武蔵大学

Objectives:We reviewed studies examining effects of non-minimum wages on health using causal inference or longitudinal data in high-income countries. We excluded studies on direct effects of minimum wages and on analyses using cross-sectional data without causal designs. Methods:Our review covered studies from public health, epidemiology, social sciences, and statistics, and published between 1974 and November 2025. Searches were conducted in Google Scholar and PubMed and supplemented by reference and citation tracing. We defined short-run (≤2 years) and long-run (≥5 years). Results:Thirty-eight studies met inclusion criteria: 20 short-run causal analyses, four short-run longitudinal studies, 12 long-run studies using causal or longitudinal methods, and two encompassing both timeframes. Instrumental variable models were most common, although many instruments (e.g., education and work experience) were invalid. Across 20+ health outcomes-most frequently self-rated health, mortality, and work limitations-results were heterogeneous. We identified recurrent methodological limitations and highlight priorities for future research. Conclusions:Although findings lack consensus, most studies, particularly long-run analyses, report that lower wages are associated with poorer health, consistent with the allostatic load and Friedman's permanent-income hypotheses. These findings carry implications for minimum-wage-and-health studies in which null findings predominate because those are only short run.

Safety-critical systems that integrate machine learning (ML) components have gained significant traction in recent years. However, current safety analysis methods lack a systematic approach to address the uncertainty inherent in ML system decision-making. Consequently, performing thorough risk analysis for ML systems with these methods is often time-consuming and makes it difficult to identify risks associated with uncertainty. In this study, we introduce a novel risk analysis method called eAI-Risk, designed for complex ML systems comprising multiple ML components. This method employs a data-driven and scenario-based approach, providing a structured framework for systematically identifying factors that contribute to high risks in ML systems and for evaluating potential countermeasures. A key feature of eAI-Risk is its capacity to identify risks within ML systems based on the classification patterns of the ML components and the action patterns defined by the system, while considering the uncertainty that characterizes ML systems. In addition, risks are quantitatively assessed from two perspectives: severity and likelihood. The product of these values generates a risk score, facilitating the prioritization of risks to address. This study illustrates the effectiveness of eAI-Risk by applying it to an automated driving system. Through the eAI-Risk process cycle, we anticipate that optimal risk mitigation strategies can be effectively identified and implemented.

IntroductionLipid nanoparticles (LNPs) are widely used as drug delivery systems (DDS) for the intracellular delivery of gene-editing tools such as the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) system. Previously, a lysine-headgroup cationic lipid (K3C16) was reported to exhibit high biocompatibility and cellular internalization via plasma membrane fusion when formulated as liposomes, suggesting its potential as a low-toxicity gene delivery material. However, the translation of this lipid into a multi-component lipid nanoparticle (LNP) matrix via microfluidic engineering for large cargo encapsulation has not been explored. In this study, K3C16 was successfully engineered for the first time as the main lipid component of LNPs to deliver large plasmid DNA encoding the CRISPR/Cas9 system.MethodsFirst, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) were evaluated as helper phospholipids to identify an optimal lipid composition. LNPs containing DOPE exhibited significantly higher transfection efficiency than those containing DOPC and were therefore selected for subsequent experiments. Next, LNPs formulated with different cationic or ionizable lipids were prepared and evaluated in HEK293 cells for transfection efficiency, cytotoxicity, and gene-editing efficiency.ResultsLNPs containing the commercial ionizable lipid SM-102 showed higher transfection and gene-editing efficiencies than the lysine-headgroup cationic lipids represent a promising and biocompatible platform for CRISPR/Cas9 plasmid delivery; however, increased cytotoxicity was observed in highly transfected cell populations. In contrast, the lysine-headgroup cationic LMPs achieved effective delivery of large plasmid DNA and CRISPR/Cas9-mediated gene editing while maintaining superior biocompatibility.ConclusionThese results demonstrate that lysine-headgroup cationic lipids represent a promising and biocompatible platform for CRISPR/Cas9 plasmid delivery. Further optimization of lipid composition and molar ratios may enhance transfection and gene-editing efficiencies, advancing the development of safer nonviral gene delivery systems.

This paper revisits long-term employment practices in Japan and provides new evidence on their evolution over 1982–2022, extending prior work to cover the most recent decades marked by the Great Recession and the COVID-19 pandemic. Using microdata from the Employment Status Survey, we estimate 10-year job retention rates by gender and educational attainment. Contrary to the widespread view that Japan’s long-term employment system has eroded since the 1990s, we find that job retention rates have remained highly stable over the past four decades. In particular, prime-age workers with at least 5 year of tenure (“core employees”) continue to exhibit consistently high retention rates, even during major economic shocks. Although job stability among young entry-level workers—especially college graduates—has declined significantly, cohort analysis shows that this early-career instability does not persist. Most workers eventually transition into stable, long-term employment relationships as they age. We also find no evidence of a decline in the overall share of core employees, although younger male cohorts exhibit temporary delays in accumulating long tenure. In addition, increases in mandatory retirement ages have contributed to higher job retention among older workers. Overall, our findings point to the resilience of Japan’s long-term employment system and suggest that policy discussions should be grounded in its continued empirical relevance.

This study addresses a fundamental challenge in the empirical application of the Capability Approach: the measurement of the “capability set” as an opportunity set. Unlike standard utility-based measures that focus solely on achieved outcomes, measuring capability requires assessing the welfare of potential activities—including those not chosen (counterfactuals). We propose a novel methodology that bridges normative social choice theory and econometric causal inference. Specifically, we interpret the Average Treatment Effect (ATE) derived from panel data fixed-effects models as capturing marginal counterfactual welfare differences between alternative actions, rather than level comparisons of chieved outcomes. Using a unique panel dataset of elderly individuals in Japan, focusing on “going-out” versus “staying-home” behavior, we evaluate the size of capability sets and the degree of “unfreedom” (the welfare gap between options). Furthermore, we propose and apply several aggregation rules—ranging from Utilitarian to Rawlsian—to construct group-level capability measures. Our empirical results demonstrate that the ranking of social groups varies significantly depending on the normative aggregation rule employed, highlighting the importance of explicitly defining the informational basis of social evaluation.