Background: Mobile health technologies show promise in addressing metabolic syndrome, but their comparative effectiveness in large-scale public health interventions remains unclear. Objective: This study aims to compare the effectiveness of wearable devices (wearable activity trackers) and mobile app–based activity trackers (built-in step counters) in promoting walking practice, improving health behaviors, and reducing metabolic syndrome risk within a national mobile health care program operated by the Korea Health Promotion Institute. Methods: This retrospective cohort study analyzed data from 46,579 participants in South Korea’s national mobile health care program (2020‐2022). Participants used wearable devices for 12 weeks, after which some switched to built-in step counters. The study collected data on demographics, health behaviors, and metabolic syndrome risk factors at baseline, 12 weeks, and 24 weeks. Outcomes included changes in walking practice, health behaviors, and metabolic syndrome risk factors. Metabolic syndrome risk was assessed based on 5 factors: blood pressure, fasting glucose, waist circumference, triglycerides, and high-density lipoprotein cholesterol. Health behaviors included low-sodium diet preference, nutrition label reading, regular breakfast consumption, aerobic physical activity, and regular walking. To address potential selection bias, propensity score matching was performed, balancing the 2 groups on baseline characteristics including age, gender, education level, occupation, insurance type, smoking status, and alcohol consumption. Results: Both wearable activity tracker and built-in step counter groups exhibited significant improvements across all evaluated outcomes. The improvement rates for regular walking practice, health behavior changes, and metabolic syndrome risk reduction were high in both groups, with percentages ranging from 45.2% to 60.8%. After propensity score matching, both device types showed substantial improvements across all indicators. The built-in step counter group demonstrated greater reductions in metabolic syndrome risk compared to the wearable device group (odds ratio [OR] 1.20, 95% CI 1.05‐1.36). No significant differences were found in overall health behavior improvements (OR 0.95, 95% CI 0.83‐1.09) or walking practice (OR 0.84, 95% CI 0.70‐1.01) between the 2 groups. Age-specific subgroup analyses revealed that the association between built-in step counters and metabolic syndrome risk reduction was more pronounced in young adults aged 19‐39 years (OR 1.35, 95% CI 1.09‐1.68). Among Android use subgroups, built-in step counters were associated with a higher reduction in health risk factors (OR 1.20, 95% CI 1.03‐1.39). Conclusions: Both wearable devices and built-in step counters effectively reduced metabolic syndrome risk in a large-scale public health intervention, with built-in step counters showing a slight advantage. The findings suggest that personalized device recommendations based on individual characteristics, such as age and specific health risk factors, may enhance the effectiveness of mobile health interventions. Future research should explore the mechanisms behind these differences and their long-term impacts on health outcomes.

Background: The demand for implementing a new listing scheme to expedite patient access to novel oncology drugs has increased in South Korea. This study was conducted to compare the prices of anticancer drugs between eight countries and to explore the feasibility of a ‘pre-listing and post-evaluation’ scheme to expedite patient access to oncology drugs. Methods: This study included 34 anticancer drugs, which were reimbursed between 1 January 2007 and 31 December 2017. The unit price and sales volume of the study drugs were collected from eight countries and IQVIA data, respectively. The prices were adjusted to estimate the ex-factory prices using the discount/rebate rate suggested by the Health Insurance Review Agency (HIRA). The four price indices of Laspeyres, Paasche, Fisher, and the unweighted index were calculated using the price in each country, the average price, and lowest price among the study countries. Each currency was converted using the currency exchange rate and purchasing power parity (PPP). The budget impact of implementing the proposed pre-listing and post-evaluation scheme on payers was calculated. Results: Based on the currency exchange rate, anticancer drug prices were higher in other countries (index range: 1.05–2.78) compared to Korea. The prices in Korea were similar to countries with the lowest prices. When the PPP was applied, prices were higher in the US, Germany, Italy, and Japan than in Korea (range: 1.10–2.13); however, the prices were lower in the UK, France, and Switzerland than in Korea (range: 0.72–0.99). The financial burden of implementing the pre-listing and post-evaluation scheme was calculated at 0.83% of the total anticancer drug sales value in Korea from 2013–2017. Conclusions: The prices of anticancer drugs in Korea were similar to the lowest prices among the seven other study countries. A pre-listing and post-evaluation scheme should be considered to improve patient access to novel anticancer drugs by reducing the reimbursement review time and uncertainties.

Study objectives: While zolpidem is considered as an example of a gender effect on drug response, there is insufficient evidence to reach a consensus. This study aimed to investigate gender differences in adverse events (AEs) of zolpidem.Methods: We estimated the difference between the reporting odds ratios (RORs) calculated in gender subgroups for the AEs signals detected in data mining using 2015–2019 Korea voluntary adverse drug events reporting system (KAERS) data. Different reporting risk by gender was evaluated by using the log RORs being significantly different by gender at the 5% significance level and the 95% confidence intervals of the gender ROR.Results: A total of 94 AE signals were detected. Among these, 35 signals showed significant disparities by gender at the 5% level or were detected only in one gender. When categorized by similarity of AEs, parasomnia including somnambulism and paroniria, and cardiovascular disorders including coronary thrombosis had higher reporting risks in women. Men were more likely to report cognitive disorders such as delirium, insomnia related disorders, and movement disorders. Among all AEs with gender differences in reporting risk, the difference in somnambulism was the most consistent and substantial.Conclusion: For several AEs associated with zolpidem, gender-based reporting disparities were evident. Notably, women exhibited a higher susbeptibility to somnambulism, potentially serious adverse effects of zolpidem. This underscores the need for further investigation into the underlying factors influencing these gender-specific reporting patterns.

Background: Poor clopidogrel metabolizers, carrying a cytochrome P450 2C19 loss-of-function allele, are more frequent among East Asians than Caucasians/White. Materials & methods: The Korea adverse event reporting system database and a case/noncase study design were used to examine the disproportionality of cardiovascular events following clopidogrel use. The US FDA's adverse event reporting system database was also analyzed for comparison. Results: In the Korea adverse event reporting system data, the clopidogrel reporting odds ratio for cardiovascular events was 7.34, more than double that of ticagrelor. In the FDA's adverse event reporting system data, the clopidogrel reporting odds ratio was 4.69, lower than that of ticagrelor. Adjustment for covariates did not change the trend. Conclusion: Considering the prevalence of poor clopidogrel metabolizers and the reported cardiovascular events among Koreans, rigorous clinical management is required for clopidogrel users.

Background: This study examines gastrointestinal adverse events (GIAEs) associated with tirzepatide using bibliometric and pharmacovigilance analyses. Research design and methods: A bibliometric analysis of Web of Science data identified research trends in tirzepatide-related adverse events, while a pharmacovigilance analysis of FAERS data (Q2 2022–Q2 2024) assessed real-world GIAEs patterns. Disproportionality, time-to-onset, univariate, and comparative analyses were conducted to evaluate reporting odds ratios (RORs), onset timing, and subgroup differences. Results: Among 110 studies, cardiovascular outcomes predominated as the research focus. FAERS data showed that nausea (27.7%) and diarrhea (12.8%) were the most frequently reported events, whereas eructation and impaired gastric emptying had the highest disproportionality. GIAEs were more common in older adults, males, and patients receiving concomitant medications, and most occurred within 3 months (median onset: 16 days). Tirzepatide had a lower ROR for GIAEs than GLP-1 receptor agonists but a higher ROR than non-GLP-1 drugs, with a greater risk in patients with type 2 diabetes (T2DM) than in those using tirzepatide for weight loss. Conclusion: Tirzepatide is associated with an increased risk of GIAEs, particularly among patients with T2DM, males, older adults, and those using concomitant medications. FAERS-based real-world evidence complements clinical trial data and highlights the need for individualized patient monitoring and management strategies.

Sparse autoencoders (SAEs) decompose transformer activations into sparse feature dictionaries, but one-shot support selection can leave many features unused and become difficult to optimize at very small active budgets. We study an SAE-like dictionary-learning procedure that combines a learned TopK encoder with a few unrolled Fast Iterative Shrinkage-Thresholding Algorithm (FISTA) steps for coefficient refinement. On GPT-2 Small layer-11 residual streams, a 12,288-feature TopK256 model with five FISTA steps achieves ΔCE = 0.0745 and a 3.09% dead-feature ratio, compared with 0.0859 and 50.76% for a matched TopK SAE. Ablations show that initialization and learned TopK feature selection explain most of the reduction in dead features, while FISTA further improves fidelity. In a 32,768-feature regime, direct exact-TopK32 training is unstable. Training becomes feasible by retaining a TopK128 candidate support while reducing the final active budget from TopK128 to TopK64 and then TopK32. Five refinement steps reach ΔCE = 0.2442 and 17.84% dead features after 264 million token presentations. A residual-gain-ranked loss over candidates at ranks 33–48 improves near-boundary feature recruitment with a modest fidelity cost. These results show that lightweight iterative correction can improve fixed-budget sparse decomposition without replacing one-shot SAE inference with a full optimization solver.