Find Research

A major aim is to identify novel genetic determinants of atherosclerotic vascular disease. We will also investigate whether gene-gene interactions influence susceptibility to Myocardial Infarction and PAD.

This NIH Roadmap grant uses Mayo’s open-source LexGrid technology to manage and deliver terminologies and ontologies pertinent to biomedical research in the clinical and basic science domains.

Poorly executed care transitions as patients move from one delivery setting to another lead to fragmented care evidenced by duplication of services, inappropriate and conflicting recommendations, medication errors, and patient confusion and distress. The lack of timely transfer of essential clinical information is a major barrier to effective care transitions. Three large Minnesota-based health care delivery systems have come together to use information technology to enhance communication at care transitions of patients with congestive heart failure and to evaluate its impact. The long-term objective which is crucial to the well being of the 100 million Americans with chronic illness and a rapidly growing population of elderly is to enable more informed clinical decisions which should result in safer and higher quality care of patients undergoing transitions. In addition, information exchange will be employed to facilitate the development of complete medication lists to be used as part of each organization's medication reconciliation process. The partners will execute an implementation plan for a community-shared electronic medical record abstract designed to enhance the care and safety of those crossing delivery sites. This abstract will be available near the point of care. Using a federated information model, the abstract will be a composite derived from the electronic medical records of each partner. The abstract will hold pertinent information such as a problem list, medications, allergies, recent procedures, and baseline physical, physiologic, and cognitive function supplemented by recent prescription claims history. The specific aims are to successfully execute the implementation plan, to evaluate the effect the shared abstract has on the care of patients with CHF by measuring indicators prior to the intervention and during both of a two-phased intervention using health information exchange, and to demonstrate the integration of other clinical users into this information exchange once the project is operational and has documented its connection and participation standards.

The University of Minnesota pathway prediction system (UM-PPS, http://umbbd.msi.umn.edu/predict/) recognizes functional groups in organic compounds that are potential targets of microbial catabolic reactions, and predicts transformations of these groups based on biotransformation rules. Rules are based on the University of Minnesota biocatalysis/biodegradation database (http://umbbd.msi.umn.edu/) and the scientific literature. As rules were added to the UM-PPS, more of them were triggered at each prediction step. The resulting combinatorial explosion is being addressed in four ways. Biodegradation experts give each rule an aerobic likelihood value of Very Likely, Likely, Neutral, Unlikely or Very Unlikely. Users now can choose whether they view all, or only the more aerobically likely, predicted transformations. Relative reasoning, allowing triggering of some rules to inhibit triggering of others, was implemented. Rules were initially assigned to individual chemical reactions. In selected cases, these have been replaced by super rules, which include two or more contiguous reactions that form a small pathway of their own. Rules are continually modified to improve the prediction accuracy; increasing rule stringency can improve predictions and reduce extraneous choices. The UM-PPS is freely available to all without registration. Its value to the scientific community, for academic, industrial and government use, is good and will only increase.

The program's long term objective is to improve the health status of lung transplant recipients by providing a means for the early detection of acute bronchopulmonary events using home monitoring of pulmonary function and symptoms, resulting in earlier and more timely intervention and improved outcome. The program will use a randomized controlled trial to evaluate and compare manual nurse triage with the computerized triage system in terms of patients health and quality of life, and the time required to fully utilize each triage system.

The overall objective of the study is to evaluate the effectiveness of a telehealth care coordination and case management nursing intervention for children with complex special health care needs. The study objective will be realized using a three armed randomized controlled trial with a control group receiving standard care, an intervention group receiving advanced practice nurse (APN) delivered telephone care coordiantion and case management services, and another intervention group receiving APN-delivered telephone and video care coordination and case management services.

The primary goal of this AHRQ funded proposal is to conduct a randomized clinical trial to evaluate the effectiveness of an integrated electronic health record system that includes an eMedical Record (EMR), eDental Record (EDR), and a Personal eHealth Record (PHR) to improve the quality and safety of dental care for patients with chronic illnesses. The emergence of EDRs in dental offices offers new opportunities for measurement and improvement of the quality and safety of care of by dentists. Furthermore, the combination of an EMR, PHR, and EDR into an integrated electronic health record system adds great potential for improved health information exchange, enhanced communication, and improved care, particularly for patients with chronic illnesses.

This project funded by NIDCR and housed at the University of Minnesota provides a mechanism to collect and dissemination of clinical data, retrieved implants, and biological specimens for TMJ disorders. The information system, database, biological samples, and retrieved implants are currently available to basic and clinical researchers. These materials will help in understanding the pathology of TMJD and provide information for the development of new TMJ implants. Materials available include the following: DNA isolated from blood, serum, saliva, paraffin TMJ tissue blocks, frozen TMJ tissue blocks, and retrieved implants.

The research conducted here uses modeling and data mining technologies to discover and structure clinical policies that most effectively reduce risk of cardiovascular events in complex patients with diabetes. The work is proceeding in two steps: (a) We develop methodology to identify physician treatment strategies (combinations of pharmaceutical agents, timing of clinical interventions, complexity of regimen, risky prescribing events) that minimize cost or risk of major cardiovascular complications in complex patients with diabetes, and (b) We apply computational modeling and data mining techniques to identify the optimal combinations of pharmaceutical agents to minimize pharmaceutical costs while achieving pre-specified degrees of reduction in risk of major cardiovascular complications in complex patients with diabetes. Results of this work will contribute to the important ongoing debate about comparative effectiveness of alternative clinical policies for complex patients with diabetes, including cost data needed to inform the development of clinical guidelines and public policy for the care of complex patients, whose needs are not well addressed by existing guidelines. Methods used in this project will provide a useful prototype for comparative effectiveness research that can be applied to diverse clinical domains and patient populations.

Randomized clinical trials (RCTs) have identified clinical goals that are independently related to risks of future cardiovascular events [1] [2]. Prior studies have also estimated projected risks of future events for patients with type 2 diabetes. However, existing diabetes risk models have a number of limitations: (a) Most are based on selected and unrepresentative RCT populations. (b) Models such as Framingham [3] and UKPDS [4] are dated. (c) Archimedes [5] has been developed as a general disease model, but it is a pure simulation model, not a risk prediction model. (d) Existing risk prediction models rely primarily on regression and similar approaches that are not well suited to handle missing data. This project focuses on developing functional, machine learning models of risk as well as methodology for constructing such models in other datasets and settings. These models will enable novel and potentially important new approaches to assessing comparative effectiveness of alternative clinical policies for complex patients, including estimates of long term risk needed to inform the development of clinical guidelines and public policy for the care of patients.

Utilizing electronic record data to analyze the relationship of changes in lab test values with changes in clinical actions. These data could be used to define analytic performance criteria.

Family history information has emerged as an increasingly important tool for clinical care and research. This project aims to utilize family history data more efficiently from the electronic health record and several additional disparate data sources. Knowledge representation and effective natural language processing techniques to extract this data and understand the "sub-language" of family history data in text are main subprojects. An electronic tool to perform "high-throughput screening" to identify high risk kindreds for familial cancers is being developed, along with a multi-institutional evaluation to compare automated techniques across institutions.

This research focuses on text data in community-based care and the informatics challenges in its automated use, which remains understudied, not well characterized, and of unclear value. As point-of-care documentation becomes more widespread and adopted into electronic health records, identification of gaps in the content and use of classification standards will become increasingly vital to allow accurate and functional healthcare documentation. The project specifically seeks to characterize content and use of text in relationship to standardized Omaha System terms and to describe language structure of text on a subset of summary “Clinical Care Notes”.

In this project funded by the Agency for Healthcare Reserach and Quality, we followed and evaluated the implementation and impact of a suite of new Health Information Technologies on patient safety and quality in a large children’s hospital.

In this project funded by Comprehensive Health Services Inc., we developed and tested a new predictive model of workforce health based on socio-demographic, individual, behavioral, job, industry, and environmental characteristics and risk factors.

The objective of this project is to develop and validate an assay for the prognosis of Stage I adenocarcinoma of the lung. Currently, stage remains the most significant predictor of survival, but is insufficient to accurately predict outcome for a given individual. For example, 5-year survival for stage I lung cancer patients treated by complete resection has been as low as 50%. Thus, up to 50% of patients will relapse and die of their disease, usually within two years of surgical treatment. Therefore, there exists a critical need for new methods to predict survival as well as patient response to lung cancer treatment, assisting clinicians with adjuvant therapy decisions. This project aims to discover and evaluate candidate biomarkers, develop and validate a multivariate model based on these markers and finally develop a multi-analyte assay that can be translated to a clinical test for the prognosis of Stage I lung adenocarcinoma.

Current clinicopathological evaluation of prostate cancer patients cannot accurately predict systemic progression following Radical Retropubic Prostatectomy (RRP). This is particularly true for patients with intermediate and high Gleason scores (GS≥7), who comprise nearly 35% of the prostate cancer patients that undergo RRP at Mayo Clinic. For this group of patients, the systemic progression rate is close to 13%. Stratification of these patients based on the likelihood of systemic progression and death due to prostate cancer is a key prerequisite for a more individualized approach to therapy. The goal of this proposal is the construction and independent validation of a prognostic model that combines clinicopathological parameters with molecular prognostic markers to identify those patients with GS≥7 at RRP at greatest risk to progress to systemic disease following RRP. Current understanding of prostate cancer biology is that multiple molecular events and signaling pathways are involved in its development and progression. Importantly, our work and studies by other investigators indicate that many of these events and pathways are not common in all cases (Kosari et al., 2008), emphasizing the significance of cancer heterogeneity considerations in designing prognostic models. For example, even the most common gene fusion in prostate cancer, TMPRSS2-ETS (Tomlins et al., 2005), is present only in 60% of cases. We have recently shown that a simple multi-variable model using molecular markers can significantly improve the predictive accuracy of clinical and pathologic variables (0.8 AUC in ROC analysis) when applied specifically to the GS≥7 patient population in a case-control setting where all clinically available prognostic measures including extra-capsular extension, nodal invasion, and adjuvant therapy were either matched or balanced (see Cheville et. al. JCO 2008). Our predictive model included 3 molecular parameters, namely TOP2A, CDH10 and the TMPRSS2-ETS-family fusion status. Since the TMPRSS2-ETS fusion appears to be an early event in carcinogenesis of prostate cancer (Perner et al., 2007), our data as well as recent reports (Tomlins et al., 2008) raise the possibility that this fusion leads to a distinct molecular subtype. Therefore, it is our hypothesis that molecular prognostic models of prostate cancer progression could provide better predictive value when evaluated within the context of the TMPRSS2-ETS fusion status.

To develop an evidence-based approach to improve the health literacy status among Minnesota seniors by designing and piloting a health literacy training program at Boutwells Landing Senior Living Community; developing and reproducing a training guide for the training program; promoting and expanding the use of the training program throughout Minnesota. Funded by the Greater Midwest Region of the National Network of Libraries of Medicine/NLM ($40,000) May 2009-April 2011.

OASIS and Omaha System data from 15 home health agency electronic health records was selected, combined and used to develop predictive models for improvement of various outcomes, including improvement in ambulation and oral medication management. Different models have been developed using logistic regression and data mining techniques.

The purpose of the project is to apply medical natural language processing methods to text with the Omaha System to understand possible areas of improvement to the Omaha System, quantify the value of text in this care setting, identify user issues with electronic systems in community-based practice, and gain a better understanding of language contained within electronic text in community-based care settings. Geneivie Melton-Meaux provides expertise on NLP methods, the additional investigators have expertise with the Omaha System.

We developed and evaluated software that uses food label information to infer additional nutrition-related information (specifically, ingredient proportions) for food products. We found the software estimated information more quickly and accurately than an existing manual process. The algorithms we developed are now in use at the University of Minnesota's Nutrition Coordinating Center, and in adapted form at the United States Department of Agriculture’s Nutrient Data Laboratory.

In this project, we developed software that allows consumers to plan menus based on their dietary preferences. The users input their dietary preferences (a list of foods), and their age and gender. The software then used linear programming to derive suggested daily food amounts that met recommended nutrient intakes. Users could specify lower and upper bounds on food amounts as needed. We found that the software allowed users to construct menu plans that were palatable and nutritionally complete.

Poorly executed care transitions as patients move from one delivery setting to another lead to fragmented care evidenced by duplication of services, inappropriate and conflicting recommendations, medication errors, and patient confusion and distress. The lack of timely transfer of essential clinical information is a major barrier to effective care transitions. Three large Minnesota-based health care delivery systems have come together to use information technology to enhance communication at care transitions of patients with congestive heart failure and to evaluate its impact. The long-term objective which is crucial to the well being of the 100 million Americans with chronic illness and a rapidly growing population of elderly is to enable more informed clinical decisions which should result in safer and higher quality care of patients undergoing transitions. In addition, information exchange will be employed to facilitate the development of complete medication lists to be used as part of each organization's medication reconciliation process. The partners will execute an implementation plan for a community-shared electronic medical record abstract designed to enhance the care and safety of those crossing delivery sites. This abstract will be available near the point of care. Using a federated information model, the abstract will be a composite derived from the electronic medical records of each partner. The abstract will hold pertinent information such as a problem list, medications, allergies, recent procedures, and baseline physical, physiologic, and cognitive function supplemented by recent prescription claims history. The specific aims are to successfully execute the implementation plan, to evaluate the effect the shared abstract has on the care of patients with CHF by measuring indicators prior to the intervention and during both of a two-phased intervention using health information exchange, and to demonstrate the integration of other clinical users into this information exchange once the project is operational and has documented its connection and participation standards.

In this project funded by the Agency for Healthcare Reserach and Quality, we followed and evaluated the implementation and impact of a suite of new Health Information Technologies on patient safety and quality in a large children’s hospital.

The research conducted here uses modeling and data mining technologies to discover and structure clinical policies that most effectively reduce risk of cardiovascular events in complex patients with diabetes. The work is proceeding in two steps: (a) We develop methodology to identify physician treatment strategies (combinations of pharmaceutical agents, timing of clinical interventions, complexity of regimen, risky prescribing events) that minimize cost or risk of major cardiovascular complications in complex patients with diabetes, and (b) We apply computational modeling and data mining techniques to identify the optimal combinations of pharmaceutical agents to minimize pharmaceutical costs while achieving pre-specified degrees of reduction in risk of major cardiovascular complications in complex patients with diabetes. Results of this work will contribute to the important ongoing debate about comparative effectiveness of alternative clinical policies for complex patients with diabetes, including cost data needed to inform the development of clinical guidelines and public policy for the care of complex patients, whose needs are not well addressed by existing guidelines. Methods used in this project will provide a useful prototype for comparative effectiveness research that can be applied to diverse clinical domains and patient populations.

We developed and evaluated software that uses food label information to infer additional nutrition-related information (specifically, ingredient proportions) for food products. We found the software estimated information more quickly and accurately than an existing manual process. The algorithms we developed are now in use at the University of Minnesota's Nutrition Coordinating Center, and in adapted form at the United States Department of Agriculture’s Nutrient Data Laboratory.

In this project, we developed software that allows consumers to plan menus based on their dietary preferences. The users input their dietary preferences (a list of foods), and their age and gender. The software then used linear programming to derive suggested daily food amounts that met recommended nutrient intakes. Users could specify lower and upper bounds on food amounts as needed. We found that the software allowed users to construct menu plans that were palatable and nutritionally complete.

The objective of this project is to develop and validate an assay for the prognosis of Stage I adenocarcinoma of the lung. Currently, stage remains the most significant predictor of survival, but is insufficient to accurately predict outcome for a given individual. For example, 5-year survival for stage I lung cancer patients treated by complete resection has been as low as 50%. Thus, up to 50% of patients will relapse and die of their disease, usually within two years of surgical treatment. Therefore, there exists a critical need for new methods to predict survival as well as patient response to lung cancer treatment, assisting clinicians with adjuvant therapy decisions. This project aims to discover and evaluate candidate biomarkers, develop and validate a multivariate model based on these markers and finally develop a multi-analyte assay that can be translated to a clinical test for the prognosis of Stage I lung adenocarcinoma.

The primary goal of this AHRQ funded proposal is to conduct a randomized clinical trial to evaluate the effectiveness of an integrated electronic health record system that includes an eMedical Record (EMR), eDental Record (EDR), and a Personal eHealth Record (PHR) to improve the quality and safety of dental care for patients with chronic illnesses. The emergence of EDRs in dental offices offers new opportunities for measurement and improvement of the quality and safety of care of by dentists. Furthermore, the combination of an EMR, PHR, and EDR into an integrated electronic health record system adds great potential for improved health information exchange, enhanced communication, and improved care, particularly for patients with chronic illnesses.

A major aim is to identify novel genetic determinants of atherosclerotic vascular disease. We will also investigate whether gene-gene interactions influence susceptibility to Myocardial Infarction and PAD.

To develop an evidence-based approach to improve the health literacy status among Minnesota seniors by designing and piloting a health literacy training program at Boutwells Landing Senior Living Community; developing and reproducing a training guide for the training program; promoting and expanding the use of the training program throughout Minnesota. Funded by the Greater Midwest Region of the National Network of Libraries of Medicine/NLM ($40,000) May 2009-April 2011.

The program's long term objective is to improve the health status of lung transplant recipients by providing a means for the early detection of acute bronchopulmonary events using home monitoring of pulmonary function and symptoms, resulting in earlier and more timely intervention and improved outcome. The program will use a randomized controlled trial to evaluate and compare manual nurse triage with the computerized triage system in terms of patients health and quality of life, and the time required to fully utilize each triage system.

This NIH Roadmap grant uses Mayo’s open-source LexGrid technology to manage and deliver terminologies and ontologies pertinent to biomedical research in the clinical and basic science domains.

This project funded by NIDCR and housed at the University of Minnesota provides a mechanism to collect and dissemination of clinical data, retrieved implants, and biological specimens for TMJ disorders. The information system, database, biological samples, and retrieved implants are currently available to basic and clinical researchers. These materials will help in understanding the pathology of TMJD and provide information for the development of new TMJ implants. Materials available include the following: DNA isolated from blood, serum, saliva, paraffin TMJ tissue blocks, frozen TMJ tissue blocks, and retrieved implants.

This research focuses on text data in community-based care and the informatics challenges in its automated use, which remains understudied, not well characterized, and of unclear value. As point-of-care documentation becomes more widespread and adopted into electronic health records, identification of gaps in the content and use of classification standards will become increasingly vital to allow accurate and functional healthcare documentation. The project specifically seeks to characterize content and use of text in relationship to standardized Omaha System terms and to describe language structure of text on a subset of summary “Clinical Care Notes”.

The purpose of the project is to apply medical natural language processing methods to text with the Omaha System to understand possible areas of improvement to the Omaha System, quantify the value of text in this care setting, identify user issues with electronic systems in community-based practice, and gain a better understanding of language contained within electronic text in community-based care settings. Geneivie Melton-Meaux provides expertise on NLP methods, the additional investigators have expertise with the Omaha System.

The University of Minnesota pathway prediction system (UM-PPS, http://umbbd.msi.umn.edu/predict/) recognizes functional groups in organic compounds that are potential targets of microbial catabolic reactions, and predicts transformations of these groups based on biotransformation rules. Rules are based on the University of Minnesota biocatalysis/biodegradation database (http://umbbd.msi.umn.edu/) and the scientific literature. As rules were added to the UM-PPS, more of them were triggered at each prediction step. The resulting combinatorial explosion is being addressed in four ways. Biodegradation experts give each rule an aerobic likelihood value of Very Likely, Likely, Neutral, Unlikely or Very Unlikely. Users now can choose whether they view all, or only the more aerobically likely, predicted transformations. Relative reasoning, allowing triggering of some rules to inhibit triggering of others, was implemented. Rules were initially assigned to individual chemical reactions. In selected cases, these have been replaced by super rules, which include two or more contiguous reactions that form a small pathway of their own. Rules are continually modified to improve the prediction accuracy; increasing rule stringency can improve predictions and reduce extraneous choices. The UM-PPS is freely available to all without registration. Its value to the scientific community, for academic, industrial and government use, is good and will only increase.

Current clinicopathological evaluation of prostate cancer patients cannot accurately predict systemic progression following Radical Retropubic Prostatectomy (RRP). This is particularly true for patients with intermediate and high Gleason scores (GS≥7), who comprise nearly 35% of the prostate cancer patients that undergo RRP at Mayo Clinic. For this group of patients, the systemic progression rate is close to 13%. Stratification of these patients based on the likelihood of systemic progression and death due to prostate cancer is a key prerequisite for a more individualized approach to therapy. The goal of this proposal is the construction and independent validation of a prognostic model that combines clinicopathological parameters with molecular prognostic markers to identify those patients with GS≥7 at RRP at greatest risk to progress to systemic disease following RRP. Current understanding of prostate cancer biology is that multiple molecular events and signaling pathways are involved in its development and progression. Importantly, our work and studies by other investigators indicate that many of these events and pathways are not common in all cases (Kosari et al., 2008), emphasizing the significance of cancer heterogeneity considerations in designing prognostic models. For example, even the most common gene fusion in prostate cancer, TMPRSS2-ETS (Tomlins et al., 2005), is present only in 60% of cases. We have recently shown that a simple multi-variable model using molecular markers can significantly improve the predictive accuracy of clinical and pathologic variables (0.8 AUC in ROC analysis) when applied specifically to the GS≥7 patient population in a case-control setting where all clinically available prognostic measures including extra-capsular extension, nodal invasion, and adjuvant therapy were either matched or balanced (see Cheville et. al. JCO 2008). Our predictive model included 3 molecular parameters, namely TOP2A, CDH10 and the TMPRSS2-ETS-family fusion status. Since the TMPRSS2-ETS fusion appears to be an early event in carcinogenesis of prostate cancer (Perner et al., 2007), our data as well as recent reports (Tomlins et al., 2008) raise the possibility that this fusion leads to a distinct molecular subtype. Therefore, it is our hypothesis that molecular prognostic models of prostate cancer progression could provide better predictive value when evaluated within the context of the TMPRSS2-ETS fusion status.

Family history information has emerged as an increasingly important tool for clinical care and research. This project aims to utilize family history data more efficiently from the electronic health record and several additional disparate data sources. Knowledge representation and effective natural language processing techniques to extract this data and understand the "sub-language" of family history data in text are main subprojects. An electronic tool to perform "high-throughput screening" to identify high risk kindreds for familial cancers is being developed, along with a multi-institutional evaluation to compare automated techniques across institutions.

Utilizing electronic record data to analyze the relationship of changes in lab test values with changes in clinical actions. These data could be used to define analytic performance criteria.

The overall objective of the study is to evaluate the effectiveness of a telehealth care coordination and case management nursing intervention for children with complex special health care needs. The study objective will be realized using a three armed randomized controlled trial with a control group receiving standard care, an intervention group receiving advanced practice nurse (APN) delivered telephone care coordiantion and case management services, and another intervention group receiving APN-delivered telephone and video care coordination and case management services.

OASIS and Omaha System data from 15 home health agency electronic health records was selected, combined and used to develop predictive models for improvement of various outcomes, including improvement in ambulation and oral medication management. Different models have been developed using logistic regression and data mining techniques.

Randomized clinical trials (RCTs) have identified clinical goals that are independently related to risks of future cardiovascular events [1] [2]. Prior studies have also estimated projected risks of future events for patients with type 2 diabetes. However, existing diabetes risk models have a number of limitations: (a) Most are based on selected and unrepresentative RCT populations. (b) Models such as Framingham [3] and UKPDS [4] are dated. (c) Archimedes [5] has been developed as a general disease model, but it is a pure simulation model, not a risk prediction model. (d) Existing risk prediction models rely primarily on regression and similar approaches that are not well suited to handle missing data. This project focuses on developing functional, machine learning models of risk as well as methodology for constructing such models in other datasets and settings. These models will enable novel and potentially important new approaches to assessing comparative effectiveness of alternative clinical policies for complex patients, including estimates of long term risk needed to inform the development of clinical guidelines and public policy for the care of patients.

In this project funded by Comprehensive Health Services Inc., we developed and tested a new predictive model of workforce health based on socio-demographic, individual, behavioral, job, industry, and environmental characteristics and risk factors.