infarction of 0.91% and 10.35%, hypoglycemia of 0.94% and 8.40%, stroke of 0.85% and 7.00%, ketoacidosis of 0.30% and 3.36%, amputations of 0.15% and 2.17%, and hyperosmotic hyperketotic syndrome of 0.03% and 0.41%. Conclusions: At JPS, until present, the incidence and prevalence of the negative sequelae of our patients with diabetes was unknown; therefore, the outcomes of care for our patients with diabetes was unknown. When compared to available information published by the CDC, JPS demonstrates better outcomes for its patients with diabetes compared to the national average., Presenters Anatoli Berezovsky, [...]

Context: Electronic health records (EHR) provide an opportunity for developing decision support and other types of learning health system (LHS) initiatives. Careful understanding of the population of interest and how clients are represented in the data is essential for problem selection and for effective study design and analysis of data to solve the problem. The Alliance for Healthier Communities is one of the first primary care LHS in North America, serving complex, at-risk clients through Community Health Centres (CHCs) across Ontario, Canada. We propose that to properly understand their electronic health record data both simple statistical techniques commonly seen in descriptive epidemiology and more complex techniques from artificial intelligence will be useful. Objective: To summarize characteristics of ongoing primary care clients served by CHCs for the purpose of informing future LHS initiatives. Study Design and Analysis: Table-based summaries are given for all outcome measures. An Ising model is used to identify condition co-occurrences, non-negative matrix factorization is used to examine care provider teams, and time series clustering is used to explore care frequency patterns. Setting or Dataset: EHR data from all CHCs across Ontario, Canada in 2009- 2019. Population Studied: Clients over 18 years old in 2009 who indicated a CHC as their primary care provider and had at least one encounter in 2009-2019. Intervention/Instrument: N/A. Outcome Measures: Sociodemographic characteristics (from structured client characteristic table, e.g., age, education), clinical characteristics (20 chronic conditions, four acute conditions, multimorbidity), and health care use characteristics (providers involved, complexity, frequency). Results: There were 221,047 eligible clients. Social determinants and clinical conditions are most prominent in CHCs serving those most at risk in urban settings; the former are also more prevalent in clients with multimorbidity. Physician and nursing types provide most care, with heterogeneous combinations of other providers. There is notable within- and between- client variability in care complexity and frequency. We identify implications for future analyses such as cohort building and risk prediction. Conclusions: Simple statistical and artificial intelligence techniques can be used to effectively describe a complex primary care population; this population-level overview provides a foundation for future LHS initiatives., Presenters Jacqueline Kueper, PhD, MSc, Jennifer Rayner, PhD, Merrick Zwarenstein, MD, PhD, Dan [...]

Context: Despite the expansion of social risk data collection, little research has examined how the increased availability of social risk information influences care delivery. Objective: To explore the collection and use of social risk data in community health centers (CHC) in Texas, and the facilitators/barriers to these social care activities in pre/peri-COVID-19 pandemic periods. Study Design & Analysis: Mixed methods: 1) interviews and thematic analysis; 2) surveys and electronic health record (EHR) review and descriptive statistics. Setting: Four urban/suburban CHCs. Population Studied: Providers, staff and patients. Instrument: Semi-structured interview guide, online survey. Outcome Measures: Perspectives on integrating a range of social care activities; reach of social screening. Results: We conducted 27 interviews (15 providers; 12 staff), 97 provider surveys, and reviewed EHR data from two CHCs. There was broad support for integrating social care; support increased peri-pandemic. 90% of survey respondents reported incorporating social screening into patient conversations; 46% of those screening used standardized screening tools. 29% of all survey respondents were unaware their clinic had a standardized tool. EHR data showed that both numbers of screens per month and rates of screens per encounters increased (4% of patient encounters in 8/2019 to 44% in 2/2021). There were significant differences in whether patients were screened by race/ethnicity and language (p Conclusions: CHCs were in the early phases of social care delivery, primarily focused on awareness and assistance with little dedicated attention to adjustment or population health-level evaluations. Differences in screening reach by patient demographics raise concern that social care might exacerbate disparities. Future research should explore equity in implementation and effectiveness. Overcoming barriers to reach, sustainability and equity will require supports targeted to program design and development, workforce capacity, quality improvement, and community investments in social services., Presenters Emilia De Marchis, MD, MS, MAS, Benjamin Aceves, PhD, Na'amah Razon, MD, PhD, Laura Gottlieb, MD, [...]

Collection of data on sexual orientation, gender identity and intersex status will help to reduce health disparities that affect people from sexual and gender minority communities., Author(s): Alex S. Keuroghlian [sup.1] [sup.2] [sup.3] Author Affiliations: (1) The Fenway Institute, Fenway Health, Boston, USA (2) Massachusetts General Hospital, Boston, USA (3) Harvard Medical School, Boston, USA Compared [...]