Biomedical subjects

S Dreiseitl

Publications and source records attributed to S Dreiseitl.

10 recordsLinked to original sources

PubMed2001

A comparison of machine learning methods for the diagnosis of pigmented skin lesions.

We analyze the discriminatory power of k-nearest neighbors, logistic regression, artificial neural networks (ANNs), decision tress, and support vector machines (SVMs) on the task of classifying pigmented skin lesions as common nevi, dysplastic nevi, or melanoma. Three different classification tasks were used as benchmarks: the dichotomous problem of distinguishing common nevi from dysplastic nevi and melanoma, the dichotomous problem of distinguishing melanoma from common and dysplastic nevi, and the trichotomous problem of correctly distinguishing all three classes. Using ROC analysis to measure the discriminatory power of the methods shows that excellent results for specific classification problems in the domain of pigmented skin lesions can be achieved with machine-learning methods. On both dichotomous and trichotomous tasks, logistic regression, ANNs, and SVMs performed on about the same level, with k-nearest neighbors and decision trees performing worse.

Algorithms↗

PubMed2001

Disambiguation data: extracting information from anonymized sources.

Privacy protection is an important consideration when releasing medical databases to the research community. We show that while recent advances in anonymization algorithms provide increased levels of protection, it is still possible to calculate approximations to the original data set. In some cases, one can even uniquely reconstruct entries in a table before anonymization. In this paper, we demonstrate how knowledge of an anonymization algorithm based on ambiguating data cell entries can be used to undo the anonymization process. We investigate the effect of this algorithm and its reversal on data sets of varying sizes and distributions. It is shown that by using a computationally complex disambiguation process, information on individuals can be extracted from an anonymized data set.

Adult↗

PubMed2001

Effects of data anonymization by cell suppression on descriptive statistics and predictive modeling performance.

Protecting individual data in disclosed databases is essential. Data anonymization strategies can produce table ambiguation by suppression of selected cells. Using table ambiguation, different degrees of anonymization can be achieved, depending on the number of individuals that a particular case must become indistinguishable from. This number defines the level of anonymization. Anonymization by cell suppression does not necessarily prevent inferences from being made from the disclosed data. Preventing inferences may be important to preserve confidentiality. We show that anonymized data sets can preserve descriptive characteristics of the data, but might also be used for making inferences on particular individuals, which is a feature that may not be desirable. The degradation of predictive performance is directly proportional to the degree of anonymity. As an example, we report the effect of anonymization on the predictive performance of a model constructed to estimate the probability of disease given clinical findings.

Algorithms↗

PubMed2001

Hiding information by cell suppression.

Joining relational data can jeopardize patient confidentiality if disseminated data for research can be joined with publicly available data containing, for example, explicit identifiers. Ambiguity in data hinders the construction of primary keys that are of importance when joining data tables. We define two values to be indiscernible if they are the same or at least one of them is a special value. Two rows in a data table are indiscernible if their corresponding entries are indiscernible. We further define a table to be k-ambiguous if each row is indiscernible from at least k rows in the same table. We present two simple heuristics to make a table k-ambiguous by cell suppression, and compare them on example data.

Algorithms↗

PubMed2000

Computer-aided epiluminescence microscopy of pigmented skin lesions: the value of clinical data for the classification process.

Early melanoma is often difficult to differentiate from benign pigmented skin lesions (PSLs). Digital epiluminescence microscopy (DELM) and automated image analysis could represent possible aids for inexperienced clinicians. We designed an automated computerized image analysis system that has the potential for use as an additional tool for the differentiation of melanoma from dysplastic naevi and common naevi. The PC-based pilot system was attached to a common DELM system as the image source. Digital images of PSLs were automatically segmented and a panel of 107 morphological parameters were measured. Additionally, seven clinical parameters were evaluated and used as an additional source of information. Neural networks were then trained to distinguish melanoma from benign PSLs. One class of networks was trained solely based on the morphometric features, whereas the second class of networks was trained on the combination of morphometric and clinical features. The automatic segmentation algorithm was correct in 96% of cases. Using three-way receiver operating characteristic (ROC) analysis, for networks trained solely on morphometric features the volume under surface (VUS) was 0.617 (SD 0.036). The performance was significantly better for networks trained on the combination of both morphometric and clinical features (VUS = 0.682, SD 0.035). In a dichotomous model, distinguishing benign lesion (common naevi + dysplastic naevi) from melanoma, the area under the curve (AUC) from two-way ROC analysis was 0.942 (SD 0.018) for networks trained solely on morphometric features and 0.968 (SD 0.012) for those trained on the combination of clinical and morphometric data (P= NS). Automated feature extraction from PSLs and the training of neural networks as classifiers has thus shown satisfactory performance in a large scale experiment. The addition of clinical data significantly increases the diagnostic performance for distinguishing three classes of lesions (i.e. common naevi, dysplastic naevi and melanoma). Such integrated systems hold promise as a decision aid for the diagnosis of PSLs.

Algorithms↗

PubMed2000

The interpretation of test results.

BACKGROUND: Dermatologists need to interpret an increasing number of research studies and diagnostic tests. Understanding the techniques for interpreting test results and making decisions based upon those tests represent important tools for decision making for both clinicians and researchers. OBJECTIVE: This article focuses briefly on the key parameters of diagnostic tests: sensitivity, specificity, prevalence, predictive values, likelihood ratios, and the concept of receiver-operating-characteristic (ROC) curves. A simple example is presented in a step-by-step manner. CONCLUSION: The principles of interpreting test results are easy to learn and applicable in daily clinical routine. Therefore, dermatologists should be familiar with the concepts outlined in this paper.

Enzyme-Linked Immunosorbent Assay↗

PubMed2000

Risk stratification in heart failure using artificial neural networks.

Accurate risk stratification of heart failure patients is critical to improve management and outcomes. Heart failure is a complex multisystem disease in which several predictors are categorical. Neural network models have successfully been applied to several medical classification problems. Using a simple neural network, we assessed one-year prognosis in 132 patients, consecutively admitted with heart failure, by classifying them in 3 groups: death, readmission and one-year event-free survival. Given the small number of cases, the neural network model was trained using a resampling method. We identified relevant predictors using the Automatic Relevance Determination (ARD) method, and estimated their mean effect on the 3 different outcomes. Only 9 individuals were misclassified. Neural networks have the potential to be a useful tool for making prognosis in the domain of heart failure.

Disease-Free Survival↗

PubMed2000

Building knowledge in a complex preterm birth problem domain.

Data mining methods used a racially diverse sample (n = 19,970) of pregnant women and 1,622 variables that were collected in Duke's TMR electronic patient record over a 10-year period. Different statistical and data mining methods were similar when compared using receiver operating characteristic (ROC) curves. Best results found that seven demographic variables yielded .72 and addition of hundreds of other clinical variables added only .03 to the area under the curve (AUC). Similar results across methods suggest that results were data-driven and not method-dependent, and that demographic variables may offer a small set of parsimonious variables with predictive accuracy in a racially diverse population. Work to determine relevant variables for improved predictive accuracy is ongoing.

Area Under Curve↗

PubMed1999

Evaluating variable selection methods for diagnosis of myocardial infarction.

This paper evaluates the variable selection performed by several machine-learning techniques on a myocardial infarction data set. The focus of this work is to determine which of 43 input variables are considered relevant for prediction of myocardial infarction. The algorithms investigated were logistic regression (with stepwise, forward, and backward selection), backpropagation for multilayer perceptrons (input relevance determination), Bayesian neural networks (automatic relevance determination), and rough sets. An independent method (self-organizing maps) was then used to evaluate and visualize the different subsets of predictor variables. Results show good agreement on some predictors, but also variability among different methods; only one variable was selected by all models.

Algorithms↗

PubMedDate not supplied

Comparing three-class diagnostic tests by three-way ROC analysis.

Three-way ROC surfaces are based on a generalization of dichotomous ROC analysis to three-class diagnostic tests. The discriminatory power of three-class diagnostic tests is measured by the volume under the ROC surface. This measure can be given a probabilistic interpretation similar to the equivalence of the c-index to the area under the ROC curve. This article presents a method to calculate nonparametric estimates of the variance of the volume under the surface using Mann-Whitney U statistics. As a simple extension of this result, it is possible to calculate covariance estimates for the volume under the surface. This allows the statistical comparison of two tests used for diagnostic tasks with three possible outcomes. The formulas derived are validated on synthetic data and applied to a three-class data set of pigmented skin lesions. It is shown that a neural network algorithm trained on clinical data and lesion features performs better than one trained on only the lesion features.

Humans↗