Brian T. Denton
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Brian T. Denton has a total of 98 co-authors. Here are the top co-authors by number of shared publications:
- Christine Barnett
Department of Industrial and Operations Engineering, University of Michigan, Ann Arbor, MI
6 shared publications James E. Montie
Department of Urology, University of Michigan, Ann Arbor, Michigan 48109
6 shared publicationsJohn T. Wei
Department ofUrology, University of Michigan, Ann Arbor, MI, USA
4 shared publicationsTodd M. Morgan
Division of Urologic Oncology, Urology, University of Michigan, Ann Arbor, Michigan, USA
4 shared publications- Zheng Zhang
School of Management, Zhejiang University, Hangzhou 310058, China
4 shared publications David C. Miller
Department of Urology, University of Michigan, Ann Arbor, Michigan 48109
3 shared publications
Latest Publications
- Open Access
- 3 March 2026
- Published by Wolters Kluwer Health in Kidney360
- Vol. 7 (7) , 1625-1632
Adherence to Potassium Citrate, Changes in 24-hour Stone Risk Parameters, and Recurrent Kidney Stone Events
Background: Patient adherence limits the effectiveness of preventive pharmacologic therapy (PPT) for urinary stone disease (USD), highlighting the need for accurate adherence monitoring. We assessed associations between potassium citrate adherence measures, changes in 24-hour urine parameters, and the risk of recurrent stone events. Methods: In this retrospective cohort study, we identified adults enrolled in Medicare with USD, a baseline 24-hour urine collection (2010-2019), hypocitraturia or low urine pH, a prescription for potassium citrate monotherapy, and a follow-up urine collection. We defined two adherence measures: (1) percentage of days covered (PDC) using pharmacy claims and (2) change in urinary potassium between baseline and follow-up divided by medication dose (ΔK/dose). We assessed the association of these measures with changes in urinary citrate and pH using analysis of covariance, and the association between adherence and recurrent stone events using Cox models. Results: Among 793 patients meeting study criteria, 504 (63.6%) were adherent based on pharmacy claims (PDC≥80%) and 389 (49.1%) based on ΔK/dose≥0.5. Compared to PDC≥80%, ΔK/dose≥0.5 had a stronger association with change in urinary citrate ( p p <0.001). Compared to PDC≥80%, patients with PDC<80% had a higher risk of recurrent stone events (hazard ratio [HR] 1.431; 95% confidence interval [CI], 1.049-1.952), whereas ΔK/dose was not statistically significantly associated with recurrent stone events (HR 1.032; 95% CI, 0.740-1.439). Conclusions: There are clinically meaningful differences in potassium citrate adherence measures. ΔK/dose was more strongly associated with the change in urinary citrate and pH. However, only PDC was significantly associated with recurrent stone events.
- 3 March 2026
- Published by Wolters Kluwer Health in Kidney360
- Vol. 7 (7) , 1625-1632
Open AccessAdherence to Potassium Citrate, Changes in 24-hour Stone Risk Parameters, and Recurrent Kidney Stone Events
Background: Patient adherence limits the effectiveness of preventive pharmacologic therapy (PPT) for urinary stone disease (USD), highlighting the need for accurate adherence monitoring. We assessed associations between potassium citrate adherence measures, changes in 24-hour urine parameters, and the risk of recurrent stone events. Methods: In this retrospective cohort study, we identified adults enrolled in Medicare with USD, a baseline 24-hour urine collection (2010-2019), hypocitraturia or low urine pH, a prescription for potassium citrate monotherapy, and a follow-up urine collection. We defined two adherence measures: (1) percentage of days covered (PDC) using pharmacy claims and (2) change in urinary potassium between baseline and follow-up divided by medication dose (ΔK/dose). We assessed the association of these measures with changes in urinary citrate and pH using analysis of covariance, and the association between adherence and recurrent stone events using Cox models. Results: Among 793 patients meeting study criteria, 504 (63.6%) were adherent based on pharmacy claims (PDC≥80%) and 389 (49.1%) based on ΔK/dose≥0.5. Compared to PDC≥80%, ΔK/dose≥0.5 had a stronger association with change in urinary citrate ( p p <0.001). Compared to PDC≥80%, patients with PDC<80% had a higher risk of recurrent stone events (hazard ratio [HR] 1.431; 95% confidence interval [CI], 1.049-1.952), whereas ΔK/dose was not statistically significantly associated with recurrent stone events (HR 1.032; 95% CI, 0.740-1.439). Conclusions: There are clinically meaningful differences in potassium citrate adherence measures. ΔK/dose was more strongly associated with the change in urinary citrate and pH. However, only PDC was significantly associated with recurrent stone events.
- Open Access
- 13 October 2025
- Published by Springer Nature in Health Care Management Science
- Vol. 28 (4) , 738-758
Can past variants of SARS-CoV-2 predict the impact of future variants? Machine learning for early warning of US counties at risk
In this paper, we determine whether machine learning (ML) models created using data from the novel SARS-CoV-2 Alpha variant can prospectively predict county-level incidence of emerging variants, validated using data of the Omicron variant. We first select publicly-available sociodemographic, economic, and health-related characteristics of 3140 United States (US) counties at the time of the confirmed early US outbreak of the novel SARS-CoV-2 virus in March 2020 for analysis. Our primary result is the set of US counties that experienced the upper quartile of population-adjusted Omicron variant incidence at certain period (e.g., 100 days) after Omicron variant’s appearance in the US. We show more predictive results by incorporating additional data and features (e.g., human mobility) that can be acquired dynamically after the outbreak, to improve prediction accuracy at the cost of additional waiting time and effort. Towards the goal of decision support, we aim to prospectively evaluate our models’ ability to classify and rank US counties at risk. We measure their classification performance using the area under receiver operating characteristic curve score with 95% confidence intervals. We further calculate the proportion of the top counties by Omicron incidence that our models correctly identify, and compare their score with those of individual county-level features that can serve as a heuristic predictive performance baseline. Our results show that ML algorithms predict county-level Omicron variant incidence with better performance than natural heuristics that decision makers might otherwise use. More generally, historical data from the first wave of a novel pandemic can help predict the incidence of future variants and strengthen state or federal pandemic response interventions. Highlights: Data-driven predictive models that capture patterns from early viral variants can support policymaking related to emerging viral variants. County-level sociodemographic, health, and economic characteristics are predictive of early COVID-19 outcomes in the United States (US). Machine learning models trained on early US county-level COVID-19 outcomes are additionally predictive of county-level SARS-CoV-2 Omicron variant outcomes. County-level machine learning models can be used as a critical policymaking tool given the inevitability of novel emerging viruses.
- 13 October 2025
- Published by Springer Nature in Health Care Management Science
- Vol. 28 (4) , 738-758
Open AccessCan past variants of SARS-CoV-2 predict the impact of future variants? Machine learning for early warning of US counties at risk
In this paper, we determine whether machine learning (ML) models created using data from the novel SARS-CoV-2 Alpha variant can prospectively predict county-level incidence of emerging variants, validated using data of the Omicron variant. We first select publicly-available sociodemographic, economic, and health-related characteristics of 3140 United States (US) counties at the time of the confirmed early US outbreak of the novel SARS-CoV-2 virus in March 2020 for analysis. Our primary result is the set of US counties that experienced the upper quartile of population-adjusted Omicron variant incidence at certain period (e.g., 100 days) after Omicron variant’s appearance in the US. We show more predictive results by incorporating additional data and features (e.g., human mobility) that can be acquired dynamically after the outbreak, to improve prediction accuracy at the cost of additional waiting time and effort. Towards the goal of decision support, we aim to prospectively evaluate our models’ ability to classify and rank US counties at risk. We measure their classification performance using the area under receiver operating characteristic curve score with 95% confidence intervals. We further calculate the proportion of the top counties by Omicron incidence that our models correctly identify, and compare their score with those of individual county-level features that can serve as a heuristic predictive performance baseline. Our results show that ML algorithms predict county-level Omicron variant incidence with better performance than natural heuristics that decision makers might otherwise use. More generally, historical data from the first wave of a novel pandemic can help predict the incidence of future variants and strengthen state or federal pandemic response interventions. Highlights: Data-driven predictive models that capture patterns from early viral variants can support policymaking related to emerging viral variants. County-level sociodemographic, health, and economic characteristics are predictive of early COVID-19 outcomes in the United States (US). Machine learning models trained on early US county-level COVID-19 outcomes are additionally predictive of county-level SARS-CoV-2 Omicron variant outcomes. County-level machine learning models can be used as a critical policymaking tool given the inevitability of novel emerging viruses.
- Open Access
- 1 December 2024
- Published by Wolters Kluwer Health in JU Open Plus
- Vol. 2 (12)
Beneficial Impact of Follow-Up 24-Hour Urine Testing on Adherence to Pharmacological Therapy for Kidney Stone Prevention
Purpose: Preventive pharmacological therapy (PPT) is recommended for certain patients with urinary stone disease to reduce their recurrence risk, but they frequently do not adhere. With causal inference methods, we evaluated whether follow-up urine testing could be used to improve PPT adherence. Materials and Methods: Using claims data from Optum’s deidentified Clinformatics Data Mart database (2008-2019), we identified patients with urinary stone disease who were prescribed PPT (thiazide diuretic, alkali therapy, or allopurinol) after baseline 24-hour urine testing and assessed their adherence with the proportion of days covered method. After distinguishing between patients who did and did not receive follow-up urine testing, we examined the association between follow-up testing and PPT adherence with logistic regression. We then emulated a target trial using propensity score matching, stratification, and doubly robust estimators to estimate the equivalent causal effect. Results: Among 4166 patients, 496 (11.9%) completed both baseline and follow-up testing. Follow-up testing was associated with a 9.0% increase in adherence odds after adjusting for sociodemographic, behavioral, and disease-related and treatment-related patient characteristics. Furthermore, we estimated a 7.1% increase in proportion of days covered caused by follow-up testing using causal inference methods. Subgroup analyses indicated the most significant effects are among women, those with low baseline medication adherence, and those not at high risk of stone formation. Conclusions: These data suggest that follow-up urine testing after starting PPT may have a beneficial effect on long-term medication adherence.
- 1 December 2024
- Published by Wolters Kluwer Health in JU Open Plus
- Vol. 2 (12)
Open AccessBeneficial Impact of Follow-Up 24-Hour Urine Testing on Adherence to Pharmacological Therapy for Kidney Stone Prevention
Purpose: Preventive pharmacological therapy (PPT) is recommended for certain patients with urinary stone disease to reduce their recurrence risk, but they frequently do not adhere. With causal inference methods, we evaluated whether follow-up urine testing could be used to improve PPT adherence. Materials and Methods: Using claims data from Optum’s deidentified Clinformatics Data Mart database (2008-2019), we identified patients with urinary stone disease who were prescribed PPT (thiazide diuretic, alkali therapy, or allopurinol) after baseline 24-hour urine testing and assessed their adherence with the proportion of days covered method. After distinguishing between patients who did and did not receive follow-up urine testing, we examined the association between follow-up testing and PPT adherence with logistic regression. We then emulated a target trial using propensity score matching, stratification, and doubly robust estimators to estimate the equivalent causal effect. Results: Among 4166 patients, 496 (11.9%) completed both baseline and follow-up testing. Follow-up testing was associated with a 9.0% increase in adherence odds after adjusting for sociodemographic, behavioral, and disease-related and treatment-related patient characteristics. Furthermore, we estimated a 7.1% increase in proportion of days covered caused by follow-up testing using causal inference methods. Subgroup analyses indicated the most significant effects are among women, those with low baseline medication adherence, and those not at high risk of stone formation. Conclusions: These data suggest that follow-up urine testing after starting PPT may have a beneficial effect on long-term medication adherence.
- Open Access
Who Goes Next? Optimizing the Allocation of Adherence-Improving Interventions
Long-term adherence to medication is a critical factor in preventing chronic diseases, such as cardiovascular disease. To address poor adherence, physicians may recommend adherence-improving interventions; however, such interventions are costly and limited in their availability. Knowing which patients will stop adhering helps distribute the available resources more effectively. We developed a binary integer program (BIP) model to select patients for adherence-improving intervention under budget constraints. We further studied a long-term adherence prediction model using dynamic logistic regression (DLR) model that uses patients' claim data, medical health factors, demographics, and monitoring frequencies to predict the risk of future non-adherence. We trained and tested our predictive model to longitudinal data for cardiovascular disease in a large cohort of patients taking medication for cholesterol control seen in the national Veterans Affairs health system. Our study shows the importance of including past adherence to increase prediction accuracy. Finally, we assess the potential benefits of using the prediction model by proposing an algorithm that combines the DLR and BIP models to decrease the number of CVD events in a population.
Open AccessWho Goes Next? Optimizing the Allocation of Adherence-Improving Interventions
Long-term adherence to medication is a critical factor in preventing chronic diseases, such as cardiovascular disease. To address poor adherence, physicians may recommend adherence-improving interventions; however, such interventions are costly and limited in their availability. Knowing which patients will stop adhering helps distribute the available resources more effectively. We developed a binary integer program (BIP) model to select patients for adherence-improving intervention under budget constraints. We further studied a long-term adherence prediction model using dynamic logistic regression (DLR) model that uses patients' claim data, medical health factors, demographics, and monitoring frequencies to predict the risk of future non-adherence. We trained and tested our predictive model to longitudinal data for cardiovascular disease in a large cohort of patients taking medication for cholesterol control seen in the national Veterans Affairs health system. Our study shows the importance of including past adherence to increase prediction accuracy. Finally, we assess the potential benefits of using the prediction model by proposing an algorithm that combines the DLR and BIP models to decrease the number of CVD events in a population.
- Open Access
Updating Clinical Risk Stratification Models Using Rank-Based Compatibility: Approaches for Evaluating and Optimizing Clinician-Model Team Performance
As data shift or new data become available, updating clinical machine learning models may be necessary to maintain or improve performance over time. However, updating a model can introduce compatibility issues when the behavior of the updated model does not align with user expectations, resulting in poor user-model team performance. Existing compatibility measures depend on model decision thresholds, limiting their applicability in settings where models are used to generate rankings based on estimated risk. To address this limitation, we propose a novel rank-based compatibility measure, $C^R$, and a new loss function that aims to optimize discriminative performance while encouraging good compatibility. Applied to a case study in mortality risk stratification leveraging data from MIMIC, our approach yields more compatible models while maintaining discriminative performance compared to existing model selection techniques, with an increase in $C^R$ of $0.019$ ($95\%$ confidence interval: $0.005$, $0.035$). This work provides new tools to analyze and update risk stratification models used in clinical care.Open AccessUpdating Clinical Risk Stratification Models Using Rank-Based Compatibility: Approaches for Evaluating and Optimizing Clinician-Model Team Performance
As data shift or new data become available, updating clinical machine learning models may be necessary to maintain or improve performance over time. However, updating a model can introduce compatibility issues when the behavior of the updated model does not align with user expectations, resulting in poor user-model team performance. Existing compatibility measures depend on model decision thresholds, limiting their applicability in settings where models are used to generate rankings based on estimated risk. To address this limitation, we propose a novel rank-based compatibility measure, $C^R$, and a new loss function that aims to optimize discriminative performance while encouraging good compatibility. Applied to a case study in mortality risk stratification leveraging data from MIMIC, our approach yields more compatible models while maintaining discriminative performance compared to existing model selection techniques, with an increase in $C^R$ of $0.019$ ($95\%$ confidence interval: $0.005$, $0.035$). This work provides new tools to analyze and update risk stratification models used in clinical care. - 1 November 2022
- Published by SAGE Publications in Production and Operations Management
- Vol. 31 (11) , 4021-4037
Optimization of active surveillance strategies for heterogeneous patients with prostate cancer
Prostate cancer (PCa) is common in American men with long latent periods, during which the disease is asymptomatic. Active surveillance is a monitoring strategy commonly used for patients diagnosed with low‐risk PCa who may harbor latent high‐risk PCa. The optimal monitoring strategy attempts to minimize the disutility of testing while ensuring that the patient is detected at the earliest time when the disease progresses. Unfortunately, guidelines for the active surveillance of PCa are often one‐size‐fits‐all strategies that ignore the heterogeneity among multiple patient types. In contrast, personalized strategies based on partially observable Markov decision process (POMDP) models are challenging to implement in practice given the large number of possible strategies that can be used. This article presents a two‐stage stochastic programming approach that selects a set of strategies for predefined cardinality based on patients' disease risks. The first‐stage decision variables include binary variables for the selection of periods at which to test patients in each strategy and the assignment of multiple patient types to strategies. The objective is to maximize a weighted reward function that considers the need for cancer detection, missed detection, and cost of monitoring patients. We discuss the structure and complexity of the model and reformulate a logic‐based Bender's decomposition formulation that can solve realistic instances to optimality. We present a case study for the active surveillance of PCa and show that our model results in strategies that vary in intensity according to patient disease risk. Finally, we show that our model can generate a small number of strategies that can significantly improve the existing “one‐size‐fits‐all” guideline strategies used in practice.
- 1 November 2022
- Published by SAGE Publications in Production and Operations Management
- Vol. 31 (11) , 4021-4037
Optimization of active surveillance strategies for heterogeneous patients with prostate cancer
Prostate cancer (PCa) is common in American men with long latent periods, during which the disease is asymptomatic. Active surveillance is a monitoring strategy commonly used for patients diagnosed with low‐risk PCa who may harbor latent high‐risk PCa. The optimal monitoring strategy attempts to minimize the disutility of testing while ensuring that the patient is detected at the earliest time when the disease progresses. Unfortunately, guidelines for the active surveillance of PCa are often one‐size‐fits‐all strategies that ignore the heterogeneity among multiple patient types. In contrast, personalized strategies based on partially observable Markov decision process (POMDP) models are challenging to implement in practice given the large number of possible strategies that can be used. This article presents a two‐stage stochastic programming approach that selects a set of strategies for predefined cardinality based on patients' disease risks. The first‐stage decision variables include binary variables for the selection of periods at which to test patients in each strategy and the assignment of multiple patient types to strategies. The objective is to maximize a weighted reward function that considers the need for cancer detection, missed detection, and cost of monitoring patients. We discuss the structure and complexity of the model and reformulate a logic‐based Bender's decomposition formulation that can solve realistic instances to optimality. We present a case study for the active surveillance of PCa and show that our model results in strategies that vary in intensity according to patient disease risk. Finally, we show that our model can generate a small number of strategies that can significantly improve the existing “one‐size‐fits‐all” guideline strategies used in practice.
- 26 October 2022
- Published by Springer Nature in Health Care Management Science
- Vol. 26 (1) , 93-116
Monitoring policy in the context of preventive treatment of cardiovascular disease
Preventing chronic diseases is an essential aspect of medical care. To prevent chronic diseases, physicians focus on monitoring their risk factors and prescribing the necessary medication. The optimal monitoring policy depends on the patient’s risk factors and demographics. Monitoring too frequently may be unnecessary and costly; on the other hand, monitoring the patient infrequently means the patient may forgo needed treatment and experience adverse events related to the disease. We propose a finite horizon and finite-state Markov decision process to define monitoring policies. To build our Markov decision process, we estimate stochastic models based on longitudinal observational data from electronic health records for a large cohort of patients seen in the national U.S. Veterans Affairs health system. We use our model to study policies for whether or when to assess the need for cholesterol-lowering medications. We further use our model to investigate the role of gender and race on optimal monitoring policies.
- 26 October 2022
- Published by Springer Nature in Health Care Management Science
- Vol. 26 (1) , 93-116
Monitoring policy in the context of preventive treatment of cardiovascular disease
Preventing chronic diseases is an essential aspect of medical care. To prevent chronic diseases, physicians focus on monitoring their risk factors and prescribing the necessary medication. The optimal monitoring policy depends on the patient’s risk factors and demographics. Monitoring too frequently may be unnecessary and costly; on the other hand, monitoring the patient infrequently means the patient may forgo needed treatment and experience adverse events related to the disease. We propose a finite horizon and finite-state Markov decision process to define monitoring policies. To build our Markov decision process, we estimate stochastic models based on longitudinal observational data from electronic health records for a large cohort of patients seen in the national U.S. Veterans Affairs health system. We use our model to study policies for whether or when to assess the need for cholesterol-lowering medications. We further use our model to investigate the role of gender and race on optimal monitoring policies.
- 8 August 2022
- Published by Taylor & Francis in IISE Transactions
- Vol. 55 (1) , 94-105
Frontiers of medical decision-making in the modern age of data analytics
Recent decades have seen considerable advances in developing Industrial Engineering/Operations Research (IE/OR) models for improving decision-making in healthcare. These approaches span the full range of descriptive, predictive, and prescriptive models for supporting patients' and clinicians' decision-making. The pervasive use of information technology to collect and store electronic health records, insurance claims, genomic information, and other observational data has opened new doors for developing, validating, and applying these types of data-driven IE/OR models. This article describes opportunities at the frontier of medical decision-making, emphasizing the intersection of medicine, data analytics, and operations research. Many of the examples covered intersect the fields of statistics, machine learning, and artificial intelligence. A series of motivating examples illustrate the possibilities and some promising future research directions.
- 8 August 2022
- Published by Taylor & Francis in IISE Transactions
- Vol. 55 (1) , 94-105
Frontiers of medical decision-making in the modern age of data analytics
Recent decades have seen considerable advances in developing Industrial Engineering/Operations Research (IE/OR) models for improving decision-making in healthcare. These approaches span the full range of descriptive, predictive, and prescriptive models for supporting patients' and clinicians' decision-making. The pervasive use of information technology to collect and store electronic health records, insurance claims, genomic information, and other observational data has opened new doors for developing, validating, and applying these types of data-driven IE/OR models. This article describes opportunities at the frontier of medical decision-making, emphasizing the intersection of medicine, data analytics, and operations research. Many of the examples covered intersect the fields of statistics, machine learning, and artificial intelligence. A series of motivating examples illustrate the possibilities and some promising future research directions.
- 10 May 2021
- Published by Taylor & Francis in IISE Transactions
- Vol. 53 (10) , 1-16
Multi-model Markov decision processes
Markov decision processes (MDPs) have found success in many application areas that involve sequential decision making under uncertainty, including the evaluation and design of treatment and screening protocols for medical decision making. However, the data used to parameterize the model can influence what policies are recommended, and multiple competing data sources are common in many application areas, including medicine. In this article, we introduce the Multi-model Markov decision process (MMDP) which generalizes a standard MDP by allowing for multiple models of the rewards and transition probabilities. Solution of the MMDP generates a single policy that maximizes the weighted performance over all models. This approach allows the decision maker to explicitly trade-off conflicting sources of data while generating a policy of the same level of complexity for models that only consider a single source of data. We study the structural properties of this problem and show that it is at least NP-hard. We develop exact methods and fast approximation methods supported by error bounds. Finally, we illustrate the effectiveness and the scalability of our approach using a case study in preventative blood pressure and cholesterol management that accounts for conflicting published cardiovascular risk models.
- 10 May 2021
- Published by Taylor & Francis in IISE Transactions
- Vol. 53 (10) , 1-16
Multi-model Markov decision processes
Markov decision processes (MDPs) have found success in many application areas that involve sequential decision making under uncertainty, including the evaluation and design of treatment and screening protocols for medical decision making. However, the data used to parameterize the model can influence what policies are recommended, and multiple competing data sources are common in many application areas, including medicine. In this article, we introduce the Multi-model Markov decision process (MMDP) which generalizes a standard MDP by allowing for multiple models of the rewards and transition probabilities. Solution of the MMDP generates a single policy that maximizes the weighted performance over all models. This approach allows the decision maker to explicitly trade-off conflicting sources of data while generating a policy of the same level of complexity for models that only consider a single source of data. We study the structural properties of this problem and show that it is at least NP-hard. We develop exact methods and fast approximation methods supported by error bounds. Finally, we illustrate the effectiveness and the scalability of our approach using a case study in preventative blood pressure and cholesterol management that accounts for conflicting published cardiovascular risk models.
- 1 May 2021
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 69 (3) , 774-794
OR Practice–Data Analytics for Optimal Detection of Metastatic Prostate Cancer
We used data-analytics approaches to develop, calibrate, and validate predictive models, to help urologists in a large statewide collaborative make prostate cancer staging decisions on the basis of individual patient risk factors. The models were validated using statistical methods based on bootstrapping and evaluation on out-of-sample data. These models were used to design guidelines that optimally weigh the benefits and harms of radiological imaging for the detection of metastatic prostate cancer. The Michigan Urological Surgery Improvement Collaborative, a statewide medical collaborative, implemented these guidelines, which were predicted to reduce unnecessary imaging by more than 40% and limit the percentage of patients with missed metastatic disease to be less than 1%. The effects of the guidelines were measured after implementation to confirm their impact on reducing unnecessary imaging across the state of Michigan.
- 1 May 2021
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Operations Research
- Vol. 69 (3) , 774-794
OR Practice–Data Analytics for Optimal Detection of Metastatic Prostate Cancer
We used data-analytics approaches to develop, calibrate, and validate predictive models, to help urologists in a large statewide collaborative make prostate cancer staging decisions on the basis of individual patient risk factors. The models were validated using statistical methods based on bootstrapping and evaluation on out-of-sample data. These models were used to design guidelines that optimally weigh the benefits and harms of radiological imaging for the detection of metastatic prostate cancer. The Michigan Urological Surgery Improvement Collaborative, a statewide medical collaborative, implemented these guidelines, which were predicted to reduce unnecessary imaging by more than 40% and limit the percentage of patients with missed metastatic disease to be less than 1%. The effects of the guidelines were measured after implementation to confirm their impact on reducing unnecessary imaging across the state of Michigan.