Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (16 results)

  • Article
  • Peer Reviewed
UC Riverside Previously Published Works (2017)
    Cover page: Unified Inference for Sparse and Dense Longitudinal Data in Time‐varying Coefficient Models
    • Thesis
    • Peer Reviewed
    UC Santa Barbara Electronic Theses and Dissertations (2024)

    This dissertation contains three chapters in labor economics and applied microeconomics.

    In Chapter 1, I provide the first empirical evidence on the effect of having a female officer in the primary unit on the discovery of domestic violence (DV) in physical abuse incidents. DV is a crime that usually has female victims and is often under-reported. This chapter studies the effect of having a female officer dispatched in the primary unit on the discovery of DV in physical abuse incidents in Milwaukee and Chicago. Using three-year calls for service data and conditional random assignment of officers in the dispatch process, the chapter finds that the existence of a female officer in the primary unit scales up the likelihood of discovering DV in physical abuse incidents by 10% in Milwaukee. Analysis of data from Chicago indicates a similar effect. These results indicate that female officers play an important role in discovering DV.

    In Chapter 2, we study the effect of college basketball game days on sexual assault. Basketball games are an important part of college identity and social activities. This chapter studies the effect of college basketball game days on the probability of having local sexual assault reports. Using crime data from universities with top basketball programs and local law enforcement agencies, this chapter shows that home game days have little effect on the probability of sexual assault reports, while away game days scale up the probability by 14%. This finding is different from those found for football, which likely reflects differences in viewing and partying behavior across the two sports.

    In Chapter 3, I study the relationship between school lunch and nutrition. School lunch is an important channel of students' nutrition intake. This chapter studies the potential mechanisms for effects of the Healthy, Hunger-Free Kids Act using metabolic analysis. Using data from the National Health and Nutrition Examination Survey, the chapter finds that this policy decreased the probability of having high total cholesterol by about 30% for students who usually eat school lunches. This result is largely driven by the significant decrease in the proportion of students with high LDL cholesterol and triglycerides. The results reported in the chapter show the powerful impact of this policy on adolescent health.

      Cover page: Essays in Labor Economics and Applied Microeconomics
      • Thesis
      • Peer Reviewed
      UCLA Electronic Theses and Dissertations (2022)

      Humans, even young infants, are adept at perceiving and understanding complex indoor scenes and events. Holistic scene understanding involves abundant aspects, including 3D human pose, objects, physical relations, functionality, etc. Besides the physical and functional configuration of the scene, interpreting human actions and goal-oriented tasks is a higher-level goal, and requires reasoning about the complex structures in activities along the temporal dimension. When multiple people are in the scene, collaborations and communications inevitably happen, in both verbal and non-verbal forms. Despite the recent remarkable progress in artificial intelligence, building an intelligent machine with human-like perception and reasoning capability for the aforementioned complex tasks remains a significant and challenging problem.In this dissertation, we study the holistic scene understanding and goal-directed multi-agent event parsing by identifying the critical problems from various perspectives. We first propose a framework for holistic 3D scene parsing and human pose estimation, with a particular focus on human-object interaction and physical commonsense reasoning. Contact information is critical in modeling the fine-grained human-object relations from visual cues. We demonstrate how to extract meaningful contact information from 2D images and its usefulness in 3D human pose estimation. Then we introduce our efforts in understanding goal-directed actions, concurrent multi-tasks, and collaborations among multi-agents. Finally, we investigate the two typical types of human communications by proposing a spatial and temporal model for shared attention and examining the power of both language and gesture under the embodied reference setting.

        Cover page: Holistic Scene Understanding and Goal-directed Multi-agent Event Parsing
        • Article
        • Peer Reviewed
        UC Riverside Previously Published Works (2015)
        In this article, a novel adaptive estimation is proposed for varying coefficient models. Unlike the traditional least squares based methods, the proposed approach can adapt to different error distributions. An efficient EM algorithm is provided to implement the proposed estimation. The asymptotic properties of the resulting estimator are established. Both simulation studies and real data examples are used to illustrate the finite sample performance of the new estimation procedure. The numerical results show that the gain of the new procedure over the least squares estimation can be quite substantial for non-Gaussian errors.
          Cover page: Adaptive estimation for varying coefficient models
          • Article
          • Peer Reviewed
          Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 44 (2022)
          Decision makers face a nontrivial problem when evaluating how much time to invest in an uncertain future prospect. Un- conditional persistence is not always advantageous; rather, different levels of persistence are favored in environments with different temporal statistics. Previous studies using foraging- like decision-making tasks have found that people can rapidly recalibrate their persistence behavior—becoming either more or less willing to tolerate delay—after a short period of direct experience with the temporal statistics of a new environment. Furthermore, substantial individual variation is apparent both in baseline levels of persistence and in the flexibility of re- calibration across environments. However, it is unknown to what degree such variation reflects trait-like individual differences in contrast to session-specific measurement noise. Here we investigated the test-retest reliability of individual variation in behavioral persistence in a computerized decision-making task. We conducted an online experiment in which participants (n=141 after exclusions) performed the task on two occasions separated by a three-week interval. We evaluated the test- retest reliability of several behavior-derived indices, including: a descriptive estimate of overall willingness to wait, a contrast measure reflecting flexibility of recalibration across environments, and individual-level parameter estimates derived from a reinforcement learning model of adaptive persistence. The results showed strong evidence for stable, trait-like individual variation in multiple aspects of persistence-related decision- making behavior. Our findings establish a foundation for future investigations of associations between task-derived parameters of decision behavior and other cognitive and motivational traits.
            Cover page: Test-retest reliability of task-based measures of voluntary persistenceCreative Commons 'BY' version 4.0 license
            • Thesis
            • Peer Reviewed
            UCLA Electronic Theses and Dissertations (2024)

            Childhood cancer, defined as cancer diagnoses in children aged 0 to 19 years, is the second leading cause of death in children in the Unites States and Europe. The etiology of childhood cancer remains unclear. Established risk factors include genetic predispositions, high dose ionizing radiation, and prior chemotherapy. This dissertation examines three types of environmental exposures in relation to childhood cancer: 1) neonatal exposure to PFAS and the risk of retinoblastoma; 2) maternal residential proximity to oil and gas development during pregnancy and the risk of childhood cancer; and 3) parental occupational exposures to organic solvents and the risk of childhood cancer. The first study utilized 501 retinoblastoma cases and 899 controls born 1983-2011 from a population-based case-control study. We identified PFOS, PFOA, and PFNA feature intensities in neonatal DBS from California GDSP. Among all children, above-mean PFOS levels increased the risk of retinoblastoma overall (aOR: 1.29; 95% CI: 1.00, 1.67) and unilateral retinoblastoma (aOR: 1.42; 95% CI: 1.03,1.97). In children of Mexico-born mothers, above-mean PFOS levels increased the risk of retinoblastoma overall (aOR: 1.67; 95% CI: 1.06, 2.66) and bilateral retinoblastoma (aOR: 2.06; 95% CI: 1.12, 3.92). In children of US-born mothers, the risk of unilateral retinoblastoma increased by 15% for each IQR increase in PFOS (95% CI: 0.99, 1.35) and by 71% (95% CI: 1.04, 2.90) for above-mean PFOS levels. PFOA increased the risk of retinoblastoma overall (aOR: 1.41; 95% CI: 1.00, 2.02 for above-mean levels, aOR: 1.06; 95% CI: 0.98, 1.16 per IQR increase) for children of US-born mothers. The second study utilized the same population-based case-control study in California with children born 1998-2016. We estimated exposure during pregnancy to any well (i.e., active or inactive) and inactive wells located within 1km and 3km of residences and assessed low and high exposures (categorized by median) within this proximity. The risk of medulloblastoma increased among children of mothers exposed to any well within 3km (≥1 well: adjusted Odds Ratio (aOR): 1.40; 95% confidence interval (CI): 1.08, 1.82; low, aOR = 1.27; 95% CI = 0.93, 1.74; high, aOR = 1.55; 95% CI = 1.14, 2.11). The risk of rhabdomyosarcoma consistently increased across the exposure levels of any well within 3km (≥1 well: aOR = 1.64; 95% CI = 1.28, 2.09; low, aOR = 1.69; 95% CI = 1.28, 2.23; high, aOR = 1.57; 95% CI = 1.16, 2.12). Similar associations were observed for teratoma with any well within 3km (≥1 well: aOR = 1.42; 95% CI = 0.99, 2.05; low, aOR = 1.35; 95% CI = 0.88, 2.07; high, aOR = 1.51; 95% CI = 0.98, 2.33). Overall, the results were consistent across the buffer sizes and no differences were found between any well and inactive wells. The third study utilized a linkage of four Danish data registries, including 10442 cases and 261050 controls born 1968-2013. Using JEMs, we identified and examined the role of parental occupational exposure to 3 hydrocarbon groups (i.e., aromatic, aliphatic/alicyclic, chlorinated hydrocarbons) and 4 individual hydrocarbons (i.e., toluene, DCM, TCE, 1,1,1-TCA) in two biologically relevant time periods: 1) 3 months preconception to birth for fathers; and 2) during pregnancy for mothers and the risk of childhood cancer in offspring. Paternal exposure to aromatic (high: aOR: 1.31; 95% CI: 0.95, 1.81) and chlorinated hydrocarbons (high: aOR: 1.82; 95% CI: 1.18, 2.82) increased the risk of glioma. The risk of AML was higher among fathers highly exposed to aliphatic/alicyclic hydrocarbons (high: aOR: 1.87; 95% CI: 1.24, 2.82). Paternal exposure to1,1,1-trichloroethane increased the risk of osteosarcoma (high: aOR: 2.29; 95% CI: 1.25, 4.18). The risk of medulloblastoma increased in offspring of mothers ever exposed to DCM (aOR: 1.69; 95% CI: 1.01, 2.84), TCE (aOR; 1.86; 95% CI: 1.12, 3.10), and toluene (aOR; 1.92; 95% CI: 1.07, 3.46). The risk of ALL increased in children of mothers ever exposed to aromatic hydrocarbons (aOR; 1.29; 95% CI: 1.01, 1.64), and 1,1,1-TCA (aOR; 1.30; 95% CI: 1.01, 1.69), while risk of NHL was associated with maternal exposure to chlorinated hydrocarbons (aOR; 1.94; 95% CI: 1.05, 3.57).

              • Article
              • Peer Reviewed
              UCLA Previously Published Works (2024)

              Background

              Neonatal per- and polyfluoroalkyl substance (PFAS) exposure can disrupt hormonal homeostasis and induce neuro- and immunotoxicity in children. In this exploratory study, we investigated associations between PFAS levels in neonatal dried blood spots and retinoblastoma risk.

              Materials and methods

              This study included 501 retinoblastoma cases born from 1983 to 2011 and 899 controls frequency-matched by birth year (20:1 matching ratio), born to 755 US-born and 366 Mexico-born mothers in California. Perfluorooctanesulfonic acid (PFOS), perflurooctanoic acid (PFOA), and perfluorononanoic acid (PFNA) feature intensities were identified from neonatal blood spots from California newborn Genetic Disease Screening Program. Using logistic regression, we assessed whether an interquartile range (IQR) increase of PFAS levels or having above-mean levels of PFAS in blood affects retinoblastoma risk overall or its subtypes (i.e., unilateral, bilateral). We assessed children of US-born and Mexico-born mothers, separately.

              Results and discussion

              Among all children, above-mean PFOS levels at birth increased the odds of retinoblastoma overall by 29% (95% Confidence Interval (CI): 1.00, 1.67) and unilateral retinoblastoma by 42% (95% CI: 1.03, 1.97). For children of Mexico-born mothers, we estimated the highest odds of retinoblastoma overall (adjusted odds ratio (aOR): 1.67; 95% CI: 1.06, 2.66) and bilateral retinoblastoma (aOR: 2.06; 95% CI: 1.12, 3.92) with above-mean PFOS levels. Among children of US-born mothers, higher PFOS levels increased the odds of unilateral retinoblastoma by 15% (95% CI: 0.99, 1.35) for each IQR increase and by 71% among children with above-mean PFOS levels (95% CI: 1.04, 2.90). In addition, for children of US-born mothers, PFOA increased the odds of retinoblastoma overall (aOR: 1.41; 95% CI: 1.00, 2.02 for above-mean levels, aOR: 1.06; 95% CI: 0.98, 1.16 per IQR increase). PFNA was not associated with retinoblastoma risk.

              Conclusions

              Our results suggested that PFOS and PFOA might contribute to retinoblastoma risk in children born in California.
                • Article
                • Peer Reviewed
                UCLA Previously Published Works (2024)

                Background

                Maternal solvent exposure has been suspected to increase offspring cancer risk. The study aimed to evaluate the associations between maternal residential exposure to solvents from industrial pollution during pregnancy and childhood cancer.

                Methods

                The present study included 15,744 cancer cases (aged 0-19 years at diagnosis) identified from California Cancer Registry and 283,141 controls randomly selected from California Birth Registry (20:1 frequency-matched by birth year: 1998-2016). We examined industrial releases of tetrachloroethylene and 1,1,1-trichloroethane within 3 km of the birth address, while we used a 5 km buffer for carbon disulfide. We calculated the total exposure from all linked Toxic Release Inventory sites during each index pregnancy and assigned "ever/never" and "high/low exposed/unexposed" exposure, using median values. We performed quadratic decay models to estimate cancer risks associated with maternal solvent exposure in pregnancy.

                Results

                1,1,1-Trichloroethane was associated with rhabdomyosarcoma (adjusted Odds Ratio (aOR): 1.96; 95% Confidence Interval (CI): 1.16, 3.32) in the "ever exposed" group. Ever exposure to carbon disulfide was associated with increased risks of medulloblastoma (OR = 1.85, 95% CI 1.01, 3.40) and ependymoma (OR = 1.63, 95% CI 0.97, 2.74).

                Conclusions

                Overall, our findings suggested maternal residential exposure to solvents from industrial sources might be associated with elevated childhood cancer risks.
                  Cover page: Maternal residential exposure to solvents from industrial sources during pregnancy and childhood cancer risk in California
                  • Article
                  • Peer Reviewed
                  UC Davis Previously Published Works (2023)
                  Cellular senescence is a state of proliferative arrest, and the development of carcinoma can be suppressed by conferring tumor cell senescence. Recently, we found that carnitine palmitoyltransferase 1C (CPT1C) controls tumor cell proliferation and senescence via regulating lipid metabolism and mitochondrial function. Here, 13C-metabolic flux analysis (13C-MFA) was performed and the results revealed that CPT1C knockdown in MDA-MB-231 cells significantly induced cellular senescence accompanied by altered fatty acid metabolism. Strikingly, stearate synthesis was decreased while oleate was increased. Furthermore, stearate significantly inhibited proliferation while oleate reversed the senescent phenotype induced by silencing CPT1C in MDA-MB-231 cells as well as PANC-1 cells. A939572, an inhibitor of stearoyl-Coenzyme A desaturase 1, had the same effect as stearate to inhibit cellular proliferation. These results demonstrated that stearate and oleate are involved in CPT1C-mediated tumor cellular senescence, and the regulation of stearate/oleate rate via inhibition of SCD-1 could be an additional strategy with depletion of CPT1C for cancer therapy.
                    • Article
                    • Peer Reviewed
                    UC San Francisco Previously Published Works (2022)
                    Our objective was to develop deep learning models with chest radiograph data to predict healthcare costs and classify top-50% spenders. 21,872 frontal chest radiographs were retrospectively collected from 19,524 patients with at least 1-year spending data. Among the patients, 11,003 patients had 3 years of cost data, and 1678 patients had 5 years of cost data. Model performances were measured with area under the receiver operating characteristic curve (ROC-AUC) for classification of top-50% spenders and Spearman ρ for prediction of healthcare cost. The best model predicting 1-year (N = 21,872) expenditure achieved ROC-AUC of 0.806 [95% CI 0.793-0.819] for top-50% spender classification and ρ of 0.561 [0.536-0.586] for regression. Similarly, for predicting 3-year (N = 12,395) expenditure, ROC-AUC of 0.771 [0.750-0.794] and ρ of 0.524 [0.489-0.559]; for predicting 5-year (N = 1779) expenditure ROC-AUC of 0.729 [0.667-0.729] and ρ of 0.424 [0.324-0.529]. Our deep learning model demonstrated the feasibility of predicting health care expenditure as well as classifying top 50% healthcare spenders at 1, 3, and 5 year(s), implying the feasibility of combining deep learning with information-rich imaging data to uncover hidden associations that may allude to physicians. Such a model can be a starting point of making an accurate budget in reimbursement models in healthcare industries.
                      Cover page: Prediction of future healthcare expenses of patients from chest radiographs using deep learning: a pilot study
                      Top