Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Datasets as Topic”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9Linked to original sources
PubMed1999

Seamless multiresolution display of portable wavelet-compressed images.

Image storage, display, and distribution have been difficult problems in radiology for many years. As improvements in technology have changed the nature of the storage and display media, demand for image portability, faster image acquisition, and flexible image distribution is driving the development of responsive systems. Technology, such as the wavelet-based multiresolution seamless image database (MrSID) portable image format (PIF), is enabling image management solutions that address the shifting "point-of-care." The MrSID PIF employs seamless, multiresolution technology, which allows the viewer to determine the size of the image to be viewed, as well as the position of the viewing area within the image dataset. In addition the MrSID PIF allows control of the compression ratio of decompressed images. This capability offers the advantage of very rapid image recall from storage devices and portability for rapid transmission and distribution using the internet or wide-area networks. For example, in teleradiology, the radiologist or other physician desiring to view images at a remote location has full flexibility in being able to choose a quick display of an overview image, a complete display of a full diagnostic quality image, or both without compromising communication bandwidth. The MrSID algorithm will satisfy Joint Photographic Experts Group (JPEG) 2000 standards, thereby being compatible with future versions of the Digital Imaging and Communications in Medicine (DICOM) standard for image data compression.

Algorithms↗
PubMed2001

Discovering common stem-loop motifs in unaligned RNA sequences.

Post-transcriptional regulation of gene expression is often accomplished by proteins binding to specific sequence motifs in mRNA molecules, to affect their translation or stability. The motifs are often composed of a combination of sequence and structural constraints such that the overall structure is preserved even though much of the primary sequence is variable. While several methods exist to discover transcriptional regulatory sites in the DNA sequences of coregulated genes, the RNA motif discovery problem is much more difficult because of covariation in the positions. We describe the combined use of two approaches for RNA structure prediction, FOLDALIGN and COVE, that together can discover and model stem-loop RNA motifs in unaligned sequences, such as UTRs from post-transcriptionally coregulated genes. We evaluate the method on two datasets, one a section of rRNA genes with randomly truncated ends so that a global alignment is not possible, and the other a hyper-variable collection of IRE-like elements that were inserted into randomized UTR sequences. In both cases the combined method identified the motifs correctly, and in the rRNA example we show that it is capable of determining the structure, which includes bulge and internal loops as well as a variable length hairpin loop. Those automated results are quantitatively evaluated and found to agree closely with structures contained in curated databases, with correlation coefficients up to 0.9. A basic server, Stem-Loop Align SearcH (SLASH), which will perform stem-loop searches in unaligned RNA sequences, is available at http://www.bioinf.au.dk/slash/.

Algorithms↗
PubMed1995

Commentary and opinion: III. Some nonontological and functionally unconnected views on current issues in the analysis of PET datasets.

Strother et al. (1995) and Friston (1995) both raise important issues and provide useful reviews of various aspects of PET data analysis. Statisticians would not assume that any single piece of methodology would answer all questions about a type of data in a variety of experimental and observational contexts. The fundamental importance of hypothesis-driven inference, based on well designed experiments, cannot be overestimated for its ability to progress scientific understanding in an orderly manner. However, hypothesis-generating experiments are also vital in their own right. In practice, we generally do not have the luxury of both types of experiment, and we should note Strother et al.'s comment on the importance of extracting as much information as possible from each dataset. Friston (1995) also sees formal testing methods and exploratory methods such as principal components analysis as complementary. The correct approach would therefore seem to be (a) to select methods for formal and exploratory data analysis from the rich existing tool kit of statistical procedures, (b) to modify these as necessary to deal with special PET problems such as multiplicity, (c) to be aware of the assumptions underlying the methods being used and to investigate the problems that can arise if these assumptions fail to hold, (d) to appreciate the complexity of both PET data and of the potential questions that can be asked of it, and (e) to be aware of the limitations of any statistical analysis and the need for caution in interpreting conclusions not based on any predefined hypothesis.

Analysis of Variance↗
PubMed1990

A comparison of infection control software for use by hospital epidemiologists in meeting the new JCAHO standards.

To choose a microcomputer software package for our hospital epidemiology division, the two leading commercial software packages for infection control, AICE (ICPA, Inc., Austin, Texas) and NOS0-3 (Epi Systematics, Inc., Ft. Meyers, Florida), were compared for the types of epidemiologic analysis likely to be required to satisfy new Joint Commission on Accreditation of Healthcare Organizations (JCAHO) 1990 Infection Control Standards. The test dataset was a surgical database of 3,235 operations with 292 (9%) wound infections. Though NOSO-3 was more flexible in terms of the amount of data items one could record, it required seven times longer to learn, nine times more disk space to store and two times as long to enter cases than AICE. Six simple infection control reports (i.e., line listings, crosstabulations, stratified rates and graphs) required only seven computing steps and approximately 11 minutes to process with AICE, but 22 steps and over two hours with NOSO-3. All analytic results from AICE agreed with the results obtained with the Statistical Analysis System (SAS, SAS Institute, Inc., Cary, North Carolina), but analyses such as service-specific rates performed with NOSO-3 differed because of a design flaw in the NOSO-3 data structure.

Cross Infection↗
PubMed2001

Maps of the brain.

We review recent developments in brain mapping and computational anatomy that have greatly expanded our ability to analyze brain structure and function. The enormous diversity of brain maps and imaging methods has spurred the development of population-based digital brain atlases. These atlases store information on how the brain varies across age and gender, across time, in health and disease, and in large human populations. We describe how brain atlases, and the computational tools that align new datasets with them, facilitate comparison of brain data across experiments, laboratories, and from different imaging devices. The major methods are presented for the construction of probabilistic atlases, which store information on anatomic and functional variability in a population. Algorithms are reviewed that create composite brain maps and atlases based on multiple subjects. We show that group patterns of cortical organization, asymmetry, and disease-specific trends can be resolved that may not be apparent in individual brain maps. Finally, we describe the creation of four-dimensional (4D) maps that store information on the dynamics of brain change in development and disease. Digital atlases that correlate these maps show considerable promise in identifying general patterns of structural and functional variation in human populations, and how these features depend on demographic, genetic, cognitive, and clinical parameters.

Algorithms↗
PubMed1999

Short Form 36 (SF-36) Health Survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, the Health Survey for England and the Oxford Healthy Life Survey.

BACKGROUND: Population norms for the attributes included in measurement scales are required to provide a standard with which scores from other study populations can be compared. This study aimed to obtain population norms for the Short Form 36 (SF-36) Health Survey Questionnaire, derived from a random sample of the population in Britain who were interviewed at home, and to make comparisons with other commonly used norms. METHODS: The method was a face-to-face interview survey of a random sample of 2056 adults living at home in Britain (response rate 78 per cent). Comparisons of the SF-36 scores derived from this sample were made with the Health Survey for England and the Oxford Healthy Life Survey. RESULTS: Controlling for age and sex, many of mean scores on the SF-36 dimensions differed between the three datasets. The British interview sample had better total means for Physical Functioning, Social Functioning, Mental Health, Energy/Vitality, and General Health Perceptions. The Health (interview) Survey for England had the lowest (worst) total mean scores for Physical Functioning, Social Functioning, Role Limitations (physical), Bodily Pain, and Health Perceptions. The postal sample in central England had the lowest (worst) total mean scores for Role Limitations (emotional), Mental Health and Energy/Vitality. CONCLUSION: Responses obtained from interview methods may suffer more from social desirability bias (resulting in inflated SF-36 scores) than postal surveys. Differences in SF-36 means between surveys are also likely to reflect question order and contextual effects of the questionnaires. This indicates the importance of providing mode-specific population norms for the various methods of questionnaire administration.

Adolescent↗
PubMed2000

Methodological issues in the use of guidelines and audit to improve clinical effectiveness in breast cancer in one United Kingdom health region.

AIMS: To develop a system to improve and monitor clinical performance in the management of breast cancer patients in one United Kingdom health region. DESIGN: An observational study of the changes brought about by the introduction of new structures to influence clinical practice and monitor change. SETTING: North Thames (East) Health region, comprising seven purchasing health authorities and 21 acute hospitals treating breast cancer. SUBJECTS: The multi-disciplinary breast teams in 21 hospitals and an audit sample of 419 (28%) of the breast cancer patients diagnosed in 1992 in the region. INTERVENTIONS: Evidence-based interventions for changing clinical practice: regional guidelines, senior clinicians acting as < >, audit of quality rather than cost of services, ownership of data by clinicians, confidential feed-back to participants and education. OUTCOME MEASURES: Qualitative measures of organizational and behavioural change. Quantitative measures of clinical outcomes compared to guideline targets and to results from previous studies within this population. RESULTS: Organizational changes included the involvement, participation of and feedback to 16 specialist surgeons and their multidisciplinary teams in 21 hospitals. Regional clinical guidelines were developed in 6 months and the dataset piloted within 9 months. The audit cycle was completed within 2 years. The pilot study led to prospective audit at the end of 2 years for all breast cancers in the region and a 15-fold increase in high quality clinical information for these patients. Changes in clinical practice between 1990 and 1992 were observed in the use of chemotherapy (up from 17-23%) and axillary surgery (up from 46-76%). CONCLUSIONS: The approach used facilitated rapid change and found a balance between local involvement (essential for sustainability within a hospital setting) and regional standardization (essential for comparability across hospitals). The principles of the approach are generalized to other cancers and to other parts of the UK and abroad.

Breast Neoplasms↗
PubMed1998

The distribution of health care costs and their statistical analysis for economic evaluation.

OBJECTIVE: Where patient level data are available on health care costs, it is natural to use statistical analysis to describe the differences in cost between alternative treatments. Health care costs are, however, commonly considered to be skewed, which could present problems for standard statistical tests. This review examines how authors report the distributional form of health care cost data and how they have analysed their results. METHOD: A review of cost-effectiveness studies that collected patient-level data on health care costs. To supplement the review, five datasets on health care costs are examined. Consideration is given to the use of parametric methods on the transformed scale and to non-parametric methods of analysing skewed cost data. RESULTS: Since economic analysis requires estimation in monetary units, the usefulness of transformation-based methods is limited by the inability to retransform cost differences to the original scale. Non-parametric rank sum methods were also found to be of limited use for economic analysis, partly due to the focus on hypothesis testing rather than estimation. Overall, the non-parametric approach of bootstrapping was found to offer a useful test of the appropriateness of parametric assumptions and an alternative method of estimation where those assumptions were found not to hold. CONCLUSIONS: Guidelines for the analysis of skewed health care cost data are offered.

Cost-Benefit Analysis↗
PubMed1997

Brainvox: an interactive, multimodal visualization and analysis system for neuroanatomical imaging.

A study of cognition emerging from a neurobiological perspective, as opposed to one emerging from a purely computational or psychological perspective, begins with observations of the human brain in normal and pathological states and is furthered by the investigation of hypotheses which are articulated using neuroanatomical nomenclature. Brainvox is an interactive three-dimensional brain imaging software package designed to permit such research through the support of the description and quantification of brain pathology in magnetic resonance images and of the experimental investigation of human cognition in lesion and functional imaging studies. Important general features of Brainvox, for these purposes, are: (1) adaptation of volume rendering for brain lesions and for corendered datasets; (2) shared memory architecture, which enables the user to identify and label anatomical structures, while inspecting the brain in multiple views simultaneously; (3) modular program design, including interlocking command-line utilities, which make Brainvox extensible and empower users without programming expertise to implement new analysis techniques through Unix shell scripting; and (4) full integration of three-dimensional tools for visualization with tools for analysis. Specific features include a new object templating technique (MAP-3) for studies of groups of brain-lesioned subjects, a complete and extensible suite of command-line processing utilities, a three-dimensional optimal graph-searching tool, and a method for planning PET slices and matching MR and PET slices (MP_FIT).

Artificial Intelligence↗
PubMed1999

Impact on quality of life during an allergen challenge research trial.

BACKGROUND: Quality of life (QOL) issues resulting from participation in an allergy research trial, or indeed any clinical trial, is not documented in the medical literature. OBJECTIVE: To determine whether participating in a trial where allergic symptoms are induced has a significant impact on subjects' QOL, and to quantify extent and duration. METHODS: Subjects were recruited from a trial utilizing a controlled allergen environment to assess anti-allergic medications. A QOL survey (consisting of the Rhinoconjunctivitis Quality of Life Questionnaire [RQLQ] & the SF-36) was completed at screening, on study day, and approximately 2 weeks post-study. Follow-up was sought from subjects' whose QOL was significantly worse than baseline. RESULTS: Of 219 trial participants, 206 completed both screening and study surveys; 141 returned at least one follow-up survey; and 136 constructed the final dataset. Mean overall scores at follow-up via RQLQ were significantly better than screening (P < .001). Significant decreases in QOL from baseline on study day occurred in social function on the SF-36 (P = .026) and in domains of sleep (P = .019), non-nasal symptoms (P = .05), ocular symptoms (P < .001), and nasal symptoms (P < .001) on the RQLQ. Average post-study follow-up was 17.1 days (range = 5 to 55 days). CONCLUSION: Subjects participating in a trial involving allergic symptom induction experienced a decrease of QOL in parameters specific to rhinoconjunctivitis and social function. Subjects' QOL returned to or improved over baseline within 2 1/2 weeks. Positive QOL findings are important to studies where symptoms are induced and also have relevance to standard Phase 3 drug trials.

Allergens↗
PubMed2001

What is bioinformatics? A proposed definition and overview of the field.

BACKGROUND: The recent flood of data from genome sequences and functional genomics has given rise to new field, bioinformatics, which combines elements of biology and computer science. OBJECTIVES: Here we propose a definition for this new field and review some of the research that is being pursued, particularly in relation to transcriptional regulatory systems. METHODS: Our definition is as follows: Bioinformatics is conceptualizing biology in terms of macromolecules (in the sense of physical-chemistry) and then applying "informatics" techniques (derived from disciplines such as applied maths, computer science, and statistics) to understand and organize the information associated with these molecules, on a large-scale. RESULTS AND CONCLUSIONS: Analyses in bioinformatics predominantly focus on three types of large datasets available in molecular biology: macromolecular structures, genome sequences, and the results of functional genomics experiments (e.g. expression data). Additional information includes the text of scientific papers and "relationship data" from metabolic pathways, taxonomy trees, and protein-protein interaction networks. Bioinformatics employs a wide range of computational techniques including sequence and structural alignment, database design and data mining, macromolecular geometry, phylogenetic tree construction, prediction of protein structure and function, gene finding, and expression data clustering. The emphasis is on approaches integrating a variety of computational methods and heterogeneous data sources. Finally, bioinformatics is a practical discipline. We survey some representative applications, such as finding homologues, designing drugs, and performing large-scale censuses. Additional information pertinent to the review is available over the web at http://bioinfo.mbb.yale.edu/what-is-it.

Computational Biology↗
PubMed1992

Evaluation of a 3D reconstruction algorithm for multi-slice PET scanners.

A fully 3D reconstruction algorithm based on filtered backprojection was evaluated for the reconstruction of data obtained with multi-slice positron emission tomography (PET) scanners which have had the septa removed. This algorithm uses forward-projection through the reconstructed images of a 2D subset of the data to complete the 3D dataset thus satisfying the condition of shift invariance. This is followed by 3D filtered backprojection. Axial sampling was doubled by combining adjacent polar angles, thus improving reconstructed axial resolution. The algorithm was tested using real and simulated datasets and gave high quality reconstructions without artifacts over a wide range of imaging conditions. Events are placed accurately throughout the imaging volume as determined by measurements with a MRI/PET registration phantom. The forward-projection step leads to degradation in image resolution due to insufficient axial and transaxial sampling. This effect is amplified if multiple iterations of the algorithm are used, with little decrease in image noise. Changing the filter employed in the initial 2D reconstruction can be used to alter the noise and resolution characteristics of the 3D images. This algorithm has proved very robust at reconstructing 3D PET data and is relatively fast. Those small problems which exist can be attributed to detector sampling problems, especially in the axial direction, which is a consequence of the geometry of these scanners, which are designed primarily for 2D data acquisition.

Algorithms↗
PubMed2001

Determinants of patient recruitment in a multicenter clinical trials group: trends, seasonality and the effect of large studies.

BACKGROUND: We examined whether quarterly patient enrollment in a large multicenter clinical trials group could be modeled in terms of predictors including time parameters (such as long-term trends and seasonality), the effect of large trials and the number of new studies launched each quarter. We used the database of all clinical studies launched by the AIDS Clinical Trials Group (ACTG) between October 1986 and November 1999. Analyses were performed in two datasets: one included all studies and substudies (n = 475, total enrollment 69,992 patients) and the other included only main studies (n = 352, total enrollment 57,563 patients). RESULTS: Enrollment differed across different months of the year with peaks in spring and late fall. Enrollment accelerated over time (+27 patients per quarter for all studies and +16 patients per quarter for the main studies, p < 0.001) and was affected by the performance of large studies with target sample size > 1,000 (p < 0.001). These relationships remained significant in multivariate autoregressive modeling. A time series based on enrollment during the first 32 quarters could forecast adequately the remaining 21 quarters. CONCLUSIONS: The fate and popularity of large trials may determine the overall recruitment of multicenter groups. Modeling of enrollment rates may be used to comprehend long-term patterns and to perform future strategic planning.

Clinical Trials as Topic↗
PubMed2000

Sample size calculations for cluster randomised trials. Changing Professional Practice in Europe Group (EU BIOMED II Concerted Action).

OBJECTIVES: Cluster randomised trials, in which groups of individuals are randomised, are increasingly being used in the health field. Adopting a clustered approach has implications for the design of such trials, and sample size calculations need to be inflated to accommodate for the clustering effect. Reliable estimates of intracluster correlation coefficients (ICCs) are required for robust sample size calculations to be made; however, little empirical evidence is available on their likely size, and on factors which influence their magnitude. The aim of this study was to generate empirical estimates of ICCs and to explore factors which may affect their magnitude. METHODS: Empirical estimates of ICCs were calculated for both process variables and patient outcomes from a number of datasets of primary and secondary care implementation studies. RESULTS: Estimates of ICCs varied according to setting and type of outcome. Estimates of ICCs for process variables were higher than those for patient outcomes, and estimates derived from secondary care were higher than those from primary care. ICCs for process variables in primary care were of the order of 0.05-0.15, whilst those in secondary care were of the order of 0.3. Estimates for patient outcomes in primary care were generally lower than 0.05. CONCLUSIONS: Adopting cluster randomisation has implications for the design, size and analysis of clinical trials. This study gives an insight into the potential size of ICCs in primary and secondary care, and provides a practical guide to researchers to aid the planning of future studies in this area.

Cluster Analysis↗
PubMed1996

Validation testing of the SEER real-time digital holter monitor.

OBJECTIVES: Perioperative myocardial ischemia, detected by off-line Holter ST-segment monitoring, has been associated with adverse cardiac outcome. Technical advances in digital signal processing have facilitated development of digital Holter recorders that allow 24- to 48-hour recording, full disclosure storage, and "real-time" quantitative analysis of ST-segment levels. These recorders may be useful for "on-line" clinical detection of perioperative ischemia. However, little data are available, independent of manufacturers' claims, to validate their accuracy. Using a previously validated digital electrocardiogram (ECG) simulator, a commercially available device was evaluated. DESIGN: Laboratory bench study. SETTING: Not applicable. PARTICIPANTS: Not applicable. INTERVENTIONS: Not applicable. MEASUREMENTS AND MAIN RESULTS: Custom digital ECG waveform templates were programmed for use with a commercially available ECG simulator (M311 ECG simulator; Fogg Systems, Inc, Aurora, CO). For each template, ST-segment morphology (horizontal elevation or depression, downsloping depression), QRS duration (80 v 120 msec) and the presence or absence of a P wave were manipulated, yielding six unique QRS shapes. For each shape, the degree of ST-segment deviation was altered over a wide range. ST-segment values from the simulator (measured at 60 msec after the J point) ranged from +10 to -18 mm. The SEER digital Holter recorder (Marquette Electronics, Milwaukee, WI) was tested. One hundred twenty-six measurements of ST-segment deviation were input to the SEER at each of two testing sessions. The ST-segment value from the recorder in the "noninteractive" analysis mode was obtained, and the two results averaged for comparison with the expected simulator value. Variability of ST-segment measurement over a continuous 1-hour period of simulator input was also assessed. Sixty-seven percent of measurements were within 95% to 100% of expected, whereas 90% were within 90% to 110%. The regression equation for the complete dataset was SEER output (mm) = -0.47 + 1.015 * simulator input, R2 = 0.99. The mean observed-to-expected value ratio was 100% +/- 6% (+/-SD), range 80% to 114%. The mean deviation in millimeters from expected for all measurements was 0.10 +/- 0.20 mm, median 0.05 mm, range -0.25 to +0.60 mm. For the 72 measurements obtained by 5-minute sampling over 1 hour of continuous simulator input for each of the six QRS shapes, the mean percent difference between observed and expected values was 0.5% +/- 4.5%, median 0.0%, with a mean coefficient of variation of 2.7% (median 1.9%). CONCLUSIONS: Using a digital ECG simulator, it was found that the SEER recorder analyzed ST-segment deviation with a high degree of accuracy. These findings, along with its full disclosure reporting capabilities, suggest it may be useful in perioperative risk stratification. However, accuracy in the clinical setting remains to be validated.

Electrocardiography, Ambulatory↗
PubMed1997

Statistical analysis of highly skewed immune response data.

This paper considers methods of statistical analysis for highly skewed immune response data. Observations from population studies of immunological variables are rarely normally distributed between individuals; typically the distribution shows extreme levels of skewness. In some situations, skewness remains considerable even after transforming the data. Using resampling techniques, applied to several actual datasets of ELISA assay data, we consider the robustness of normal parametric methods, e.g. t tests and linear regression. Despite the skewness of the transformed data, we demonstrate that such methods are quite robust depending on the number of observations, type of analysis and severity of skewness. We also illustrate how bootstrap resampling can be used to provide a valid alternative method of analysis that can be used either for checking normal parametric analysis or as a direct method of analysis. We illustrate this combined approach by analysing real data to test for association between human serum antibodies to malaria merozoite surface proteins, MSP1 and MSP2, and resistance to clinical malaria, and confirm the protective effect of antibodies to MSP1 and demonstrated a similar protective effect for some antibodies to MSP2.

Animals↗
PubMed2000

Clinical and economic outcomes assessment in nuclear cardiology.

The future of nuclear medicine procedures, as understood within our current economic climate, depends upon its ability to provide relevant clinical information at similar or lower comparative costs. With an ever-increasing emphasis on cost containment, outcome assessment forms the basis of preserving the quality of patient care. Today, outcomes assessment encompasses a wide array of subjects including clinical, economic, and humanistic (i.e., quality of life) outcomes. For nuclear cardiology, evidence-based medicine would require a threshold level of evidence in order to justify the added cost of any test in a patient's work-up. This evidence would include large multicenter, observational series as well as randomized trial data in sufficiently large and diverse patient populations. The new movement in evidence-based medicine is also being applied to the introduction of new technologies, in particular when comparative modalities exist. In the past 5 years, we have seen a dramatic shift in the quality of outcomes data published in nuclear cardiology. This includes the use of statistically rigorous risk-adjusted techniques as well as large populations (i.e., > 500 patients) representing multiple diverse medical care settings. This has been the direct result of the development of multiple outcomes databases that have now amassed thousands of patients worth of data. One of the benefits of examining outcomes in large patient datasets is the ability to assess individual endpoints (e.g., cardiac death) as compared with smaller datasets that often assess combined endpoints (e.g., death, myocardial infarction, or unstable angina). New technologies for the diagnosis of coronary artery disease have contributed to the rising costs of care. In the United States and in Europe, costs of care have risen dramatically, consuming an ever-increasing amount of available resources. The overuse of diagnostic angiography often leads to unnecessary revascularization that does not lead to improvement in outcome. Thus, the potential exists that stress SPECT imaging, a highly effective diagnostic tool, could effect substantial change in reducing inappropriate use of an invasive procedure resulting in cost effective cardiac care. A synthesis of current economic evidence in gated SPECT imaging will be presented. In conclusion, a current state of the evidence review is presented on the clinical and economic data using nuclear cardiology imaging.

Angiography↗
PubMed1996

Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.

Multivariable regression models are powerful tools that are used frequently in studies of clinical outcomes. These models can use a mixture of categorical and continuous variables and can handle partially observed (censored) responses. However, uncritical application of modelling techniques can result in models that poorly fit the dataset at hand, or, even more likely, inaccurately predict outcomes on new subjects. One must know how to measure qualities of a model's fit in order to avoid poorly fitted or overfitted models. Measurement of predictive accuracy can be difficult for survival time data in the presence of censoring. We discuss an easily interpretable index of predictive discrimination as well as methods for assessing calibration of predicted survival probabilities. Both types of predictive accuracy should be unbiasedly validated using bootstrapping or cross-validation, before using predictions in a new data series. We discuss some of the hazards of poorly fitted and overfitted regression models and present one modelling strategy that avoids many of the problems discussed. The methods described are applicable to all regression models, but are particularly needed for binary, ordinal, and time-to-event outcomes. Methods are illustrated with a survival analysis in prostate cancer using Cox regression.

Clinical Trials as Topic↗