Serge Aleshin‐Guendel

ORCID: 0000-0001-5906-5667
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Census and Population Estimation
  • COVID-19 and healthcare impacts
  • Data Quality and Management
  • Data-Driven Disease Surveillance
  • COVID-19 epidemiological studies
  • Health and Conflict Studies
  • Prostate Cancer Diagnosis and Treatment
  • Prostate Cancer Treatment and Research
  • Privacy-Preserving Technologies in Data
  • Insurance, Mortality, Demography, Risk Management
  • Climate Change and Health Impacts
  • Cardiac Arrest and Resuscitation
  • Cancer, Lipids, and Metabolism
  • Opioid Use Disorder Treatment
  • Bayesian Methods and Mixture Models
  • Statistical Methods and Bayesian Inference
  • Pain Management and Opioid Use
  • Web Data Mining and Analysis
  • Global Cancer Incidence and Screening
  • Data Mining Algorithms and Applications
  • Data Management and Algorithms
  • Colorectal Cancer Screening and Detection
  • Forensic and Genetic Research
  • Agricultural risk and resilience
  • Distributed systems and fault tolerance

University of Washington
2020-2024

United States Census Bureau
2024

Statistical Research (United States)
2024

Duke University
2024

Fred Hutch Cancer Center
2018

The World Health Organization has a mandate to compile and disseminate statistics on mortality, we have been tracking the progression of COVID-19 pandemic since beginning 20201. Reported mortality are problematic for many countries owing variations in testing access, differential diagnostic capacity inconsistent certification as cause death. Beyond what is directly attributable it, caused extensive collateral damage that led losses lives livelihoods. Here report comprehensive consistent...

10.1038/s41586-022-05522-2 article EN cc-by Nature 2022-12-14

Estimating the true mortality burden of COVID-19 for every country in world is a difficult, but crucial, public health endeavor. Attributing deaths, direct or indirect, to problematic. A more attainable target "excess deaths," number deaths particular period, relative that expected during "normal times," and we develop model this The excess requires two numbers, total former unavailable many countries, so modeling required such countries. are based on historic data, producing estimates these...

10.1214/22-aoas1673 article EN The Annals of Applied Statistics 2023-05-01

ABSTRACT The problem of estimating the size a population based on subset individuals observed across multiple data sources is often referred to as capture-recapture or multiple-systems estimation. This fundamentally missing problem, where number unobserved represents data. As with any estimation requires users make an untestable identifying assumption in order estimate from If appropriate cannot be found for set, no should produced that models different assumptions can produce arbitrarily...

10.1093/biomtc/ujad028 article EN Biometrics 2024-01-29

Merging datafiles containing information on overlapping sets of entities is a challenging task in the absence unique identifiers, and further complicated when some are duplicated datafiles. Most approaches to this problem have focused linking two files assumed be free duplicates, or detecting which records single file duplicates. However, it common practice encounter scenarios that fit somewhere between beyond these settings. We propose Bayesian approach for general setting multifile record...

10.1080/01621459.2021.2013242 article EN Journal of the American Statistical Association 2021-12-07

Abstract As a part of its mandate to compile and disseminate statistics on mortality, the World Health Organization (WHO) has been tracking progression COVID-19 pandemic since beginning 2020. However, reported are problematic for number countries due variations in testing access, differential diagnostic capacity inconsistencies applications standards correctly certify as cause-of-death. In addition, caused extensive collateral damage beyond what is directly attributable it. Consequently, WHO...

10.21203/rs.3.rs-1673759/v1 preprint EN cc-by Research Square (Research Square) 2022-07-06

In studies of cancer risk, detection bias arises when risk factors are associated with screening patterns, affecting the likelihood and timing diagnosis. To eliminate in a screened cohort, we propose modeling latent onset estimating association between rather than We apply this framework to estimate increase prostate black race family history using data from SELECT prevention trial, which men were biopsied according community practices. A positive was hazard ratio (HR) 1.8, lower...

10.1177/0163278720984203 article EN Evaluation & the Health Professions 2021-01-28

207 Background: Optimal utilization of novel therapies for advanced prostate cancer is challenging without a validated surrogate efficacy endpoint. Ongoing trials are using durable undetectable specific antigen (PSA) levels as marker efficacy. The proportion patients and clinical relevance those with prolonged PSA after short course androgen deprivation therapy (ADT) uncertain. Methods: University Washington Caisis database was queried radical prostatectomy who received 6–12 months ADT...

10.1200/jco.2018.36.6_suppl.207 article EN Journal of Clinical Oncology 2018-02-20

Estimating the true mortality burden of COVID-19 for every country in world is a difficult, but crucial, public health endeavor. Attributing deaths, direct or indirect, to problematic. A more attainable target "excess deaths", number deaths particular period, relative that expected during "normal times", and we estimate this all countries on monthly time scale 2020 2021. The excess requires two numbers, total former unavailable many countries, so modeling required these countries. are based...

10.48550/arxiv.2205.09081 preprint EN other-oa arXiv (Cornell University) 2022-01-01

Summary We organize the discussants’ major comments into following categories: sensitivity analyses, zero counts, model selection, marginal no-highest-order interaction (NHOI) assumption, and usefulness of our proposed framework.

10.1093/biomtc/ujad033 article EN Biometrics 2024-01-29

Abstract The under-5 mortality rate (U5MR), a critical health indicator, is typically estimated from household surveys in lower and middle income countries. Spatio-temporal disaggregation of survey data can lead to highly variable estimates U5MR, necessitating the usage smoothing models which borrow information across space time. assumptions common may be unrealistic when certain time periods or regions are expected have shocks relative their neighbors, oversmoothing U5MR estimates. In this...

10.1093/biostatistics/kxae030 article EN public-domain Biostatistics 2024-08-05

Entity resolution is the process of merging and removing duplicate records from multiple data sources, often in absence unique identifiers. Bayesian models for entity allow one to include a priori information, quantify uncertainty important applications, directly estimate partition records. Markov chain Monte Carlo (MCMC) sampling primary computational method approximate posterior inference this setting, but due high dimensionality space partitions, there are no agreed upon standards...

10.1146/annurev-statistics-040522-114848 article EN other-oa Annual Review of Statistics and Its Application 2024-04-22

Background Optimal utilization of novel therapies for advanced prostate cancer is challenging without a validated surrogate efficacy endpoint. Ongoing trials are using durable undetectable prostate‐specific antigen (PSA) levels as marker efficacy. The clinical relevance prolonged PSA after short course androgen deprivation therapy (ADT) uncertain. Methods University Washington Caisis database was queried radical prostatectomy patients who received 6‐12 months ADT biochemical recurrence...

10.1002/pros.23666 article EN The Prostate 2018-07-10

Latent class models have recently become popular for multiple-systems estimation in human rights applications. However, it is currently unknown when a given family of latent identifiable this context. We provide necessary and sufficient conditions on the number classes needed to be identifiable. Along way we mechanism verifying identifiability that allow individual heterogeneity.

10.48550/arxiv.2008.09865 preprint EN other-oa arXiv (Cornell University) 2020-01-01

The under-5 mortality rate (U5MR), a critical health indicator, is typically estimated from household surveys in lower and middle income countries. Spatio-temporal disaggregation of survey data can lead to highly variable estimates U5MR, necessitating the usage smoothing models which borrow information across space time. assumptions common may be unrealistic when certain time periods or regions are expected have shocks relative their neighbors, oversmoothing U5MR estimates. In this paper, we...

10.48550/arxiv.2309.00724 preprint EN other-oa arXiv (Cornell University) 2023-01-01

e17073 Background: Optimal utilization of novel therapies for advanced prostate cancer is challenging without a validated surrogate efficacy endpoint. Durable undetectable prostate-specific antigen (PSA) levels are being used in ongoing trials as marker efficacy. The clinical relevance patients with prolonged PSA after short course androgen deprivation therapy (ADT) uncertain. Methods: University Washington Caisis database was queried radical prostatectomy who received 6–12 months ADT...

10.1200/jco.2018.36.15_suppl.e17073 article EN Journal of Clinical Oncology 2018-05-20

Merging datafiles containing information on overlapping sets of entities is a challenging task in the absence unique identifiers, and further complicated when some are duplicated datafiles. Most approaches to this problem have focused linking two files assumed be free duplicates, or detecting which records single file duplicates. However, it common practice encounter scenarios that fit somewhere between beyond these settings. We propose Bayesian approach for general setting multifile record...

10.48550/arxiv.2110.03839 preprint EN other-oa arXiv (Cornell University) 2021-01-01

The problem of estimating the size a population based on subset individuals observed across multiple data sources is often referred to as capture-recapture or multiple-systems estimation. This fundamentally missing problem, where number unobserved represents data. As with any estimation requires users make an untestable identifying assumption in order estimate from If appropriate cannot be found for set, no should produced that models different assumptions can produce arbitrarily estimates...

10.48550/arxiv.2101.09304 preprint EN other-oa arXiv (Cornell University) 2021-01-01
Coming Soon ...