Biomedical subjects

L Ohno-Machado

Publications and source records attributed to L Ohno-Machado.

At least 19 recordsLinked to original sources

PubMed2001

Vascular closure devices and the risk of vascular complications after percutaneous coronary intervention in patients receiving glycoprotein IIb-IIIa inhibitors.

Vascular closure devices offer advantages over traditional means of obtaining hemostasis after percutaneous coronary intervention (PCI) in terms of patient comfort and time to ambulation. We investigate whether such devices also reduce the risk of vascular complications in selected patient populations. We conducted a retrospective analysis of all patients who underwent PCI at our institution between January 1998 and December 1999. Of 3,151 consecutive patients, 3,027 were eligible to receive vascular closure devices. Of these, 1,485 received a closure device and 1,409 received glycoprotein IIb-IIIa antagonists. The overall vascular complication rate, as defined by the need for surgical repair or transfusion, or the development of arteriovenous fistula, pseudoaneurysm, or large hematoma, was 4.20%. By univariate analysis, the use of closure devices was associated with a lower vascular complication rate (3.03% vs 5.52%; p = 0.002) and a shorter length of hospital stay (2.77 vs 3.97 days, p <0.001). Multivariate analysis showed a significant reduction in vascular complications with closure devices (odds ratio 0.59, p = 0.007). For the subgroup of patients receiving glycoprotein IIb-IIIa antagonists, the use of closure devices was associated with an even more pronounced reduction in the risk of vascular complications (odds ratio 0.45, p <0.008). Thus, the use of closure devices in selected patients undergoing PCI is associated with a low rate of vascular complications and decreased length of stay. This benefit was most marked for patients receiving glycoprotein IIb-IIIa antagonists.

Age Distribution↗

PubMed2001

Simplified risk score models accurately predict the risk of major in-hospital complications following percutaneous coronary intervention.

The objectives of this analysis were to develop and validate simplified risk score models for predicting the risk of major in-hospital complications after percutaneous coronary intervention (PCI) in the era of widespread stenting and use of glycoprotein IIb/IIIa antagonists. We then sought to compare the performance of these simplified models with those of full logistic regression and neural network models. From January 1, 1997 to December 31, 1999, data were collected on 4,264 consecutive interventional procedures at a single center. Risk score models were derived from multiple logistic regression models using the first 2,804 cases and then validated on the final 1,460 cases. The area under the receiver operating characteristic (ROC) curve for the risk score model that predicted death was 0.86 compared with 0.85 for the multiple logistic model and 0.83 for the neural network model (validation set). For the combined end points of death, myocardial infarction, or bypass surgery, the corresponding areas under the ROC curves were 0.74, 0.78, and 0.81, respectively. Previously identified risk factors were confirmed in this analysis. The use of stents was associated with a decreased risk of in-hospital complications. Thus, risk score models can accurately predict the risk of major in-hospital complications after PCI. Their discriminatory power is comparable to those of logistic models and neural network models. Accurate bedside risk stratification may be achieved with these simple models.

Angioplasty, Balloon, Coronary↗

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

Using patient-reportable clinical history factors to predict myocardial infarction.

Using a derivation data set of 1253 patients, we built several logistic regression and neural network models to estimate the likelihood of myocardial infarction based upon patient-reportable clinical history factors only. The best performing logistic regression model and neural network model had C-indices of 0.8444 and 0.8503, respectively, when validated on an independent data set of 500 patients. We conclude that both logistic regression and neural network models can be built that successfully predict the probability of myocardial infarction based on patient-reportable history factors alone. These models could have important utility in applications outside of a hospital setting when objective diagnostic test information is not yet be available.

Databases, Factual↗

PubMed2001

Unsupervised learning from complex data: the matrix incision tree algorithm.

Analysis of large-scale gene expression data requires novel methods for knowledge discovery and predictive model building as well as clustering. Organizing data into meaningful structures is one of the most fundamental modes of learning. DNA microarray data set can be viewed as a set of mutually associated genes in a high-dimensional space. This paper describes a novel method to organize a complex high-dimensional space into successive lower-dimensional spaces based on the geometric properties of the data structure in the absence of a priori knowledge. The matrix incision tree algorithm reveals the hierarchical structural organization of observed data by determining the successive hyperplanes that 'optimally' separate the data hyperspace. The algorithm was tested against published data sets yielding promising results.

Algorithms↗

PubMed2001

Effects of case removal in prognostic models.

Constructing and updating prognostic models that learn from training cases is a time-consuming task. The more compact, and yet informative, the training sets are, the faster one can build and properly evaluate such models. We have compared different regression diagnostic methods for selection and removal of training cases in prognostic models. Univariate determinations were performed using classical regression diagnostic statistics. Multivariate determinations were performed using (1) a sequential "backward" selection of cases, and (2) a non-sequential genetic algorithm. The genetic algorithm produced final models that kept few cases and retained predictive capability. A genetic algorithm approach to case selection may be better suited for guiding removal of cases in training sets than a univariate or a sequential multivariate approach, possibly because of its ability to detect sets of cases that are influential en bloc but may not be sufficiently influential when considered in isolation.

Algorithms↗

PubMed2001

Training in medical informatics: combining onsite and online instruction.

The Internet is promoting active exchange of teaching materials and discussion among geographically distant collaborators. We envision that training in medical informatics can be better achieved if both onsite and online instruction are combined, provided that cultural and technological barriers are anticipated and the training program is prepared accordingly. We describe our Brazil/USA program in medical informatics, which includes components of on-site and online education, and discuss lessons learned during its ongoing implementation. Three onsite courses and one workshop have been planned, and two online courses are being developed.

Brazil↗

PubMed2001

Finding appropriate clinical trials: evaluating encoded eligibility criteria with incomplete data.

We describe our work on creating a system that selects appropriate clinical trials by automating the evaluation of eligibility criteria. We developed a data model of eligibility for breast cancer clinical trials, upon which the criteria were encoded. Standard vocabularies are utilized to represent concepts used in the system, and retrieve their hierarchical relationships. The system incorporates Bayesian networks to handle missing patient information. Protocols are ranked by the belief that the patient is eligible for each of them. In a preliminary evaluation, we found good agreement (kappa 0.86) between the system and an independent physician in selection of protocols, but poor agreement (kappa 0.24) in protocol ranking. We conclude that our approach is feasible, and potentially useful in assisting both physicians and patients in the task of selecting appropriate trials.

Bayes Theorem↗

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↗

PubMed2001

Generation of dynamically configured check lists for intra-operative problems using a set of covering algorithms.

We present a prototype of a decision support system for anesthesia that applies set covering theory. The system is designed to generate dynamically configured check-lists for intra-operative problems. These lists have the potential to help anesthesiologists detect and manage problems in a timely manner. The items in the lists consist of major complications that should be considered for a particular case. A set covering algorithm that accommodates multiple problem sets was used to implement the prototype. A simulated case and the system behavior are presented. The ultimate goals of a system such as the one presented are to function as an intelligent alarm module for electronic monitors and to facilitate the task of correcting intra-operative problems.

Algorithms↗

PubMed2000

A genetic algorithm approach to multi-disorder diagnosis.

One of the common limitations of expert systems for medical diagnosis is that they make an implicit assumption that multiple disorders do not co-occur in a single patient. The need for this simplifying assumption stems from the fact that finding minimal sets of disorders that cover all symptoms for a given patient is generally computationally intractable (NP-hard). In this paper, we explain the need for performing multi-disorder diagnosis, review previous approaches, formulate the problem using set theory notation, and propose the use of a search method based on a genetic algorithm. We test the algorithm and compare it to another approach using a simple example. The genetic algorithm performs well independently of the order of symptoms, and has the potential to perform multi-disorder diagnosis using existing or newly developed knowledge bases.

Algorithms↗

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↗

PubMed2000

Major complications after angioplasty in patients with chronic renal failure: a comparison of predictive models.

Novel modeling approaches were investigated to predict major complications in patients with chronic renal failure (CRF) or end-stage renal disease (ESRD) undergoing percutaneous transluminal coronary angioplasty (PTCA). The following hypotheses were explored: (1) Pre-angioplasty patient risk factors, demographic characteristics and procedural information may be used to predict major complications after PTCA; and (2) Rough sets and artificial neural nets (ANN) may be used to build models that are better than standard logistic regression models. Several variables were found to be predictive of major complications for patients with CRF or ESRD undergoing PTCA. The presence of shock at presentation portends poor outcome but congestive heart failure and prior history of myocardial infarction increases the risk tenfold and 25-fold, respectively. The discriminatory ability of the ANN model was better than both Rough Sets and Logistic Regression for the test set.

Analysis of Variance↗

PubMed2000

Decision trees and fuzzy logic: a comparison of models for the selection of measles vaccination strategies in Brazil.

In 1997, health authorities of the state of São Paulo, Brazil designed a vaccination campaign against measles based on a decision model that utilized fuzzy logic. The chosen mass vaccination strategy was implemented and changed the natural course of the epidemic in that state. We have built a model using a decision tree and compare it to the fuzzy logic model. Using essentially the same set of assumptions about this problem, we contrast the two approaches. The models identify the same strategy as being the best one, but exhibit differences in the ranking of the remaining strategies.

Brazil↗

PubMed2000

GLIF3: the evolution of a guideline representation format.

The Guideline Interchange Format (GLIF) is a language for structured representation of guidelines. It was developed to facilitate sharing clinical guidelines. GLIF version 2 enabled modeling a guideline as a flowchart of structured steps, representing clinical actions and decisions. However, the attributes of structured constructs were defined as text strings that could not be parsed, and such guidelines could not be used for computer-based execution that requires automatic inference. GLIF3 is a new version of GLIF designed to support computer-based execution. GLIF3 builds upon the framework set by GLIF2 but augments it by introducing several new constructs and extending GLIF2 constructs to allow a more formal definition of decision criteria, action specifications and patient data. GLIF3 enables guideline encoding at three levels: a conceptual flowchart, a computable specification that can be verified for logical consistency and completeness, and an implementable specification that can be incorporated into particular institutional information systems.

Decision Support Techniques↗