XTXi Tan

Xi Tan

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xi Tan has a total of 96 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Predicting Treatment Failure With Sodium-Glucose Cotransporter-2 Inhibitors in People With Type 2 Diabetes: Novel Artificial Intelligence and Machine Learning Approach

    by Doyoung Kwak, Xi Tan, Yuanjie Liang, Caroline Swift, Chalak Muhammad, Xu Shi

    Background: The rate of treatment failure with sodium-glucose cotransporter-2 inhibitors (SGLT2i) is high among individuals with type 2 diabetes (T2D). Accurately predicting SGLT2i treatment failure is important for improving the clinical management of T2D. Objective: The study aimed to use machine learning (ML) models to identify factors predicting treatment failure with SGLT2i in T2D and to evaluate model performance. Methods: This retrospective observational cohort study included adults with T2D treated with SGLT2i (2016-2024). The primary outcome was overall treatment failure with SGLT2i during follow-up (≥180 days after SGLT2i initiation). The secondary outcome was subtypes of treatment failure with SGLT2i (treatment discontinuation, failure with action, and inertial failure) or nonfailure, which was defined as not meeting the definition for one of the failure subtypes. Variables potentially associated with treatment failure were assessed during the year before SGLT2i treatment initiation (analysis 1) and the year before SGLT2i treatment failure (analysis 2). Using these variables, ML models—logistic regression (LR), multilayer perceptron (MLP), extreme gradient boosting (XGBoost), and Transformer—were used to identify significant predictors of the outcomes. Model performance metrics (accuracy, area under the curve, precision, recall, and F 1 -score) were calculated. Using Shapley Additive Explanations methodology, key features were identified based on their impact on model predictions. LR and Transformer models using key features were further evaluated for their potential to support the development of a risk score for predicting treatment failure with SGLT2i. Results: Among all individuals in the study (N=62,222), 71% (n=44,156) had treatment failure with SGLT2i. Across subtypes, failure with action (n=23,839, 38.3%) was more common than treatment discontinuation (n=16,449, 26.4%) and inertial failure (n=3868, 6.2%). Model performance was moderate in both analyses. In analysis 1, the accuracy ranged from 0.72 to 0.73 for predicting overall treatment failure and from 0.56 to 0.57 for predicting the subtype of treatment failure. In analysis 2, the accuracy ranged from 0.74 to 0.75 for predicting overall treatment failure and from 0.61 to 0.63 for predicting the subtype of treatment failure. XGBoost, MLP, and Transformer models showed small improvements compared with LR. Using the top 9 key features identified from the Shapley Additive Explanations analysis, the Transformer model performed similarly in accuracy and area under the curve to its counterpart using the full feature set. Conclusions: Performance across the LR, MLP, XGBoost, and Transformer models was moderate. The advanced ML models performed slightly better than LR. Overall, the results suggest that further model advancements and increased data availability are needed to better predict treatment failure with SGLT2i. The LR coefficients from the key features model may inform the development of a risk score to predict SGLT2i treatment failure. Accurate prediction could inform individualized treatment planning for individuals with T2D.

  • Open Access

    Cardiometabolic and Renal Outcomes in Semaglutide Users with Type 2 Diabetes Achieving Glycemic and Weight Goals: An Observational Cohort Study

    by Xi Tan, Wan-Lun Tsai, Yuanjie Liang, Caroline Swift, Gang Fang, Chalak Muhammad, Adam de Havenon

    Introduction Within the cardiovascular-kidney-metabolic syndrome (CKM) framework, semaglutide has demonstrated benefits beyond glycemic control and weight loss in clinical trials. However, most real-world studies in type 2 diabetes (T2D) have limited assessment of broader cardiometabolic and renal outcomes. We evaluated CKM-relevant outcomes among individuals with T2D who achieved substantial hemoglobin A1c (HbA1c) and weight improvements after initiating semaglutide in real-world settings. Methods This observational pre-post study used Optum’s de-identified Market Clarity Data from January 1, 2007, to June 30, 2024. The primary cohort comprised individuals with T2D who achieved glycemic control (HbA1c < 7%) and weight loss (≥ 5%) goals after semaglutide initiation. We compared baseline (1 year before initiation) with 1st-year and 2nd-year follow-up for cardiometabolic endpoints (3-point and 5-point major adverse cardiovascular events [MACE]), cardiometabolic risk factors, and renal outcomes. Sensitivity analysis was performed in an exploratory cohort of patients in the top tertile of Hb1Ac reduction and weight loss. Results We identified 413 patients in the primary cohort (mean age 59.6 years and balanced sex distribution). Significant reductions were observed in low-density lipoprotein cholesterol (LDL-C), very-low-density lipoprotein cholesterol (VLDL-C), total cholesterol, triglycerides, systolic blood pressure, and diastolic blood pressure and high-density lipoprotein cholesterol (HDL-C) increased at both 1 and 2 years after semaglutide initiation (all p < 0.001). Fewer than 1% of patients experienced a ≥ 40% decline in estimated glomerular filtration rate (eGFR) during follow-up. Mean change in urine albumin-to-creatinine ratio (UACR) were − 20.13 mg/g in the 1st year (p < 0.001) and − 54.03 mg/g in the 2nd year (p < 0.001). The event rate of 3-point MACE decreased from 26.82 per 1000 person-years (PY) during baseline to 22.31 per 1000 PY in the 2nd year. The sensitivity analysis showed consistent results. Conclusion In this real-world study, semaglutide users who achieved glycemic and weight goals exhibited marked improvements in cardiometabolic and renal outcomes over 2 years.

  • Open Access

    Comparison of cardiovascular outcomes between once‐weekly semaglutide and dulaglutide in adults with type 2 diabetes and established atherosclerotic cardiovascular disease in the United States

    by Xi Tan, Yuanjie Liang, Lin Xie, Joanna Harton, Cynthia Gutierrez, Chalak Muhammad, Caroline Swift, Adam de Havenon

    Aims: This study aims to compare the risk of major adverse cardiovascular events (MACE) among United States individuals with type 2 diabetes (T2D) and atherosclerotic cardiovascular disease (ASCVD) treated with once‐weekly semaglutide vs. dulaglutide. Materials and methods: This was a retrospective cohort study using Optum's de‐identified Clinformatics Data Mart (Optum CDM) from 1 January 2007 through 30 September 2024. New initiators of semaglutide or dulaglutide ≥18 years with both T2D and ASCVD were included. The index date was the first date of a prescription claim for semaglutide or dulaglutide within the index medication identification period (1 January 2018 through 31 March 2024). The primary outcome was 3‐point MACE (stroke, myocardial infarction [MI], cardiovascular [CV]‐related death). Entropy balancing was applied to balance baseline characteristics. Weighted incidence rates per 1000 person‐years and doubly robust Cox proportional hazard ratios were reported. Results: There were 75 243 enrolees included (semaglutide, 42 007; dulaglutide, 33 236). The mean age was 68.2 and 69.3 years (unweighted) in the semaglutide and dulaglutide cohorts, respectively. After balancing, standardized mean differences were p < 0.001). Conclusions: Among United States Medicare Optum CDM enrolees with T2D and ASCVD, semaglutide was associated with reduced risks of CV outcomes compared with dulaglutide. These results help to address an important evidence gap in selecting glucagon‐like peptide‐1 receptor agonists for this high‐risk population.


  • Open Access

    Comparative effectiveness of OW GLP-1 RAs and other glucose-lowering therapies among Medicare beneficiaries with T2D and ASCVD

    by Xi Tan, Joanna Harton, Cynthia Gutierrez, Yuanjie Liang, Lin Xie, Chalak Muhammad, Caroline Swift, Adam de Havenon


  • Open Access

    Healthcare Resource Utilization and Costs in Individuals Who Discontinue Liraglutide and Who Switch from Liraglutide to Once-Weekly Injectable Semaglutide

    by Xi Tan, Brenna L. Brady, Lin Xie, Yurek Paprocki

    Introduction This study evaluated healthcare resource utilization (HCRU) and costs in the USA for people with type 2 diabetes (T2D) who discontinued the injectable glucagon-like peptide 1 receptor agonist (GLP-1 RA) once-daily liraglutide for T2D (with no other glucose-lowering agent added) or switched from liraglutide to the GLP-1 RA once-weekly semaglutide for T2D. Methods In this observational cohort study, we utilized claims data (Merative MarketScan [Merative, Ann Arbor, MI, USA] Commercial and Medicare Database; January 1, 2017–March 31, 2021) to compare HCRU and costs between individuals who discontinued liraglutide (“discontinuers”) and those who switched from liraglutide to semaglutide (“switchers”). Patients were indexed between January 1, 2018 and March 31, 2020. Outcomes were compared between discontinuers and switchers over the 360-day post-index period using stabilized inverse probability of treatment weighting. Results Characteristics of the two cohorts were balanced after weighting. Switchers had significantly lower HCRU in inpatient and emergency department (ED) settings compared with discontinuers. Mean [standard deviation] total medical costs were significantly lower for switchers ($8513 [$18,931]) than for discontinuers ($13,585 [$52,011], p < 0.001), driven by reduced inpatient costs (2.6 times lower) and ED costs (1.6 times lower). Conclusion This analysis demonstrates that the cohort of people switching from liraglutide to semaglutide was associated with significantly lower HCRU and costs when compared with people discontinuing liraglutide only. These findings imply that switching to semaglutide could be a good option for people with T2D for whom liraglutide is no longer optimal.

  • Open Access

    Macrovascular and microvascular complications in US Medicare enrollees with type 2 diabetes with and without atherosclerotic cardiovascular disease

    by Tyler J. Dunn, Xi Tan, Joanna Harton, Sonia Kim, Lin Xie, Cory Gamble, Daniel Rotroff

    Aims: To assess the incidence of macrovascular and microvascular complications in US Medicare enrollees diagnosed with T2D with and without established ASCVD. Materials and Methods: We conducted a retrospective cohort study using Medicare fee‐for‐service claims data from 1 January 2006, through 31 December 2021. Baseline demographic and clinical characteristics were assessed in the 1‐year prior to indexing. Cumulative incidences of various diabetes complications were assessed until the first microvascular or macrovascular complication of interest, the end of the study period or death. Results: A total of 2 326 726 patients and 640 666 patients met study inclusion/exclusion criteria for the T2D cohort and T2D + ASCVD sub‐cohort, respectively. The incidence rate of any macrovascular event was 483.34 per 10 000 person‐years in the T2D cohort. Overall, the 1‐year cumulative incidence rate of any macrovascular event was 3.90%. Coronary heart disease (T2D, 3.24%; T2D + ASCVD, 8.10%) and peripheral artery disease (T2D, 1.97%; T2D + ASCVD, 7.33%) were the macrovascular events with the greatest 1‐year cumulative incidence. Patients developed microvascular complications at a rate of 1569.28 per 10 000 person‐years in the T2D cohort and 1859.80 per 10 000 person‐years in the T2D + ASCVD sub‐cohort. The 1‐year cumulative incidence of any microvascular event was 16.88% in the T2D cohort and 21.16% in the T2D + ASCVD sub‐cohort. Neuropathy and nephropathy were the microvascular events with the greatest 1‐year cumulative incidence in both cohorts: T2D, 8.34% and 7.02%; T2D + ASCVD, 10.65% and 9.12%, respectively. Conclusions: The frequencies of macrovascular and microvascular complications highlight the importance of annual cardiovascular risk assessment in patients with T2D, especially those with established ASCVD. Plain Language Summary: What is the context and purpose of this research study?: People with type 2 diabetes (T2D), especially those with hardening of the arteries or atherosclerotic cardiovascular disease (ASCVD), are at risk of developing problems with large and small blood vessels throughout their bodies. These problems can lead to serious health risks and death. We wanted to determine how many Medicare people with T2D, with and without ASCVD, developed problems with their large and small blood vessels over time. What was done?: We used Medicare claims data to measure how many people with T2D, and a subset with T2D+ASCVD, developed problems with these blood vessels from January 1, 2006, through December 31, 2021. What were the main results?: In the year after developing T2D, we found that 3.90% of patients developed problems with their large blood vessels. The most common large blood vessel problems that occurred within the first year of developing T2D (or T2D+ASCVD) were coronary heart disease (T2D, 3.24%; T2D+ASCVD, 8.10%) and peripheral artery disease (T2D, 1.97%; T2D+ASCVD, 7.33%). In the year after developing T2D (or T2D+ASCVD), we found that 16.88% of people with T2D and 21.16% of people with T2D+ASCVD developed problems with their small blood vessels. The most common small blood vessel problems were nerve damage (T2D, 8.34%; T2D+ASCVD, 10.65%) and kidney disease (T2D, 7.02%; T2D+ASCVD, 9.12%) in the first year. What is the originality and relevance of this study?: This was the first study where we measured large and small blood vessel problems in people with T2D+ASCVD in a Medicare population. This study also provides an update to previous studies that measured large and small blood vessel problems in people with T2D. By highlighting the frequency of large and small blood vessel problems in people with T2D, we hope to raise awareness about the importance of annual screening in people with T2D, especially people with existing ASCVD, and manage according to the established guidelines.


  • Open Access

    Recent trends in GLP-1 RA and SGLT2i use among people with type 2 diabetes and atherosclerotic cardiovascular disease in the USA

    by Aaron King, Xi Tan, Neil Dhopeshwarkar, Rhonda Bohn, Katherine Dea, Charles E Leonard, Adam de Havenon

    Introduction This study aimed to assess recent trends in the US use of glucagon-like peptide-1 receptor agonist (GLP-1 RA) and sodium-glucose cotransporter 2 inhibitor (SGLT2i) in people with type 2 diabetes (T2D) and atherosclerotic cardiovascular disease (ASCVD), including incident use following newly diagnosed ASCVD. Research design and methods This real-world, retrospective observational study used de-identified data from the TriNetX Dataworks–USA network. A longitudinal analysis of cross-sectional data (interval: January 01, 2018 to December 31, 2022) assessed the yearly prevalent use of GLP-1 RA and SGLT2i. A nested cohort study (January 01, 2017 to January 31, 2023) assessed the proportions of patients with T2D newly prescribed GLP-1 RAs and SGLT2is after incident ASCVD diagnosis. Results Prevalent use of GLP-1 RA and/or SGLT2i increased from 9.2% of patients in 2018 to 27.1% in 2022, with eligible annual patient numbers ranging from 279,474 to 348,997. GLP-1 RA-alone use rose from 5.2% to 9.9% and SGLT2i-alone use rose from 2.8% to 12.2% over this interval. Incident use of GLP-1 RA and/or SGLT2i within the year following ASCVD diagnosis increased from 5.9% to 17.0% (2018–2022). For GLP-1 RA alone, this increase was from 3.6% to 7.8%, while for SGLT2i alone, it was from 1.8% to 7.0%. Conclusions Use of GLP-1 RAs/SGLT2is in patients with T2D and ASCVD has increased in recent years in the USA, but remains suboptimal given the prevalence of ASCVD and its high morbidity and mortality.


  • Open Access

    Comparing Machine Learning and Advanced Methods with Traditional Methods to Generate Weights in Inverse Probability of Treatment Weighting: The INFORM Study

    by Doyoung Kwak, Yuanjie Liang, Xu Shi, Xi Tan

    Purpose: Observational research provides valuable insights into treatments used in patient populations in real-world settings. However, confounding is likely to occur if there are differences in patient characteristics associated with both the exposure and outcome between the groups being evaluated. One approach to reduce confounding and facilitate unbiased comparisons is inverse probability of treatment weighting (IPTW) using propensity scores. Machine learning (ML) and entropy balancing can potentially be used in generating propensity scores for IPTW, but there is limited literature on this application. We aimed to assess the feasibility of applying these methods for reducing confounding in observational studies. These methods were assessed in a study comparing cardiovascular outcomes in adults with type 2 diabetes and established atherosclerotic cardiovascular disease taking once-weekly glucagon-like peptide-1 receptor agonists or dipeptidyl peptidase-4 inhibitors. Methods: We applied advanced methods to generate the propensity scores compared to the original logistic regression method in terms of covariate balance. After calculating weights, a weighted Cox proportional hazards model was used to calculate the sample average treatment effect. Support Vector Classification, Support Vector Regression, XGBoost, and LightGBM were the ML models used. Entropy balancing was also performed on features identified in the original cardiovascular outcomes study. Results: Accuracy (range: 0.71 to 0.73), area under the curve (0.77 to 0.79), precision (0.53 to 0.60), recall (0.66 to 0.68), and F1 score (0.60 to 0.64) were similar between all of the advanced propensity score methods and traditional logistic regression. Among ML models, only XGBoost achieved balance in all measured baseline characteristics between the two treatment groups, closely approximating the performance of the original logistic regression. Entropy balancing weights provided the best performance among all models in balancing baseline characteristics, achieving near perfect balancing. Conclusion: Among the advanced methods examined, entropy balancing weights performed the best for optimizing balancing and can produce similar results compared to traditional logistic regression.