Martin Schumacher
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
Research Interests
Martin Schumacher has a total of 142 co-authors. Here are the top co-authors by number of shared publications:
- Arthur Allignol
Institute of Medical Biometry and Medical Informatics University Medical Center Freiburg Freiburg Center for Data Analysis and Modelling University of Freiburg
13 shared publications - Jan Beyersmann
Institute of Medical Biometry and Medical Informatics University Medical Center Freiburg Freiburg Center for Data Analysis and Modelling University of Freiburg
13 shared publications - Martin Wolkewitz
Institute of Medical Biometry and Statistics, Faculty of Medicine and Medical Center, University of Freiburg, 79104 Freiburg, Germany
10 shared publications - Gerta Rücker
University of Freiburg
8 shared publications - Harald Binder
Institute of Medical Biometry and Statistics (IMBI), Faculty of Medicine and Medical Center, University of Freiburg, Germany
8 shared publications - Gabi Schulgen/Kristiansen
Burgunderweg 1, 78464 Konstanz, Deutschland
6 shared publications
Latest Publications
- Open Access
Investigator-initiated versus industry-sponsored trials – Visibility and relevance of randomized controlled trials in clinical practice guidelines (IMPACT)
Background: The goal of evidence-based medicine is to make clinical decisions based on the best available, relevant evidence. For this to be possible, studies such as randomized controlled trials (RCTs), which are widely considered to provide the best evidence of all forms of primary research, must be visible and have an impact on clinical practice guidelines. We further investigated the impact of publicly and commercially sponsored RCTs on clinical practice guidelines by measuring direct and indirect impactful citations and the time to guideline impact. Methods: We considered the sample from the IMPACT study, where a total of 691 RCTs (120 German investigator-initiated trials (IITs), 200 international IITs, 171 German industry-sponsored trials (ISTs) and 200 international ISTs) was sampled from registries and followed prospectively. Their publications in the form of original articles and systematic reviews (SRs) citing these articles had already been identified, as well as clinical practice guidelines (CPGs) which cited either the articles or the SRs. The methods and results of this effort were already published. In this investigation we aimed to better characterize the impact of RCTs in CPGs. Therefore, we identified all citations of the original articles and SRs in the citing CPGs and classified them into impactful and non-impactful. This allowed us to calculate an estimate for the guideline impact of a trial. In addition, we estimated the time-to-guideline-impact, defined as the time to a direct and indirect impactful citation in a CPG. Direct means that the publication of a trial was cited in the main text of a CPG. Indirect means that the publication was cited and included in the findings of a SR and the SR was cited in the main text of a CPG. We also investigated to what extent pre-defined study characteristics influenced the guideline impact using multivariable negative binomial regression as well as the time-to-guideline impact using multivariable Cox proportional hazards regression. Results: Overall, 22% of RCTs impacted a CPG. For international ISTs, only 15% of trials had an impact in CPGs. Overall, of the 405 associated guidelines, 331 were impacted. Larger trials were associated with more impactful main text citations in CPGs and earlier time-to-guideline impact, while international industry-sponsored trials were associated with smaller impact on CPGs and longer time-to-guideline impact. IITs funded by governmental bodies in Germany reached an impact on CPGs that is on par with German ISTs or international IITs and ISTs. Conclusion: This study demonstrated that a considerable number of trials previously identi-fied as being linked to CPGs have had impact in those CPGs (85%). International ISTs seem to have a lower impact on CPGs, and fewer of them influence CPGs at all.
Open AccessInvestigator-initiated versus industry-sponsored trials – Visibility and relevance of randomized controlled trials in clinical practice guidelines (IMPACT)
Background: The goal of evidence-based medicine is to make clinical decisions based on the best available, relevant evidence. For this to be possible, studies such as randomized controlled trials (RCTs), which are widely considered to provide the best evidence of all forms of primary research, must be visible and have an impact on clinical practice guidelines. We further investigated the impact of publicly and commercially sponsored RCTs on clinical practice guidelines by measuring direct and indirect impactful citations and the time to guideline impact. Methods: We considered the sample from the IMPACT study, where a total of 691 RCTs (120 German investigator-initiated trials (IITs), 200 international IITs, 171 German industry-sponsored trials (ISTs) and 200 international ISTs) was sampled from registries and followed prospectively. Their publications in the form of original articles and systematic reviews (SRs) citing these articles had already been identified, as well as clinical practice guidelines (CPGs) which cited either the articles or the SRs. The methods and results of this effort were already published. In this investigation we aimed to better characterize the impact of RCTs in CPGs. Therefore, we identified all citations of the original articles and SRs in the citing CPGs and classified them into impactful and non-impactful. This allowed us to calculate an estimate for the guideline impact of a trial. In addition, we estimated the time-to-guideline-impact, defined as the time to a direct and indirect impactful citation in a CPG. Direct means that the publication of a trial was cited in the main text of a CPG. Indirect means that the publication was cited and included in the findings of a SR and the SR was cited in the main text of a CPG. We also investigated to what extent pre-defined study characteristics influenced the guideline impact using multivariable negative binomial regression as well as the time-to-guideline impact using multivariable Cox proportional hazards regression. Results: Overall, 22% of RCTs impacted a CPG. For international ISTs, only 15% of trials had an impact in CPGs. Overall, of the 405 associated guidelines, 331 were impacted. Larger trials were associated with more impactful main text citations in CPGs and earlier time-to-guideline impact, while international industry-sponsored trials were associated with smaller impact on CPGs and longer time-to-guideline impact. IITs funded by governmental bodies in Germany reached an impact on CPGs that is on par with German ISTs or international IITs and ISTs. Conclusion: This study demonstrated that a considerable number of trials previously identi-fied as being linked to CPGs have had impact in those CPGs (85%). International ISTs seem to have a lower impact on CPGs, and fewer of them influence CPGs at all.
- 27 February 2023
- Published by Wiley in Biometrical Journal
- Vol. 65 (4) , e2200360
Letter to the editor regarding the paper “New weighting methods when cases are only a subset of events in a nested case‐control study” by Qian M. Zhou, Xuan Wang, Yingye Zheng, and Tianxi Cai
- 27 February 2023
- Published by Wiley in Biometrical Journal
- Vol. 65 (4) , e2200360
Letter to the editor regarding the paper “New weighting methods when cases are only a subset of events in a nested case‐control study” by Qian M. Zhou, Xuan Wang, Yingye Zheng, and Tianxi Cai
- Open Access
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Computational Biology and Bioinformatics
- Vol. 18 (6) , 2635-2648
Netboost: Boosting-Supported Network Analysis Improves High-Dimensional Omics Prediction in Acute Myeloid Leukemia and Huntington’s Disease
State-of-the art selection methods fail to identify weak but cumulative effects of features found in many high-dimensional omics datasets. Nevertheless, these features play an important role in certain diseases. We present Netboost, a three-step dimension reduction technique. First, a boosting-based filter is combined with the topological overlap measure to identify the essential edges of the network. Second, sparse hierarchical clustering is applied on the selected edges to identify modules and finally module information is aggregated by the first principal components. We demonstrate the application of the newly developed Netboost in combination with CoxBoost for survival prediction of DNA methylation and gene expression data from 180 acute myeloid leukemia (AML) patients and show, based on cross-validated prediction error curve estimates, its prediction superiority over variable selection on the full dataset as well as over an alternative clustering approach. The identified signature related to chromatin modifying enzymes was replicated in an independent dataset, the phase II AMLSG 12-09 study. In a second application we combine Netboost with Random Forest classification and improve the disease classification error in RNA-sequencing data of Huntington's disease mice. Netboost is a freely available Bioconductor R package for dimension reduction and hypothesis generation in high-dimensional omics applications.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Computational Biology and Bioinformatics
- Vol. 18 (6) , 2635-2648
Open AccessNetboost: Boosting-Supported Network Analysis Improves High-Dimensional Omics Prediction in Acute Myeloid Leukemia and Huntington’s Disease
State-of-the art selection methods fail to identify weak but cumulative effects of features found in many high-dimensional omics datasets. Nevertheless, these features play an important role in certain diseases. We present Netboost, a three-step dimension reduction technique. First, a boosting-based filter is combined with the topological overlap measure to identify the essential edges of the network. Second, sparse hierarchical clustering is applied on the selected edges to identify modules and finally module information is aggregated by the first principal components. We demonstrate the application of the newly developed Netboost in combination with CoxBoost for survival prediction of DNA methylation and gene expression data from 180 acute myeloid leukemia (AML) patients and show, based on cross-validated prediction error curve estimates, its prediction superiority over variable selection on the full dataset as well as over an alternative clustering approach. The identified signature related to chromatin modifying enzymes was replicated in an independent dataset, the phase II AMLSG 12-09 study. In a second application we combine Netboost with Random Forest classification and improve the disease classification error in RNA-sequencing data of Huntington's disease mice. Netboost is a freely available Bioconductor R package for dimension reduction and hypothesis generation in high-dimensional omics applications.
- Open Access
- 31 August 2021
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 21 (1) , 1-20
Investigator initiated trials versus industry sponsored trials - translation of randomized controlled trials into clinical practice (IMPACT)
Background: Healthcare decisions are ideally based on clinical trial results, published in study registries, as journal articles or summarized in secondary research articles. In this research project, we investigated the impact of academically and commercially sponsored clinical trials on medical practice by measuring the proportion of trials published and cited by systematic reviews and clinical guidelines. Methods: We examined 691 multicenter, randomized controlled trials that started in 2005 or later and were completed by the end of 2016. To determine whether sponsorship/funding and place of conduct influence a trial’s impact, we created four sub-cohorts of investigator initiated trials (IITs) and industry sponsored trials (ISTs): 120 IITs and 171 ISTs with German contribution compared to 200 IITs and 200 ISTs without German contribution. We balanced the groups for study phase and place of conduct. German IITs were funded by the German Research Foundation (DFG), the Federal Ministry of Education and Research (BMBF), or by another non-commercial research organization. All other trials were drawn from the German Clinical Trials Register or ClinicalTrials.gov. We investigated, to what extent study characteristics were associated with publication and impact using multivariable logistic regressions. Results: For 80% of the 691 trials, results were published as result articles in a medical journal and/or study registry, 52% were cited by a systematic review, and 26% reached impact in a clinical guideline. Drug trials and larger trials were associated with a higher probability to be published and to have an impact than non-drug trials and smaller trials. Results of IITs were more often published as a journal article while results of ISTs were more often published in study registries. International ISTs less often gained impact by inclusion in systematic reviews or guidelines than IITs. Conclusion: An encouraging high proportion of the clinical trials were published, and a considerable proportion gained impact on clinical practice. However, there is still room for improvement. For publishing study results, study registries have become an alternative or complement to journal articles, especially for ISTs. IITs funded by governmental bodies in Germany reached an impact that is comparable to international IITs and ISTs.
- 31 August 2021
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 21 (1) , 1-20
Open AccessInvestigator initiated trials versus industry sponsored trials - translation of randomized controlled trials into clinical practice (IMPACT)
Background: Healthcare decisions are ideally based on clinical trial results, published in study registries, as journal articles or summarized in secondary research articles. In this research project, we investigated the impact of academically and commercially sponsored clinical trials on medical practice by measuring the proportion of trials published and cited by systematic reviews and clinical guidelines. Methods: We examined 691 multicenter, randomized controlled trials that started in 2005 or later and were completed by the end of 2016. To determine whether sponsorship/funding and place of conduct influence a trial’s impact, we created four sub-cohorts of investigator initiated trials (IITs) and industry sponsored trials (ISTs): 120 IITs and 171 ISTs with German contribution compared to 200 IITs and 200 ISTs without German contribution. We balanced the groups for study phase and place of conduct. German IITs were funded by the German Research Foundation (DFG), the Federal Ministry of Education and Research (BMBF), or by another non-commercial research organization. All other trials were drawn from the German Clinical Trials Register or ClinicalTrials.gov. We investigated, to what extent study characteristics were associated with publication and impact using multivariable logistic regressions. Results: For 80% of the 691 trials, results were published as result articles in a medical journal and/or study registry, 52% were cited by a systematic review, and 26% reached impact in a clinical guideline. Drug trials and larger trials were associated with a higher probability to be published and to have an impact than non-drug trials and smaller trials. Results of IITs were more often published as a journal article while results of ISTs were more often published in study registries. International ISTs less often gained impact by inclusion in systematic reviews or guidelines than IITs. Conclusion: An encouraging high proportion of the clinical trials were published, and a considerable proportion gained impact on clinical practice. However, there is still room for improvement. For publishing study results, study registries have become an alternative or complement to journal articles, especially for ISTs. IITs funded by governmental bodies in Germany reached an impact that is comparable to international IITs and ISTs.
- Open Access
- 5 May 2021
- Published by Wiley in Biometrical Journal
- Vol. 63 (7) , 1493-1506
A shared frailty model for multivariate longitudinal data on adverse event of radiation therapy
Oral mucositis is an inflammatory adverse event when treating head and neck cancer patients with radiation therapy (RT). The severity of its occurrence is believed to mainly depend on its site and the distribution of a cumulative radiation dose in the mouth area. The motivating study investigating differences in radiosensitivities (mucositis progression) at distinct sites where the severity of mucositis is assessed regularly at eight distinct sites on an ordinal scale results in multivariate longitudinal data and thus poses certain challenges. To deal with the multivariate longitudinal data in this particular setting, we take a time‐to‐event approach focusing on the first occurrence of severe mucositis at the distinct sites using the fact that the site‐specific cumulative radiation dose thought to be the main driver of oral mucositis develops over time. Thereby, we may address multivariate longitudinal processes in a simpler and more compact fashion. In this article, to find out differences in mucositis progression at eight distinct sites we propose a shared frailty model for multivariate parallel processes within individuals. The shared frailty model directly incorporating ‘process indicators’ as covariates turns out to adequately explain the differences in the parallel processes (here, mucositis progressions at distinct sites) while taking individual effects into account. The parallel result with the one from the previous analysis based on the same data but conducted with an alternative statistical methodology shows adequacy of the proposed approach.
- 5 May 2021
- Published by Wiley in Biometrical Journal
- Vol. 63 (7) , 1493-1506
Open AccessA shared frailty model for multivariate longitudinal data on adverse event of radiation therapy
Oral mucositis is an inflammatory adverse event when treating head and neck cancer patients with radiation therapy (RT). The severity of its occurrence is believed to mainly depend on its site and the distribution of a cumulative radiation dose in the mouth area. The motivating study investigating differences in radiosensitivities (mucositis progression) at distinct sites where the severity of mucositis is assessed regularly at eight distinct sites on an ordinal scale results in multivariate longitudinal data and thus poses certain challenges. To deal with the multivariate longitudinal data in this particular setting, we take a time‐to‐event approach focusing on the first occurrence of severe mucositis at the distinct sites using the fact that the site‐specific cumulative radiation dose thought to be the main driver of oral mucositis develops over time. Thereby, we may address multivariate longitudinal processes in a simpler and more compact fashion. In this article, to find out differences in mucositis progression at eight distinct sites we propose a shared frailty model for multivariate parallel processes within individuals. The shared frailty model directly incorporating ‘process indicators’ as covariates turns out to adequately explain the differences in the parallel processes (here, mucositis progressions at distinct sites) while taking individual effects into account. The parallel result with the one from the previous analysis based on the same data but conducted with an alternative statistical methodology shows adequacy of the proposed approach.
- Open Access
- 2 October 2020
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 20 (1) , 1-15
Impact of investigator initiated trials and industry sponsored trials on medical practice (IMPACT): rationale and study design
Background: The German Research Foundation (DFG) and the Federal Ministry of Education and Research (BMBF) initiated large research programs to foster high quality clinical research in the academic area. These investigator initiated trials (IITs) cover important areas of medical research and often go beyond the scope of industry sponsored trials (ISTs). The purpose of this project was to understand to what extent results of randomized controlled IITs and ISTs have an impact on medical practice, measured by their availability for decisions in healthcare and their implementation in clinical practice. We aimed to determine study characteristics influencing a trial’s impact such as type of sponsor and place of conduct. In this article, we describe the rationale and design of this project and present the characteristics of the trials included in our study cohort. Methods: The research impact of the following sub-cohorts was compared: German IITs (funded by DFG and BMBF or by other German non-commercial organizations), international IITs (without German contribution), German ISTs, and international ISTs. Trials included were drawn from the DFG−/BMBF-Websites, the German Clinical Trials Register, and from ClinicalTrials.gov. Research impact was measured as follows: 1) proportion of published trials, 2) time to publication, 3) proportion of publications appropriately indexed in biomedical databases, 4) proportion of openly accessible publications, 5) broadness of publication’s target group, 6) citation of publications by systematic reviews or meta-analyses, and 7) appearance of publications or citing systematic reviews or meta-analyses in clinical practice guidelines. We also aimed to identify study characteristics associated with the impact of trials. Results: We included 691 trials: 120 German IITs, 200 International IITs, 171 German ISTs and 200 International ISTs. The median number of participants was 150, 30% were international trials and 70% national trials, 48% drug-trials and 52% non-drug trials. Overall, 72% of the trials had one pre-defined primary endpoint, 28% two or more (max. 36). Conclusions: The results of this project deepen our understanding of the impact of biomedical research on clinical practice and healthcare policy, add important insights for the efficient allocation of scarce research resources and may facilitate providing accountability to the different stakeholders involved.
- 2 October 2020
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 20 (1) , 1-15
Open AccessImpact of investigator initiated trials and industry sponsored trials on medical practice (IMPACT): rationale and study design
Background: The German Research Foundation (DFG) and the Federal Ministry of Education and Research (BMBF) initiated large research programs to foster high quality clinical research in the academic area. These investigator initiated trials (IITs) cover important areas of medical research and often go beyond the scope of industry sponsored trials (ISTs). The purpose of this project was to understand to what extent results of randomized controlled IITs and ISTs have an impact on medical practice, measured by their availability for decisions in healthcare and their implementation in clinical practice. We aimed to determine study characteristics influencing a trial’s impact such as type of sponsor and place of conduct. In this article, we describe the rationale and design of this project and present the characteristics of the trials included in our study cohort. Methods: The research impact of the following sub-cohorts was compared: German IITs (funded by DFG and BMBF or by other German non-commercial organizations), international IITs (without German contribution), German ISTs, and international ISTs. Trials included were drawn from the DFG−/BMBF-Websites, the German Clinical Trials Register, and from ClinicalTrials.gov. Research impact was measured as follows: 1) proportion of published trials, 2) time to publication, 3) proportion of publications appropriately indexed in biomedical databases, 4) proportion of openly accessible publications, 5) broadness of publication’s target group, 6) citation of publications by systematic reviews or meta-analyses, and 7) appearance of publications or citing systematic reviews or meta-analyses in clinical practice guidelines. We also aimed to identify study characteristics associated with the impact of trials. Results: We included 691 trials: 120 German IITs, 200 International IITs, 171 German ISTs and 200 International ISTs. The median number of participants was 150, 30% were international trials and 70% national trials, 48% drug-trials and 52% non-drug trials. Overall, 72% of the trials had one pre-defined primary endpoint, 28% two or more (max. 36). Conclusions: The results of this project deepen our understanding of the impact of biomedical research on clinical practice and healthcare policy, add important insights for the efficient allocation of scarce research resources and may facilitate providing accountability to the different stakeholders involved.
- Open Access
- 8 July 2020
- Published by Wiley in Biometrical Journal
- Vol. 63 (2) , 226-246
Doug Altman: Driving critical appraisal and improvements in the quality of methodological and medical research
Doug Altman was a visionary leader and one of the most influential medical statisticians of the last 40 years. Based on a presentation in the “Invited session in memory of Doug Altman” at the 40th Annual Conference of the International Society for Clinical Biostatistics (ISCB) in Leuven, Belgium and our long‐standing collaborations with Doug, we discuss his contributions to regression modeling, reporting, prognosis research, as well as some more general issues while acknowledging that we cannot cover the whole spectrum of Doug's considerable methodological output. His statement “To maximize the benefit to society, you need to not just do research but do it well” should be a driver for all researchers. To improve current and future research, we aim to summarize Doug's messages for these three topics.
- 8 July 2020
- Published by Wiley in Biometrical Journal
- Vol. 63 (2) , 226-246
Open AccessDoug Altman: Driving critical appraisal and improvements in the quality of methodological and medical research
Doug Altman was a visionary leader and one of the most influential medical statisticians of the last 40 years. Based on a presentation in the “Invited session in memory of Doug Altman” at the 40th Annual Conference of the International Society for Clinical Biostatistics (ISCB) in Leuven, Belgium and our long‐standing collaborations with Doug, we discuss his contributions to regression modeling, reporting, prognosis research, as well as some more general issues while acknowledging that we cannot cover the whole spectrum of Doug's considerable methodological output. His statement “To maximize the benefit to society, you need to not just do research but do it well” should be a driver for all researchers. To improve current and future research, we aim to summarize Doug's messages for these three topics.
- Open Access
Risk of disease recurrence and survival in patients with multiple myeloma: A German Study Group analysis using a conditional survival approach with long‐term follow‐up of 815 patients
Background: Unlike the traditional method of overall survival prediction in patients with cancer, conditional survival predicts the survival of patients dynamically throughout the course of disease, identifying how a prognosis evolves over time.Methods: The authors assessed 815 consecutive patients with multiple myeloma through the German Study Group on Multiple Myeloma (Deutsche Studiengruppe Multiples Myelom; DSMM) incentive. Over 10 variables, including patient‐specific and multiple myeloma–specific parameters, were analyzed at the time of initial diagnosis and repeatedly during follow‐up. The probability of survival for another 5 years was calculated according to disease‐related and host‐related risks. Multivariate Cox models were used to determine baseline and updated prognostic factors for survival.Results: The median follow‐up and overall survival were 10.3 years and 5.1 years, respectively. When comparing 5‐year conditional survival probabilities from the data derived at the time of initial diagnosis with those updated over time, substantially differing prognoses were observed when follow‐up data were used. Multivariate Cox regression models for cohorts surviving 0 to 5 years demonstrated hazard ratios (HRs) for patients aged 70 years of 1, 1.68, and 3.17, respectively. These HRs for age were found to decline for patients surviving 5 years, as well as for those with advanced stages of disease (II/III) and unfavorable cytogenetics, whereas progressive disease remained an important factor in patients surviving 1 year, 3 years, and 5 years, with HRs of 1.85, 2.11, and 2.14, respectively.Conclusions: To the authors' knowledge, the current study is the first analysis of conditional survival in patients with multiple myeloma using both baseline and follow‐up risk parameters, demonstrating that regular risk assessment throughout the course of disease and complete follow‐up provide a more reliable conditional survival estimation than baseline assessment alone.
Open AccessRisk of disease recurrence and survival in patients with multiple myeloma: A German Study Group analysis using a conditional survival approach with long‐term follow‐up of 815 patients
Background: Unlike the traditional method of overall survival prediction in patients with cancer, conditional survival predicts the survival of patients dynamically throughout the course of disease, identifying how a prognosis evolves over time.Methods: The authors assessed 815 consecutive patients with multiple myeloma through the German Study Group on Multiple Myeloma (Deutsche Studiengruppe Multiples Myelom; DSMM) incentive. Over 10 variables, including patient‐specific and multiple myeloma–specific parameters, were analyzed at the time of initial diagnosis and repeatedly during follow‐up. The probability of survival for another 5 years was calculated according to disease‐related and host‐related risks. Multivariate Cox models were used to determine baseline and updated prognostic factors for survival.Results: The median follow‐up and overall survival were 10.3 years and 5.1 years, respectively. When comparing 5‐year conditional survival probabilities from the data derived at the time of initial diagnosis with those updated over time, substantially differing prognoses were observed when follow‐up data were used. Multivariate Cox regression models for cohorts surviving 0 to 5 years demonstrated hazard ratios (HRs) for patients aged 70 years of 1, 1.68, and 3.17, respectively. These HRs for age were found to decline for patients surviving 5 years, as well as for those with advanced stages of disease (II/III) and unfavorable cytogenetics, whereas progressive disease remained an important factor in patients surviving 1 year, 3 years, and 5 years, with HRs of 1.85, 2.11, and 2.14, respectively.Conclusions: To the authors' knowledge, the current study is the first analysis of conditional survival in patients with multiple myeloma using both baseline and follow‐up risk parameters, demonstrating that regular risk assessment throughout the course of disease and complete follow‐up provide a more reliable conditional survival estimation than baseline assessment alone.
- Open Access
- 30 March 2020
- Published by Wiley in Biometrical Journal
- Vol. 62 (3) , 568-582
Comparison of complex modeling strategies for prediction of a binary outcome based on a few, highly correlated predictors
Motivated by a clinical prediction problem, a simulation study was performed to compare different approaches for building risk prediction models. Robust prediction models for hospital survival in patients with acute heart failure were to be derived from three highly correlated blood parameters measured up to four times, with predictive ability having explicit priority over interpretability. Methods that relied only on the original predictors were compared with methods using an expanded predictor space including transformations and interactions. Predictors were simulated as transformations and combinations of multivariate normal variables which were fitted to the partly skewed and bimodally distributed original data in such a way that the simulated data mimicked the original covariate structure. Different penalized versions of logistic regression as well as random forests and generalized additive models were investigated using classical logistic regression as a benchmark. Their performance was assessed based on measures of predictive accuracy, model discrimination, and model calibration. Three different scenarios using different subsets of the original data with different numbers of observations and events per variable were investigated. In the investigated setting, where a risk prediction model should be based on a small set of highly correlated and interconnected predictors, Elastic Net and also Ridge logistic regression showed good performance compared to their competitors, while other methods did not lead to substantial improvements or even performed worse than standard logistic regression. Our work demonstrates how simulation studies that mimic relevant features of a specific data set can support the choice of a good modeling strategy.
- 30 March 2020
- Published by Wiley in Biometrical Journal
- Vol. 62 (3) , 568-582
Open AccessComparison of complex modeling strategies for prediction of a binary outcome based on a few, highly correlated predictors
Motivated by a clinical prediction problem, a simulation study was performed to compare different approaches for building risk prediction models. Robust prediction models for hospital survival in patients with acute heart failure were to be derived from three highly correlated blood parameters measured up to four times, with predictive ability having explicit priority over interpretability. Methods that relied only on the original predictors were compared with methods using an expanded predictor space including transformations and interactions. Predictors were simulated as transformations and combinations of multivariate normal variables which were fitted to the partly skewed and bimodally distributed original data in such a way that the simulated data mimicked the original covariate structure. Different penalized versions of logistic regression as well as random forests and generalized additive models were investigated using classical logistic regression as a benchmark. Their performance was assessed based on measures of predictive accuracy, model discrimination, and model calibration. Three different scenarios using different subsets of the original data with different numbers of observations and events per variable were investigated. In the investigated setting, where a risk prediction model should be based on a small set of highly correlated and interconnected predictors, Elastic Net and also Ridge logistic regression showed good performance compared to their competitors, while other methods did not lead to substantial improvements or even performed worse than standard logistic regression. Our work demonstrates how simulation studies that mimic relevant features of a specific data set can support the choice of a good modeling strategy.
- Open Access
- 4 March 2020
- Published by Wiley in Biometrical Journal
- Vol. 62 (3) , 790-807
On the relation between the cause‐specific hazard and the subdistribution rate for competing risks data: The Fine–Gray model revisited
The Fine–Gray proportional subdistribution hazards model has been puzzling many people since its introduction. The main reason for the uneasy feeling is that the approach considers individuals still at risk for an event of cause 1 after they fell victim to the competing risk of cause 2. The subdistribution hazard and the extended risk sets, where subjects who failed of the competing risk remain in the risk set, are generally perceived as unnatural . One could say it is somewhat of a riddle why the Fine–Gray approach yields valid inference. To take away these uneasy feelings, we explore the link between the Fine–Gray and cause‐specific approaches in more detail. We introduce the reduction factor as representing the proportion of subjects in the Fine–Gray risk set that has not yet experienced a competing event. In the presence of covariates, the dependence of the reduction factor on a covariate gives information on how the effect of the covariate on the cause‐specific hazard and the subdistribution hazard relate. We discuss estimation and modeling of the reduction factor, and show how they can be used in various ways to estimate cumulative incidences, given the covariates. Methods are illustrated on data of the European Society for Blood and Marrow Transplantation.
- 4 March 2020
- Published by Wiley in Biometrical Journal
- Vol. 62 (3) , 790-807
Open AccessOn the relation between the cause‐specific hazard and the subdistribution rate for competing risks data: The Fine–Gray model revisited
The Fine–Gray proportional subdistribution hazards model has been puzzling many people since its introduction. The main reason for the uneasy feeling is that the approach considers individuals still at risk for an event of cause 1 after they fell victim to the competing risk of cause 2. The subdistribution hazard and the extended risk sets, where subjects who failed of the competing risk remain in the risk set, are generally perceived as unnatural . One could say it is somewhat of a riddle why the Fine–Gray approach yields valid inference. To take away these uneasy feelings, we explore the link between the Fine–Gray and cause‐specific approaches in more detail. We introduce the reduction factor as representing the proportion of subjects in the Fine–Gray risk set that has not yet experienced a competing event. In the presence of covariates, the dependence of the reduction factor on a covariate gives information on how the effect of the covariate on the cause‐specific hazard and the subdistribution hazard relate. We discuss estimation and modeling of the reduction factor, and show how they can be used in various ways to estimate cumulative incidences, given the covariates. Methods are illustrated on data of the European Society for Blood and Marrow Transplantation.