Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Dataset”

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 1,387 records · Page 77Linked to original sources
PubMed1992

Reliability of data sources for poisoning deaths in Massachusetts.

The authors analyzed poisoning-related deaths in Massachusetts from 1986 and 1987 recorded in three datasets: poison center records, death certificate, and state medical examiner's office records. While 714 such deaths were found, 551 of these were prehospital deaths recorded within the medical examiner's office but not by the poison center. The poison center was not consulted in over 47% of the poisoning deaths occurring in Massachusetts hospitals. Conversely, 15% of deaths were reported to the poison center but were not found either in death certificate or medical examiner records. Concordance between all three datasets for recording the 163 poisoning-related hospital deaths was only 17%. The authors conclude that reliance on a single data source underestimates and potentially misrepresents both the numbers and types of poisoning deaths occurring in the state. They also believe the files of the medical examiner are an underappreciated, rich source of data concerning out-of-hospital deaths due to poisonings and intoxications. Their findings suggest that the regional poison center is an underused resource for the management of seriously poisoned patients. There is a need for a better working relationship between poison centers and area hospitals so that all serious intoxications and poisonings are reported to the poison center in a timely fashion.

Carbon Monoxide Poisoning↗
PubMed1990

Identification of high risk groups for breast cancer by means of logistic models.

In order to identify high risk groups for breast cancer, unconditional multiple logistic regression models based on 5 widely recognized and easily identifiable risk factors (age at menarche, at menopause and at first birth, family history of breast cancer and body mass index) were applied to a large dataset including 2085 cases and 1936 controls aged 50 or over derived from two unmatched hospital-based case-control studies conducted in Italy. Although various models provided an excellent fitting, both on the whole dataset and using a training-testing approach to a priori identified separate subset, the observed extent of variation in breast cancer risk between highest and lowest decile of the distribution was limited to a factor 2. This indicates that the 5 variables considered did not allow identification of subgroups with substantially elevated risk of breast cancer to have practical implications for screening/prophylactic treatment purposes.

Adolescent↗
PubMed1994

Comparison of chemotherapy and bone marrow transplants using two independent clinical databases.

Comparing the outcome of chemotherapy and bone marrow transplants in the absence of a randomized trial is difficult but necessary for diseases where small numbers of patients make such trials difficult if not impossible. To address this issue for adults with acute lymphoblastic leukemia in first remission, we created an empirical database using two separate datasets, one from the International Bone Marrow Transplant Registry and the other from two multicenter chemotherapy studies. Prior to combining the datasets, a study protocol was developed to define inclusion criteria, outcomes to be compared and statistical methods. The main problems of a non-randomized comparison are biases potentially introduced by differences in baseline composition of the two cohorts and differences in time-to-treatment. The source of the latter bias is different distributions of waiting times between achieving complete remission and receiving post-remission therapy. Several techniques to control these biases were evaluated; each gave qualitatively similar results. These methods can easily be applied to other clinical situations where randomized trials are not available.

Adolescent↗
PubMedDate not supplied

Use of general purpose mechanical computer assisted engineering software in orthopaedic surgical planning: advantages and limitations.

Two surgical plans were developed for an appropriately complex reconstructive orthopaedic surgery case. One plan was developed with customary methods using two-dimensional (2D) radiographs. The second plan was developed with general purpose mechanical computer assisted engineering (MCAE) software using x-ray computed tomography (CT) data. The limitations of each method are identified. To create a surgical plan using three-dimensional (3D) medical datasets and MCAE software, five necessary steps were identified: (a) data reduction; (b) contour extraction; (c) 3D model creation; (d) extraction of mass properties; (e) model idealization. The principal limitation of general purpose MCAE software is the lack of pre-processing modules with which to address the unique requirements of medical image datasets.

Acetabulum↗
PubMed2006

Modeling motor vehicle crashes using Poisson-gamma models: examining the effects of low sample mean values and small sample size on the estimation of the fixed dispersion parameter.

There has been considerable research conducted on the development of statistical models for predicting crashes on highway facilities. Despite numerous advancements made for improving the estimation tools of statistical models, the most common probabilistic structure used for modeling motor vehicle crashes remains the traditional Poisson and Poisson-gamma (or Negative Binomial) distribution; when crash data exhibit over-dispersion, the Poisson-gamma model is usually the model of choice most favored by transportation safety modelers. Crash data collected for safety studies often have the unusual attributes of being characterized by low sample mean values. Studies have shown that the goodness-of-fit of statistical models produced from such datasets can be significantly affected. This issue has been defined as the "low mean problem" (LMP). Despite recent developments on methods to circumvent the LMP and test the goodness-of-fit of models developed using such datasets, no work has so far examined how the LMP affects the fixed dispersion parameter of Poisson-gamma models used for modeling motor vehicle crashes. The dispersion parameter plays an important role in many types of safety studies and should, therefore, be reliably estimated. The primary objective of this research project was to verify whether the LMP affects the estimation of the dispersion parameter and, if it is, to determine the magnitude of the problem. The secondary objective consisted of determining the effects of an unreliably estimated dispersion parameter on common analyses performed in highway safety studies. To accomplish the objectives of the study, a series of Poisson-gamma distributions were simulated using different values describing the mean, the dispersion parameter, and the sample size. Three estimators commonly used by transportation safety modelers for estimating the dispersion parameter of Poisson-gamma models were evaluated: the method of moments, the weighted regression, and the maximum likelihood method. In an attempt to complement the outcome of the simulation study, Poisson-gamma models were fitted to crash data collected in Toronto, Ont. characterized by a low sample mean and small sample size. The study shows that a low sample mean combined with a small sample size can seriously affect the estimation of the dispersion parameter, no matter which estimator is used within the estimation process. The probability the dispersion parameter becomes unreliably estimated increases significantly as the sample mean and sample size decrease. Consequently, the results show that an unreliably estimated dispersion parameter can significantly undermine empirical Bayes (EB) estimates as well as the estimation of confidence intervals for the gamma mean and predicted response. The paper ends with recommendations about minimizing the likelihood of producing Poisson-gamma models with an unreliable dispersion parameter for modeling motor vehicle crashes.

Accidents, Traffic↗
PubMed2006

Determining the potential safety benefit of improved lighting in three pedestrian crash scenarios.

The influence of light level was determined for three pedestrian crash scenarios associated with three adaptive headlighting solutions-curve lighting, motorway lighting, and cornering light. These results were coupled to corresponding prevalence data for each scenario to derive measures of annual lifesaving potential. For each scenario, the risk associated with light level was determined using daylight saving time (DST) transitions to produce a dark/light interval risk ratio; prevalence was determined using the corresponding annual crash rate in darkness for each scenario. For curve lighting, pedestrian crashes on curved roadways were examined; for motorway lighting, crashes associated with high speed roadways were examined; and for cornering light, crashes involving turning vehicles at intersections were examined. In the curve analysis, lower dark/light crash ratios were observed for curved sections of roadway compared to straight roads. In the motorway analysis, posted speed limit was the dominant predictor of this ratio for the fatal crash dataset; road function class was the dominant predictor of the ratio for the fatal/nonfatal dataset. Finally, in the intersection crash analysis, the dark/light ratio for turning vehicles was lower than for nonturning vehicles; and the ratio at intersections was lower than at non-intersections. Relative safety need was determined by combining the dark/light ratio with prevalence data to produce an idealized measure of lifesaving potential. While all three scenarios suggested a potential for safety improvement, scenarios related to high speed roadway environments showed the greatest potential.

Accidents, Traffic↗
PubMed2004

Understanding interpretive errors in radiologists learning computed tomography colonography.

RATIONALE AND OBJECTIVES: To determine if interpretive errors in the course of learning CT colonography are secondary to failures in detection or in characterization and determine the types of lesions frequently missed. MATERIALS AND METHODS: Fifteen radiologists completed an electronic CTC training module consisting of two parts: 1) a teaching file demonstrating the varied appearances of polyps, cancers, and pitfalls in interpreting exams; and 2) a test of 50 complete CTC datasets. Following review of each test case, radiologists were asked to indicate if and where a polyp was visualized. The module then showed each neoplasm (if any) located within the dataset. For false negative examinations, radiologists indicated if the lesion was not seen, was seen but interpreted as colonic wall or fold, or was seen but interpreted as stool or fluid. RESULTS: The average sensitivity for sessile, pedunculated, and flat polyps for these novice readers was 76%, 63%, and 32%, respectively. Average sensitivity for all morphologies of cancers (annular, polypoid, flat) was high (93%, 85%, 95%), with 8/11 missed cancers being secondary to failure in detection. The most frequently missed cancer was an annular constricting tumor (5/11). Overall, 55% (73/132) of errors were failures of detection and 45% (59/132) were errors in characterization. CONCLUSION: Radiologists learning CT colonography had slightly more errors of detection than characterization, but this difference was not statistically significant. Flat and pedunculated polyps and annular constricting cancers were the most frequently missed morphologies. Examples of these abnormalities should be emphasized in CTC training programs.

Colonic Neoplasms↗
PubMed2004

The effect of varying user-selected input parameters on quantitative values in CT perfusion maps.

RATIONALE AND OBJECTIVES: Deconvolution-based software can be used to calculate quantitative maps of cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT) from first-pass computed tomography perfusion (CTP) datasets. The application of this software requires the user to select multiple input variables. The purpose of this study was to investigate the degree to which both major and minor variations of these user-defined inputs would affect the final quantitative values of CBF, CBV, and MTT. MATERIALS AND METHODS: A neuroradiologist constructed CBF, CBV, and MTT maps using standard methodology with commercially available software (GE Functool Version 1.9s) from CTP datasets of three acute stroke patients. Each map was reconstructed multiple times by systematically and independently varying the following parameters: postenhancement and preenhancement cutoff values, arterial and venous region-of-interest (ROI) placement, and arterial and venous ROI size. The resulting quantitative CTP values were compared using identical ROIs placed at the infarct core. RESULTS: Major variations of either arterial ROI placement or arterial and venous ROI size had no significant effect on the mean CBF, CBV, and MTT values at the infarct core (p > .05). Even minor variations, however, in the choice of venous ROI placement or in pre- and postenhancement cutoff values significantly altered the quantitative values for each of the CTP maps, by as much as threefold. CONCLUSION: Even minor variations of user-defined inputs can significantly influence the quantitative, deconvolution-based CTP map values of acute stroke patients. Although quantitation was robust to the choice of arterial ROI placement and arterial or venous ROI size, it was strongly dependent on the choice of venous ROI location and pre- and postenhancement cut-off values. Awareness of these results by clinicians may be important in the creation of quantitatively accurate CTP maps.

Adult↗
PubMed2005

Estimation and comparison of CAD system performance in clinical settings.

RATIONALE AND OBJECTIVES: Computer-aided detection (CAD) systems are frequently compared using free-response receiver operating characteristic (FROC) curves. While there are ample statistical methods for comparing FROC curves, when one is interested in comparing the outcomes of 2 CAD systems applied in a typical clinical setting, there is the additional matter of correctly determining the system operating point. This article shows how the effect of the sampling error on determining the correct CAD operating point can be captured. By incorporating this uncertainty, a method is presented that allows estimation of the probability with which a particular CAD system performs better than another on unseen data in a clinical setting. MATERIALS AND METHODS: The distribution of possible clinical outcomes from 2 artificial CAD systems with different FROC curves is examined. The sampling error is captured by the distribution of possible system thresholds of the classifying machine that yields a specified sensitivity. After introducing a measure of superiority, the probability of one system being superior to the other can be determined. RESULTS: It is shown that for 2 typical mammography CAD systems, each trained on independent representative datasets of 100 cases, the FROC curves must be separated by 0.20 false positives per image in order to conclude that there is a 90% probability that one is better than the other in a clinical setting. Also, there is no apparent gain in increasing the size of the training set beyond 100 cases. DISCUSSION: CAD systems for mammography are modeled for illustrative purposes, but the method presented is applicable to any computer-aided detection system evaluated with FROC curves. The presented method is designed to construct confidence intervals around possible clinical outcomes and to assess the importance of training set size and separation between FROC curves of systems trained on different datasets.

Algorithms↗
PubMed2005

Computer aided characterization of the solitary pulmonary nodule using volumetric and contrast enhancement features.

RATIONALE AND OBJECTIVES: To investigate the utility of a computer-aided diagnosis (CAD) in the task of differentiating malignant nodules from benign nodules based on quantitative features extracted from volumetric thin section CT image data acquired before and after the injection of contrast media. MATERIALS AND METHODS: 35 volumetric CT datasets of solitary pulmonary nodules (SPN) with proven diagnoses (19 malignant/16 benign) were contoured by a thoracic radiologist. All studies had at least a baseline series obtained without contrast media and at least one series following an intravenous contrast injection at 45, 90, 180, and 360 seconds. Two separate contours were created for each nodule: one including only the solid portion and another including the ground-glass component, if any, of the nodule. For each contour 31 features were calculated that measured the attenuation, shape, and enhancement of the nodule due to the injection of contrast. These features were input into a feature selection step and three different classifiers to determine if the diagnosis could be predicted from the resulting feature vector. In addition, observer input was introduced to two of the classifiers as an a priori probability of malignancy and the resulting performance was compared. Training and testing was conducted in a resubstitution and leave-one-out fashion and performance was evaluated using ROC analysis. RESULTS: In a leave-one-out testing methodology, the classifiers achieved areas under the ROC curves AZ that ranged from 0.69 to 0.92. A classifier based on logistic regression performed the best with an AZ of 0.92 while a classifier based on quadratic discriminant analysis performed the poorest (AZ, 0.69). The AZ increased when using a priori observer input in most cases reaching a maximum of 0.95. CONCLUSION: Based on this initial work with a limited number of nodules in our dataset, it appears that CAD using volumetric and contrast-enhanced data has the potential to assist radiologists in the task of differentiating solitary pulmonary nodules and in the management of these patients. Further studies with an increased number of patients are required to validate these results.

Adult↗
PubMed2005

Smoothing lung segmentation surfaces in three-dimensional X-ray CT images using anatomic guidance.

RATIONALE AND OBJECTIVES: Automatic lung segmentation in volumetric computed tomography (CT) images has been extensively investigated, and several methods have been proposed. Most methods distinguish the lung parenchyma from the surrounding anatomy based on the difference in CT attenuation values. This leads to an irregular and inconsistent lung boundary for the regions near the mediastinum, which can cause inconsistent boundaries both across subjects and within subjects scanned at different intervals of time. Processes like lung image registration and lung atlas construction can be affected by such inconsistencies. Therefore there is a need for a more consistent lung surface near the mediastinum. MATERIALS AND METHODS: This paper presents a fully automatic method for the three-dimensional smoothing of the lung boundary using information from the segmented human airway tree. First, using the segmented airway tree, we define a bounding box around the mediastinum for each lung, within which all operations are performed. We then define all generations of the airway tree distal to the right and left main stem bronchi to be part of the respective lungs and exclude all other segments. Finally, we perform a fast morphological closing with an ellipsoidal kernel to smooth the surface of the lung. RESULTS: This method has been tested by processing the segmented lungs from eight normal datasets. The mean value of the magnitude of curvedness of the lung contours, averaged over all the datasets, is 0.13 postsmoothing, compared with 4.91 before smoothing. The accuracy of the lung contours after smoothing is assessed by comparing the automatic results to manually traced smooth lung borders by a human analyst. Averaged over all volumes, the root mean square difference between human and computer borders is 0.8691 mm after smoothing, compared with 1.3012 mm before smoothing. The mean similarity index, which is an area overlap measure based on the kappa statistic, is 0.9958 (SD 0.0032), after smoothing. CONCLUSIONS: We have described a novel scheme for smoothing the lung contour around the mediastinum. The method is based on using anatomic information from the segmented airway tree. The validation results show that there is good agreement between manual and computer results. Because there are no accepted criteria for defining the lung boundary near the mediastinum, we believe our method of defining the boundary based on the structure of the airway tree provides a good basis for three-dimensional smoothing.

Algorithms↗
PubMed2006

Software tools for interactive instruction in radiologic anatomy.

RATIONALE AND OBJECTIVES: To promote active learning in an introductory Radiologic Anatomy course through the use of computer-based exercises. MATERIALS AND METHODS: DICOM datasets from our hospital PACS system were transferred to a networked cluster of desktop computers in a medical school classroom. Medical students in the Radiologic Anatomy course were divided into four small groups and assigned to work on a clinical case for 45 minutes. The groups used iPACS viewer software, a free DICOM viewer, to view images and annotate anatomic structures. The classroom instructor monitored and displayed each group's work sequentially on the master screen by running SynchronEyes, a software tool for controlling PC desktops remotely. RESULTS: Students were able to execute the assigned tasks using the iPACS software with minimal oversight or instruction. Course instructors displayed each group's work on the main display screen of the classroom as the students presented the rationale for their decisions. The interactive component of the course received high ratings from the students and overall course ratings were higher than in prior years when the course was given solely in lecture format. CONCLUSIONS: DICOM viewing software is an excellent tool for enabling students to learn radiologic anatomy from real-life clinical datasets. Interactive exercises performed in groups can be powerful tools for stimulating students to learn radiologic anatomy.

Anatomy↗
PubMed2007

Spatial localization accuracy of radiologists in free-response studies: Inferring perceptual FROC curves from mark-rating data.

RATIONALE AND OBJECTIVES: Free-response data consist of a set of mark-ratings pairs. Before analysis, the data are classified or "scored" into lesion and non-lesion localizations. The scoring is done by choosing an acceptance-radius and classifying marks within the acceptance-radius of lesion centers as lesion localizations, and all other marks are classified as non-lesion localizations. The scored data are plotted as a free-response receiver operating characteristic (FROC) curve, essentially a plot of appropriately normalized numbers of lesion localizations vs. non-lesion localizations. Scored FROC curves are frequently used to compare imaging systems and computer-aided detection (CAD) algorithms. However, the choice of acceptance-radius is arbitrary. This makes it difficult to compare curves from different studies and to estimate true performance. MATERIALS AND METHODS: To resolve this issue the concept of two types of marks is introduced: perceptual hits and perceptual misses. A perceptual hit is a mark made in response to the observer seeing the lesion. A perceptual miss is a mark made in response to the observer seeing a (lesion-like) non-lesion. A method of estimating the most probable numbers of perceptual hits and misses is described. This allows one to plot a perceptual FROC operating point and by extension a perceptual FROC curve. Unlike a scored FROC operating point, a perceptual point is independent of the choice of acceptance-radius. The method does not allow one to identify individual marks as perceptual hits or misses-only the most probable numbers. It is based on a three-parameter statistical model of the spatial distributions of perceptual hits and misses relative to lesion centers. RESULTS: The method has been applied to an observer dataset in which mammographers and residents with different levels of experience were asked to locate lesions in mammograms. The perceptual operating points suggest superior performance for the mammographers and equivalent performance for residents in the first and second mammography rotations. These results and the model validation are preliminary as they are based on a small dataset. CONCLUSION: The significance of this study is showing that it is possible to probabilistically determine if a mark resulted from seeing a lesion or a non-lesion. Using the method developed in this study one could perform acceptance-radius independent estimation of observer performance.

Image Interpretation, Computer-Assisted↗
PubMed2025

Implementing a training resource for large-scale genomic data analysis in the All of Us Researcher Workbench.

A lack of representation in genomic research and limited access to computational training create barriers for many researchers seeking to analyze large-scale genetic datasets. The All of Us Research Program provides an unprecedented opportunity to address these gaps by offering genomic data from a broad range of participants, but its impact depends on equipping researchers with the necessary skills to use it effectively. The All of Us Biomedical Researcher (BR) Scholars Program at Baylor College of Medicine aims to break down these barriers by providing early-career researchers with hands-on training in computational genomics through the All of Us Evenings with Genetics Research Program. The year-long program begins with the faculty summit, an in-person computational boot camp that introduces scholars to foundational skills for using the All of Us dataset via a cloud-based research environment. The genomics tutorials focus on genome-wide association studies (GWASs), utilizing Jupyter Notebooks and the Hail computing framework to provide an accessible and scalable approach to large-scale data analysis. Scholars engage in hands-on exercises covering data preparation, quality control, association testing, and result interpretation. By the end of the summit, participants will have successfully conducted a GWAS, visualized key findings, and gained confidence in computational resource management. This initiative expands access to genomic research by equipping early-career researchers from a variety of backgrounds with the tools and knowledge to analyze All of Us data. By lowering barriers to entry and promoting the study of representative populations, the program fosters innovation in precision medicine and advances equity in genomic research.

Humans↗
PubMed2005

A multistate comparison of patient characteristics, outcomes, and treatment practices in acute myocardial infarction.

The primary purpose of this study was to examine variations in patient characteristics, outcomes, and treatment practices in acute myocardial infarction (AMI) across 11 states. Data from 11 states were extracted from the Healthcare Cost and Utilization Project State Inpatient Dataset. Patients who had a primary diagnosis of AMI (International Classification of Diseases, Ninth Revision, Clinical Modification, code 410.x1) from 11 states were extracted from the Healthcare Cost and Utilization Project 1999 dataset. Bivariate comparisons were conducted to examine the characteristics, treatment practices, and outcomes of patients who had AMI. Multivariate regression models were used to examine the association between geographic location (and other factors) and the likelihood of in-hospital mortality, undergoing coronary artery bypass grafting (CABG), or percutaneous coronary interventions (PCIs). Results revealed considerable variations across states in practice patterns and treatment outcomes. New York had the highest average length of stay (8.2 days, p <0.01), rate of patients who had AMI being transferred (20.7%, p <0.01), and in-hospital case fatality rate (10.7%, p <0.01) and the lowest rate of alive discharges being routine (65.6%, p <0.01). PCI was performed 2 times as often as CABG for patients who had AMI (23.9% vs 11.3%, p <0.01), with patients who underwent CABG being transferred more often. Multivariate analyses showed that state of residence, age, female gender, transfer status, and number of co-morbidities were predictors of in-hospital mortality and the likelihood of undergoing CABG or PCI. In conclusion, large differences in practice patterns and treatment outcomes exist across states.

Age Distribution↗
PubMed2006

An updated meta-analysis of genome scans for hypertension and blood pressure in the NHLBI Family Blood Pressure Program (FBPP).

A meta-analysis of the results from a multicenter genome-wide linkage study for hypertension and blood pressure (BP) based on an initial sample of 6,245 individuals was published in 2003. We report here a combined linkage analysis of hypertension and BP using the complete Family Blood Pressure Program (FBPP) dataset, which includes a total of 12,028 genotyped individuals. Genome-wide linkage analyses for hypertension and BP were first performed in each of the studied ethnic group within each network and the results were combined with a meta-analysis using a modified Fisher's method of combining P values. Our meta-analysis of genome scans for the latest FBPP dataset reveals suggestive linkage for hypertension and BP at several regions on the human genome. Strong evidence for linkage at two of these regions, 2p14 and 3p14.1, have also been published in previous meta-analyses, making them good candidate locations for susceptibility variants.

Adult↗
PubMed2004

Environmental risk limits for antifouling substances.

In 1989, the EU restricted the use of tributyl-tin (TBT) and the International Maritime Organisation (IMO) decided for a world-wide ban on TBT in 2003. As a replacement for TBT, new antifouling agents are entering the market. Environmental risk limits (ERLs) are derived for substances that are used as TBT-substitutes, i.e. the compounds Irgarol 1051, dichlofluanid, ziram, chlorothalonil and TCMTB. ERLs represent the potential risk of the substances to the ecosystem and are derived using data on (eco)toxicology and environmental chemistry. Only toxicity studies with endpoints related to population dynamics are taken into account. For Irgarol 1051 especially plants appear to be sensitive; the mode of action is inhibition of photosynthetic electron transport. Despite the higher sensitivity of the plants, the calculated ERL for water based on plants only is higher than the ERL based on all data due to the lower variability in the plant only dataset. Because there is a mechanistic basis to state that plants are the most sensitive species, we propose to base the ERL for water on the plants only dataset. As dichlofluanid is highly unstable in the water phase, it is recommended to base the ERL on the metabolites formed and not on the parent compound. No toxicity data of the studied compounds for organisms living in sediments were found, the ERLs for sediment are derived with help of the equilibrium partitioning method. For dichlofluanid and chlorothalonil the ERL for soil is directly based on terrestrial data, for Irgarol 1051 and ziram the ERL for soil is derived using equilibrium partitioning. Except for Irgarol 1051, no information was encountered in the open literature on the environmental occurrence in The Netherlands of the chemicals studied. The measured concentrations for Irgarol 1051 are close to the derived ERL. For this compound it is concluded that the species composition and thereby ecosystem functioning cannot be considered as protected.

Aniline Compounds↗
PubMed2005

Predicting breast cancer survivability: a comparison of three data mining methods.

OBJECTIVE: The prediction of breast cancer survivability has been a challenging research problem for many researchers. Since the early dates of the related research, much advancement has been recorded in several related fields. For instance, thanks to innovative biomedical technologies, better explanatory prognostic factors are being measured and recorded; thanks to low cost computer hardware and software technologies, high volume better quality data is being collected and stored automatically; and finally thanks to better analytical methods, those voluminous data is being processed effectively and efficiently. Therefore, the main objective of this manuscript is to report on a research project where we took advantage of those available technological advancements to develop prediction models for breast cancer survivability. METHODS AND MATERIAL: We used two popular data mining algorithms (artificial neural networks and decision trees) along with a most commonly used statistical method (logistic regression) to develop the prediction models using a large dataset (more than 200,000 cases). We also used 10-fold cross-validation methods to measure the unbiased estimate of the three prediction models for performance comparison purposes. RESULTS: The results indicated that the decision tree (C5) is the best predictor with 93.6% accuracy on the holdout sample (this prediction accuracy is better than any reported in the literature), artificial neural networks came out to be the second with 91.2% accuracy and the logistic regression models came out to be the worst of the three with 89.2% accuracy. CONCLUSION: The comparative study of multiple prediction models for breast cancer survivability using a large dataset along with a 10-fold cross-validation provided us with an insight into the relative prediction ability of different data mining methods. Using sensitivity analysis on neural network models provided us with the prioritized importance of the prognostic factors used in the study.

Breast Neoplasms↗