YLYan Li

Yan Li

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Yan Li has a total of 68 co-authors. Here are the top co-authors by number of shared publications:

  • Almut Gertrud Winterstein

    Department of Pharmaceutical Outcomes and Policy, University of Florida, Gainesville, USA

    4 shared publications
  • Yanmin Zhu

    Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, University of Florida, Gainesville, FL, USA

    3 shared publications
  • Betty Chaar

    The University of Sydney

    2 shared publications
  • Hong-Guang Xie

    Division of Clinical Pharmacology, General Clinical Research Center, Nanjing First Hospital, Nanjing Medical University, Nanjing, China; Department of Clinical Pharmacy, Nanjing Medical University School of Pharmacy, Nanjing, China; Department of Clinical Pharmacy, School of Basic Medicine and Clinical Pharmacy, China Pharmaceutical University, Nanjing, China

    2 shared publications
  • Huilin Tang

    Center for Preventative Ophthalmology and Biostatistics, Department of Ophthalmology University of Pennsylvania Perelman School of Medicine Philadelphia Pennsylvania USA

    2 shared publications
  • Jonathan Penm

    Faculty of Medicine and Health, School of Pharmacy, The University of Sydney, Sydney, Australia

    2 shared publications

Latest Publications

  • Algorithmic Identification of Treatment‐Emergent Adverse Events From Clinical Notes Using Large Language Models: A Pilot Study in Inflammatory Bowel Disease

    by Anna L. Silverman, Madhumita Sushil, Balu Bhasuran, Dana Ludwig, James Buchanan, Rebecca Racz, Mahalakshmi Parakala, Samer El‐Kamary, Ohenewaa Ahima, Artur Belov, et al.

    Outpatient clinical notes are a rich source of information regarding drug safety. However, data in these notes are currently underutilized for pharmacovigilance due to methodological limitations in text mining. Large language models (LLMs) like Bidirectional Encoder Representations from Transformers (BERT) have shown progress in a range of natural language processing tasks but have not yet been evaluated on adverse event (AE) detection. We adapted a new clinical LLM, University of California – San Francisco (UCSF)‐BERT, to identify serious AEs (SAEs) occurring after treatment with a non‐steroid immunosuppressant for inflammatory bowel disease (IBD). We compared this model to other language models that have previously been applied to AE detection. We annotated 928 outpatient IBD notes corresponding to 928 individual patients with IBD for all SAE‐associated hospitalizations occurring after treatment with a non‐steroid immunosuppressant. These notes contained 703 SAEs in total, the most common of which was failure of intended efficacy. Out of eight candidate models, UCSF‐BERT achieved the highest numerical performance on identifying drug‐SAE pairs from this corpus (accuracy 88–92%, macro F1 61–68%), with 5–10% greater accuracy than previously published models. UCSF‐BERT was significantly superior at identifying hospitalization events emergent to medication use (P < 0.01). LLMs like UCSF‐BERT achieve numerically superior accuracy on the challenging task of SAE detection from clinical notes compared with prior methods. Future work is needed to adapt this methodology to improve model performance and evaluation using multicenter data and newer architectures like Generative pre‐trained transformer (GPT). Our findings support the potential value of using large language models to enhance pharmacovigilance.


  • Open Access

    An Alzheimer’s disease blood test study in a diverse community‐based population: Preliminary results from SEABIRD

    by Melody Li, Yan Li, Suzanne E. Schindler, Daniel Yen, Siobhan Sutcliffe, Ganesh M. Babulal, Tammie L.S. Benzinger, Eric J. Lenze, Randall J. Bateman

    Background: Alzheimer’s disease (AD) blood tests are needed to aid in the evaluation of dementia, but the performance of these tests should be evaluated in diverse groups and real‐world settings. The goals of the Study to Evaluate Amyloid in Blood and Imaging Related to Dementia (SEABIRD) were to enroll 1120 participants to determine the participant acceptability and validity (relative to amyloid PET) of a plasma amyloid‐β 42/40 test in a diverse, community‐based sample of older adults, and explore the impact of important factors (age, race, education, cognition, APOE genotype, and medical conditions) on the blood test’s performance.Method: SEABIRD measured plasma amyloid‐β 42/40 in a diverse population of older adults in the Saint Louis, Missouri, USA area. Participants completed a blood collection, cognitive screening (AD8® dementia screening interview and Montreal Cognitive Assessment [MoCA]), and a survey about their study experience and perceptions of the blood test. A subset of participants completed additional blood collection for reproducibility, amyloid PET and MRI for validation of the blood test, and the Clinical Dementia Rating® (CDR) for validation of the cognitive screening measures.Result: Of the 859 participants enrolled in this ongoing study, 20.6% self‐identified as Black or African American and the percentage of APOE ε4 carriers was 30.2% (Table 1). Compared to the general population, SEABIRD participants were more likely to be female, aged 70‐79 years, and highly educated. They were less likely to report high cholesterol, diabetes, kidney disease, and stroke, and more likely to report depression. The receiver operating characteristic area under the curve for distinguishing cognitively unimpaired (CDR = 0) from cognitively impaired (CDR>0) individuals was 0.72 for the informant‐rated AD8, 0.46 for the self‐rated AD8, 0.70 for the MoCA, and 0.79 for the AD8/MoCA composite (Table 2). Survey results indicated that the blood test was well accepted (Figure 1), but it was perceived more positively by white and highly educated individuals.Conclusion: A study of an AD blood test achieved rapid enrollment, broader diversity than typical AD cohorts, and high rates of participation. These results suggest that an AD blood test could be used for screening in a diverse clinical population.


  • Improving routine use of clinical pathway decision support through EMR integration.

    by Dinesh Kotak, Mary Mitsuko Ichiuji, Laura Asakura, Shawn Hanifin, Nancy Aye, Tatjana Kolevska, Carol Cain, Violeta Rabrenovich, Mark Lindberg, Yan Li

    432 Background: There has been an explosive growth in oncology with an evolution of precision oncology based on molecular profiling. Providing evidence-based, high quality, equitable care is a foundational mission for Kaiser Permanente (KP). Thus, KP created a systematic approach toward oncology care pathways that incorporated organizational leadership, structured participation and feedback, novel integration of clinical decision support (CDS) tools, and evaluation of efficacy, safety, and value. Our goal was to make it easy and preferable for physicians to access clinical decision support for every patient, every time. Methods: KP’s inter-regional oncology chiefs, supported by relevant teams in evidence analysis, pharmacy, and other clinical specialties, convened monthly meetings in 10 oncology sub-specialties to synthesize evidence and create care pathways for priority conditions. The pathways are published to KP’s Clinical Library (CL), a trusted, curated resource of clinical knowledge guidance and tools, available via KP’s Epic Systems electronic medical record (EMR). The pathways are represented as intuitive, graphical, clickable process diagrams, with additional clinical content available as hover text. CL uses standard Epic interfaces to receive contextual information when accessed from a patient encounter. When the physician browses the CL oncology pathway and selects a treatment recommendation, CL returns to the EMR with the specific treatment plan queued up for ordering. We began initial deployment of these inter-regional pathways in Feb 2020. We combined EMR data with CL data to understand the impact of this clinical decision support. Results: We experienced dramatic uptake of oncology clinical pathways despite offering minimal training and no external incentives. KP orders over Beacon treatment plans annually in the clinical conditions supported by our pathways. In the first quarter of deployment, clinicians ordered 9% of these protocols through clicking on CDS in CL. Over the course of 2 years, we saw pathway-based ordering grow to over 75% of eligible patients. Conclusions: We find that we have made considerable progress toward our goal of having physician users consult clinical pathways decision support before ordering oncology drug treatment protocols. As with all clinical guidance, we would not expect 100% of patients to follow the recommendation, due to clinical variation and co-morbidity, but we find that by making clinical pathways easily accessible and actionable within normal clinical workflow, physicians are consulting the latest evidence in their treatment.

  • Open Access

    Utilization Patterns of Skeletal Muscle Relaxants Among Commercially Insured Adults in the United States from 2006 to 2018

    by Yan Li, Chris Delcher, Gary M Reisfield, Yu-Jung Wei, Joshua D Brown, Almut G Winterstein

    Objective: To examine the prevalence and duration of skeletal muscle relaxant (SMR) treatment among commercially insured adults in the United States.Methods: We used the MarketScan Research Database to identify a cohort of adults 18 to 64 years who had ≥2-year continuous enrollment between 2005 and 2018. We estimated the prevalence of SMR treatment using a repeated cross-sectional design and derived treatment duration using the Kaplan-Meier method. Analyses were stratified by age group, sex, geographic region, individual SMR agent, and musculoskeletal disorder.Results: 48.7 million individuals were included. Treatment prevalence ranged from 61.5 to 68.3 per 1,000. About one-third of users did not have a preceding musculoskeletal disorder diagnosis. Cyclobenzaprine was the dominant agent accounting for >50% of prescriptions. The considerable growth in the use of baclofen, tizanidine, and methocarbamol paralleled with a decline in carisoprodol and metaxalone use. The prevalence was highest in the South while lowest in the Northeast. The median treatment duration was 14 days with 4.0%, 1.9%, and 1.0% of individuals using SMRs for more than 90, 180, and 365 days, respectively. Compared with cyclobenzaprine, patients initiating baclofen, tizanidine, and carisoprodol had longer treatment duration.Conclusions: SMRs are widely used in the United States. Their use slightly increased in recent years, but trends varied among individual agents, patient groups, and geographic regions. Despite limited evidence to support efficacy, a sizable number of U.S. adults used SMRs for long-term and off-label conditions. Further study is needed to understand determinants of treatment as well as outcomes associated with such use.


  • Medication Use for ADHD and the Risk of Driving Citations and Crashes Among Teenage Drivers: A Population-Based Cohort Study

    by Almut G. Winterstein, Yan Li, Tobias Gerhard, Stephan Linden, Jonathan J. Shuster

    Objectives: To evaluate the real-world effectiveness of ADHD medications on adverse driving outcomes in teenage drivers with ADHD. Method: We retrospectively followed 15- to 20-year-old ADHD patients with valid driver’s license to compare the risk for crashes and citations between periods with and without ADHD medication use, using Florida Medicaid records linked to Department of Motor Vehicles data from 1999 to 2004. Patient-level demographic, clinical, and driver licensing characteristics as well as county-level crash and traffic statistics were adjusted in Cox models. Results: A total of 2,049 patients had 67 crashes and 319 citations. Adjusted hazard ratios comparing ADHD medication use versus no use were 1.22 (95% confidence interval [CI] = [0.66, 1.90]) and 0.89 (95% CI = [0.69, 1.13]) for crashes and citations, respectively. Conclusion: Our study showed no evidence that ADHD medication use was associated with a reduced risk of adverse driving outcomes among teenage drivers enrolled in Medicaid programs. Limitations in interpreting this finding are presented.


  • Internal validation of Medicaid Analytic eXtract (MAX) data capture for comprehensive managed care plan enrollees from 2007 to 2010

    by Yan Li, Yanmin Zhu, Chao Chen, Xi Wang, Yoonyoung Choi, Carl Henriksen, Almut G. Winterstein

    Purpose: Because of concerns over incomplete medical encounter capture in Medicaid capitated comprehensive managed care (CMC) plans, researchers have traditionally confined analyses to fee‐for‐service (FFS) enrollees. We aimed to evaluate the usability of data for CMC enrollees in Medicaid Analytic eXtract (MAX) files for 29 states from 2007 to 2010.Methods: We applied 7 measures to MAX inpatient, other therapy, and prescription drug files for each state and study year. Four measures were based on “connectivity” criteria where we expected use of a select essential service to be closely connected to another, resulting in “service pairs.” Three measures were based on “continuity” criteria where we expected patients to continue chronically used services or treatments when they switched enrollment from FFS to CMC plans. High proportions of continuity and comparable proportions of patients with complete service pairs relative to FFS enrollees may suggest complete data capture for CMC enrollees. Data of states that met preset criteria were considered usable for research and policy analyses.Results: The completeness of CMC enrollees' data in MAX varied by states. Among 22 states having at least 5% CMC plan enrollment, data of 12 states met our quality standard and were considered usable starting in 2007. Four states had usable data starting in 2008 and one in 2009.Conclusions: The completeness of CMC enrollees' data in MAX improved over the study period. In 17 out of 29 states, CMC enrollees' data in selected years were comparable with FFS enrollees and can be considered for use in analysis.


  • Determinants of Pharmacological Treatment Initiation and Persistence in Publicly Insured Adults With Attention-Deficit/Hyperactivity Disorder

    by Yan Li, Wei Liu, Yanmin Zhu, Xinyue Liu, Almut Gertrud Winterstein

    Objectives: The objectives of this study were to describe the longitudinal trends of pharmacological treatment initiation and persistence in adult patients with attention-deficit/hyperactivity disorder (ADHD) and to explore how patients' sociodemographic and clinical characteristics affect these patterns. Methods: A retrospective cohort of 23,747 adult patients with new ADHD episodes was assembled using Medicaid extract files of 29 US states from 1999 and 2010. Sociodemographic and clinical characteristics of treatment initiators and noninitiators, and treatment persistent and nonpersistent patients were compared using multivariable logistic regression. Results: The overall treatment initiation and 1-year persistence rate was 22.5% and 34.2%, respectively. After adjustment, the propensity of initiating pharmacotherapy and being persistent tripled (2010 vs 2000; odds ratio, 3.25; 95% confidence interval, 2.61–4.04) and almost doubled (2009 vs 2000; odds ratio, 1.86; 95% confidence interval, 1.20–2.90) over the study period. Male, racial and ethnic minorities, and younger age groups had a decreased propensity for treatment initiation and persistence. The likelihood of initiating treatment was similar for rural and urban residents but varied among different US states. Severe mental comorbidities and, concordantly, use of antipsychotics or mood stabilizers reduced the propensity of treatment initiation but had little effect on persistence. Patients filling sustained release stimulants were most likely to be persistent. Finally, preexisting cardiovascular conditions reduced the likelihood of receiving treatment but were not associated with persistence. Conclusions: The propensity to initiate and persist on pharmacological treatment in adults with ADHD has increased in the last decade but remains low. Treatment patterns were closely related to several sociodemographic and clinical determinants. Our findings provide valuable information to ensure appropriate treatment choices for these patients.


  • Prevalence of ADHD in Publicly Insured Adults

    by Yanmin Zhu, Wei Liu, Yan Li, Xi Wang, Almut G. Winterstein

    Objective: To estimate the prevalence of adult ADHD diagnosis and treatment in U.S. Medicaid beneficiaries. Method: Using outpatient, inpatient, and pharmacy billing records for patients eligible for Medicaid fee-for-service benefits within 29 states from 1999 to 2010, we estimated the annual prevalence of ADHD diagnosis and treatment. Results: The prevalence of ADHD diagnosis increased from 2.20 per 1,000 patients in 1999 to 10.57 in 2010. Likewise, prevalence of ADHD treatment increased from 1.95 per 1,000 patients in 1999 to 13.16 in 2010. Between 40% and 65%, patients had ADHD drug prescription fills 6 months after ADHD diagnosis, whereas 45% to 55% of the beneficiaries with an ADHD drug prescription fill had ADHD diagnoses within 6 months before the prescription. Conclusion: In publicly insured adults, the prevalence of ADHD diagnosis and treatment increased dramatically over the years. Approximately half of the diagnosed patients are not treated, whereas half of the treated adults are not diagnosed.


    • 18 October 2014
      • Vol. 46  (5) , 809-13

    [Pharmaceutical care of serious bleeding induced by tramadol-warfarin interaction: a case report].

    by Shu-Jie Dong, Ting Zhang, Yan Li, Suo-Di Zhai

    Warfarin is a high-alert medication, which may result in bleeding if used improperly. In our case, one elderly female with atrial fibrillation had taken warfarin for more than half a year, and her international normalized ratio (INR) was maintained within the therapeutic range. The patient began to take tramadol to alleviate her shoulder pain. Three days later she presented hematuresis and had ecchymosis in her right upper arm, and in the meantime her INR rose to 10.04. Clinical pharmacists analyzed the cause for bleeding by searching relevant literature, and finally discovered the interaction between warfarin and tramadol. On the basis of that, the clinical pharmacists provided pharmaceutical care, offered specific medication education, as well as assisted the physicians to establish the medication plan for warfarin reuse. Eventually, her INR declined to reference ranges, and her hematuresis and ecchymosis were alleviated significantly. This successful case reveals that clinical pharmacy services contribute to better treatment outcomes. Clinical pharmacists can play an active role in anticoagulation management in healthcare team.

  • Factors Affecting the Implementation of Clinical Pharmacy Services in China

    by Jonathan Penm, Rebekah Moles, Holly Wang, Yan Li, Betty Chaar

    New policies in China have recently led to the implementation of clinical pharmacy services in hospitals. We explored the views of hospital administrators, pharmacy directors, clinical pharmacists, and dispensing pharmacists about the factors affecting clinical pharmacy services in China, using the framework approach and organizational theory. We conducted 30 interviews with 130 participants at 29 hospitals (both secondary and tertiary) in Beijing, Zhengzhou, Luoyang, and Shanghai. We found that the barriers to and facilitators of implementation of clinical pharmacy services slotted into the environment and participant dimensions of Scott’s adapted version of Leavitt’s organizational model. External support from government was perceived as crucial to promoting pharmacy services. It is proposed that the internationally recognized Basel Statements of the International Pharmaceutical Federation also provide a strong foundation for guiding China in implementing clinical pharmacy services.