David van der Goes
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
David van der Goes has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Nicholas Edwardson
School of Public Administration, University of New Mexico, Albuquerque, NM, United States
2 shared publications - Alexandria D Viszolay
Department of Internal Medicine, School of Medicine, University of New Mexico Health Sciences Center, 1 University of New Mexico, MSC10 5550, Albuquerque, NM, 87131, United States, 1 505-272-9709
1 shared publication Christine Hollis
New Mexico Coalition for Health Care Value, Albuquerque, NM, USA
1 shared publication- Christophe Gerard Lambert
Department of Internal Medicine, University of New Mexico, Albuquerque, NM, USA
1 shared publication - Cristian G. Bologa
Department of Internal Medicine, School of Medicine, University of New Mexico Health Sciences Center, 1 University of New Mexico, MSC10 5550, Albuquerque, NM, 87131, United States, 1 505-272-9709
1 shared publication Dawn Hunter
Office of Policy and Accountability, New Mexico Department of Health, Santa Fe, NM, USA
1 shared publication
Latest Publications
- Open Access
- 4 June 2026
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 28 (1) , e89071
Detecting Uncoded Self-Harm in Veterans’ Electronic Health Records Using Positive and Unlabeled Learning: Retrospective Cohort Study
Background: Underdiagnosis and undercoding are common across mental health conditions, particularly suicide and self-harm. This leaves health care datasets lacking reliable negative examples needed for predictive modeling, phenotype prevalence estimation, and identification of individuals at elevated risk. We use positive and unlabeled (PU) learning to address this challenge. Objective: This study aims to identify US Veterans whose self-harm events were not explicitly captured through diagnostic codes in electronic health records (EHRs) and estimate the underlying prevalence using a novel PU learning algorithm. Methods: We performed a retrospective cohort study using Veterans Health Administration EHRs (from October 1, 1999, to August 31, 2019), selecting a random 25% sample of 1,329,120 Veterans out of 5,316,480 (1,193,563 males and 135,557 females) with at least 2 years of observation. The study cohort comprised 24,625 Veterans with coded self-harm and 1,304,495 uncoded, with the mean ages of 38.39 (SD 12.17) and 48.76 (SD 15.04) years, respectively. We applied our PULSNAR (positive unlabeled learning selected not at random) algorithm to estimate the proportion of individuals with uncoded self-harm. Covariates included age, medical conditions, procedures, and clinical observations. Four experts (raters) independently reviewed charts of 97 uncoded Veterans, each selected from 1% intervals of calibrated PULSNAR probabilities from 0.01 to 0.97. Agreement was assessed among raters, PULSNAR classifications, and consensus review decisions. Post hoc calibration was used to refine prevalence estimates. Results: Of the 159,049 covariates in the dataset, PULSNAR’s Extreme Gradient Boosting (XGBoost) model identified 1302 (0.82%) as informative for classification. Only 1.85% (24,625/1,329,120) of Veterans had diagnostic codes indicating self-harm events, while PULSNAR estimated an overall prevalence of 10.46% (139,026/1,329,120) by identifying an additional α =8.77% (114,404/1,304,495) of self-harm cases among the uncoded population. Of the 97 chart-reviewed patients, 39 had documented but uncoded self-harm. PULSNAR probabilities were post hoc calibrated such that their sum over the 97 cases equaled 39, which adjusted the combined coded and imputed prevalence downward from 10.46% to 7.91% (105,133/1,329,120). By applying this calibration to shift the probabilities of all uncoded Veterans, with bootstrapping for confidence intervals, PULSNAR estimates that coded self-harm represents only 23.4% (95% CI 17.76% to 31.51%) of all documented (coded+notes) self-harm. Conclusions: Under the “selected not at random” assumption, PULSNAR provides an innovative and scalable framework for estimating the clinically documented prevalence of mental health conditions and identifying the uncoded individuals with calibrated prediction, without requiring confirmed negative labels. This method offers an alternative to time-consuming chart reviews for detecting likely cases missing structured coding capture. By addressing diagnostic undercoding of mental health conditions in EHRs, this approach has the potential to enhance the estimation of mental health prevalence and support screening, activation of automated clinical decision support, targeted intervention, better resource allocation, and research to improve outcomes in real-world settings.
- 4 June 2026
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 28 (1) , e89071
Open AccessDetecting Uncoded Self-Harm in Veterans’ Electronic Health Records Using Positive and Unlabeled Learning: Retrospective Cohort Study
Background: Underdiagnosis and undercoding are common across mental health conditions, particularly suicide and self-harm. This leaves health care datasets lacking reliable negative examples needed for predictive modeling, phenotype prevalence estimation, and identification of individuals at elevated risk. We use positive and unlabeled (PU) learning to address this challenge. Objective: This study aims to identify US Veterans whose self-harm events were not explicitly captured through diagnostic codes in electronic health records (EHRs) and estimate the underlying prevalence using a novel PU learning algorithm. Methods: We performed a retrospective cohort study using Veterans Health Administration EHRs (from October 1, 1999, to August 31, 2019), selecting a random 25% sample of 1,329,120 Veterans out of 5,316,480 (1,193,563 males and 135,557 females) with at least 2 years of observation. The study cohort comprised 24,625 Veterans with coded self-harm and 1,304,495 uncoded, with the mean ages of 38.39 (SD 12.17) and 48.76 (SD 15.04) years, respectively. We applied our PULSNAR (positive unlabeled learning selected not at random) algorithm to estimate the proportion of individuals with uncoded self-harm. Covariates included age, medical conditions, procedures, and clinical observations. Four experts (raters) independently reviewed charts of 97 uncoded Veterans, each selected from 1% intervals of calibrated PULSNAR probabilities from 0.01 to 0.97. Agreement was assessed among raters, PULSNAR classifications, and consensus review decisions. Post hoc calibration was used to refine prevalence estimates. Results: Of the 159,049 covariates in the dataset, PULSNAR’s Extreme Gradient Boosting (XGBoost) model identified 1302 (0.82%) as informative for classification. Only 1.85% (24,625/1,329,120) of Veterans had diagnostic codes indicating self-harm events, while PULSNAR estimated an overall prevalence of 10.46% (139,026/1,329,120) by identifying an additional α =8.77% (114,404/1,304,495) of self-harm cases among the uncoded population. Of the 97 chart-reviewed patients, 39 had documented but uncoded self-harm. PULSNAR probabilities were post hoc calibrated such that their sum over the 97 cases equaled 39, which adjusted the combined coded and imputed prevalence downward from 10.46% to 7.91% (105,133/1,329,120). By applying this calibration to shift the probabilities of all uncoded Veterans, with bootstrapping for confidence intervals, PULSNAR estimates that coded self-harm represents only 23.4% (95% CI 17.76% to 31.51%) of all documented (coded+notes) self-harm. Conclusions: Under the “selected not at random” assumption, PULSNAR provides an innovative and scalable framework for estimating the clinically documented prevalence of mental health conditions and identifying the uncoded individuals with calibrated prediction, without requiring confirmed negative labels. This method offers an alternative to time-consuming chart reviews for detecting likely cases missing structured coding capture. By addressing diagnostic undercoding of mental health conditions in EHRs, this approach has the potential to enhance the estimation of mental health prevalence and support screening, activation of automated clinical decision support, targeted intervention, better resource allocation, and research to improve outcomes in real-world settings.
- 28 August 2025
- Published by American Association for Cancer Research (AACR) in Cancer Prevention Research
- Vol. 18 (12) , 745-755
Colorectal Cancer Screening Modalities in the United States: Regional and Rural Variance, the Resilience of Colonoscopy, and the Rise of mt-sDNA Testing
This study examined trends in colorectal cancer screening modality utilization across the United States from 2016 to 2022, leveraging a large national claims database. The purpose was to identify national, regional, and demographic patterns in screening behavior during a period that encompassed the introduction of multitarget stool DNA testing (mt-sDNA) and the COVID-19 pandemic. Among 6.9 million colorectal cancer screenings analyzed, overall utilization increased through 2019, dipped in 2020 due to pandemic disruptions, and rebounded by 2022. Colonoscopy remained the dominant modality, with its utilization increasing in both relative and absolute terms. mt-sDNA testing experienced rapid adoption, increasing from less than 1% to 17% of all screenings, whereas fecal occult blood testing and fecal immunochemical testing declined. Multinomial logistic regression revealed that utilization patterns varied significantly by region, rurality, sex, age, and year. The Midwest and rural patients exhibited higher uptake of both colonoscopy and mt-sDNA compared with other groups, whereas the West maintained the highest reliance on fecal occult blood testing and fecal immunochemical testing. Findings highlight the nonuniform adoption of screening modalities across regions, urban and rural patients, categories of sex, and age cohorts. Understanding these patterns can inform and improve future resource allocation with the goal of increasing colorectal cancer screening uptake and adherence. Prevention Relevance: This study informs cancer prevention by revealing regional and demographic variation in colorectal cancer screening modality use. Understanding these evolving patterns can inform and improve targeted strategies and allocation of resources to improve screening uptake and adherence.
- 28 August 2025
- Published by American Association for Cancer Research (AACR) in Cancer Prevention Research
- Vol. 18 (12) , 745-755
Colorectal Cancer Screening Modalities in the United States: Regional and Rural Variance, the Resilience of Colonoscopy, and the Rise of mt-sDNA Testing
This study examined trends in colorectal cancer screening modality utilization across the United States from 2016 to 2022, leveraging a large national claims database. The purpose was to identify national, regional, and demographic patterns in screening behavior during a period that encompassed the introduction of multitarget stool DNA testing (mt-sDNA) and the COVID-19 pandemic. Among 6.9 million colorectal cancer screenings analyzed, overall utilization increased through 2019, dipped in 2020 due to pandemic disruptions, and rebounded by 2022. Colonoscopy remained the dominant modality, with its utilization increasing in both relative and absolute terms. mt-sDNA testing experienced rapid adoption, increasing from less than 1% to 17% of all screenings, whereas fecal occult blood testing and fecal immunochemical testing declined. Multinomial logistic regression revealed that utilization patterns varied significantly by region, rurality, sex, age, and year. The Midwest and rural patients exhibited higher uptake of both colonoscopy and mt-sDNA compared with other groups, whereas the West maintained the highest reliance on fecal occult blood testing and fecal immunochemical testing. Findings highlight the nonuniform adoption of screening modalities across regions, urban and rural patients, categories of sex, and age cohorts. Understanding these patterns can inform and improve future resource allocation with the goal of increasing colorectal cancer screening uptake and adherence. Prevention Relevance: This study informs cancer prevention by revealing regional and demographic variation in colorectal cancer screening modality use. Understanding these evolving patterns can inform and improve targeted strategies and allocation of resources to improve screening uptake and adherence.
- 20 March 2020
- Published by Elsevier
- p. 597-605
Costs and benefits of intraoperative neurophysiological monitoring in spinal surgeries
- 20 March 2020
- Published by Elsevier
- p. 597-605
Costs and benefits of intraoperative neurophysiological monitoring in spinal surgeries
- Open Access
- 1 January 2019
- Published by Taylor & Francis in ClinicoEconomics and Outcomes Research
- Vol. 11, 335-348
An iron triangle ROI model of health care
Objective: Few, if any, return on investment (ROI) analyses of health programs make systematic considerations of patient access, instead focusing principally on gains related to cost and quality. The objective of this study was to develop an open-source model that adds an estimation of gains in patient access to a traditional ROI analysis. A classification system for quantifying gains in patient access is proposed. Materials and methods: An Excel-based ROI model was built that not only incorporated traditional ROI considerations - cost savings and patient cases avoided - but also addressed changes in patient access. The model was then applied in a case study using New Mexico Medicaid data and two proposed initiatives - a statewide health information exchange (HIE) and a community health worker (CHW) program that focused on chronic disease patients. Savings, Health, Outreach, and Access estimates were derived from the literature. ROI estimates were produced that also incorporated relative gains in patient access. Results: Combined, the HIE and CHW programs are predicted to generate a positive ROI by the fourth year, growing to 45% by the program's tenth year. Total estimated cumulative cost for both programs after 10 years is $9,555,226. Total estimated cumulative saving for both programs after 10 years is $11,332,899. Access-related costs begin moderately in year 1 at $122,766 and grow to $1,858,274 by year 10. The model estimates an Access score of 19 in year 1. This figure grew to 380 by year 10. Conclusion: Our model shows that a rough estimation of gains in to patient access can be incorporated to traditional ROI analyses. The results of our case study suggest that a CHW program and statewide HIE can generate a positive ROI for the state's Medicaid program.
- 1 January 2019
- Published by Taylor & Francis in ClinicoEconomics and Outcomes Research
- Vol. 11, 335-348
Open AccessAn iron triangle ROI model of health care
Objective: Few, if any, return on investment (ROI) analyses of health programs make systematic considerations of patient access, instead focusing principally on gains related to cost and quality. The objective of this study was to develop an open-source model that adds an estimation of gains in patient access to a traditional ROI analysis. A classification system for quantifying gains in patient access is proposed. Materials and methods: An Excel-based ROI model was built that not only incorporated traditional ROI considerations - cost savings and patient cases avoided - but also addressed changes in patient access. The model was then applied in a case study using New Mexico Medicaid data and two proposed initiatives - a statewide health information exchange (HIE) and a community health worker (CHW) program that focused on chronic disease patients. Savings, Health, Outreach, and Access estimates were derived from the literature. ROI estimates were produced that also incorporated relative gains in patient access. Results: Combined, the HIE and CHW programs are predicted to generate a positive ROI by the fourth year, growing to 45% by the program's tenth year. Total estimated cumulative cost for both programs after 10 years is $9,555,226. Total estimated cumulative saving for both programs after 10 years is $11,332,899. Access-related costs begin moderately in year 1 at $122,766 and grow to $1,858,274 by year 10. The model estimates an Access score of 19 in year 1. This figure grew to 380 by year 10. Conclusion: Our model shows that a rough estimation of gains in to patient access can be incorporated to traditional ROI analyses. The results of our case study suggest that a CHW program and statewide HIE can generate a positive ROI for the state's Medicaid program.