PubMed Health⌕ Search

Biomedical subjects

R Caruana

Publications and source records attributed to R Caruana.

13 recordsLinked to original sources
PubMed2000

Predicting cesarean delivery with decision tree models.

OBJECTIVE: The purpose of this study was to determine whether decision tree-based methods can be used to predict cesarean delivery. STUDY DESIGN: This was a historical cohort study of women delivered of live-born singleton neonates in 1995 through 1997 (22,157). The frequency of cesarean delivery was 17%; 78 variables were used for analysis. Decision tree rule-based methods and logistic regression models were each applied to the same 50% of the sample to develop the predictive training models and these models were tested on the remaining 50%. RESULTS: Decision tree receiver operating characteristic curve areas were as follows: nulliparous, 0.82; parous, 0.93. Logistic receiver operating characteristic curve areas were as follows: nulliparous, 0.86; parous, 0.93. Decision tree methods and logistic regression methods used similar predictive variables; however, logistic methods required more variables and yielded less intelligible models. Among the 6 decision tree building methods tested, the strict minimum message length criterion yielded decision trees that were small yet accurate. Risk factor variables were identified in 676 nulliparous cesarean deliveries (69%) and 419 parous cesarean deliveries (47.6%). CONCLUSION: Decision tree models can be used to predict cesarean delivery. Models built with strict minimum message length decision trees have the following attributes: Their performance is comparable to that of logistic regression; they are small enough to be intelligible to physicians; they reveal causal dependencies among variables not detected by logistic regression; they can handle missing values more easily than can logistic methods; they predict cesarean deliveries that lack a categorized risk factor variable.

Adolescent↗
PubMed1999

Case-based explanation of non-case-based learning methods.

We show how to generate case-based explanations for non-case-based learning methods such as artificial neural nets or decision trees. The method uses the trained model (e.g., the neural net or the decision tree) as a distance metric to determine which cases in the training set are most similar to the case that needs to be explained. This approach is well suited to medical domains, where it is important to understand predictions made by complex machine learning models, and where training and clinical practice makes users adept at case interpretation.

Artificial Intelligence↗
PubMed1998

Improving the care of patients treated with hemodialysis: a report from the Health Care Financing Administration's ESRD Core Indicators Project.

To determine the impact of a quality improvement intervention on dialysis care delivered to hemodialysis patients, we studied 213 hemodialysis facilities in North Carolina, South Carolina, and Georgia. Dialysis adequacy measurements made on two random samples of 30 patients per treatment center, or all patients if fewer than 30 were treated, selected in October 1994 (preintervention) and October 1995 (postintervention) were used to estimate the facility mean urea reduction ratio (URR) and the proportion of patients with a mean URR less than 50%. The 10% of facilities (n = 22) with the highest proportion of patients with a mean URR less than 50% in the facility at preintervention were selected for an intervention that included feedback of facility-specific mean URR, educational programs, a quality improvement workshop, and monitoring until improvement was attained. Changes between preintervention and postintervention facility mean URR and proportions of patients with a URR less than 60% and 65% were used to assess the impact of the intervention. After 1 year, the mean URR had increased an average of 7% in intervention centers compared with an increase of 1.4% (P < 0.001) in the remainder of the treatment centers in the Network. There was an average reduction of 17.2% in the proportion of patients with a URR less than 65% in intervention centers compared with 4.8% in the other facilities (P < 0.001). Comparable reductions in the proportion of patients with a mean URR of less than 60% were 16.2% in intervention centers and 2.0% in comparison facilities (P < 0.001). After controlling for facility case mix and other characteristics, the intervention was independently associated with an absolute 2.4% increase in facility-specific mean URR. We conclude that the intervention was associated with improvement in hemodialysis care.

Ambulatory Care Facilities↗
PubMed1997

An evaluation of machine-learning methods for predicting pneumonia mortality.

This paper describes the application of eight statistical and machine-learning methods to derive computer models for predicting mortality of hospital patients with pneumonia from their findings at initial presentation. The eight models were each constructed based on 9847 patient cases and they were each evaluated on 4352 additional cases. The primary evaluation metric was the error in predicted survival as a function of the fraction of patients predicted to survive. This metric is useful in assessing a model's potential to assist a clinician in deciding whether to treat a given patient in the hospital or at home. We examined the error rates of the models when predicting that a given fraction of patients will survive. We examined survival fractions between 0.1 and 0.6. Over this range, each model's predictive error rate was within 1% of the error rate of every other model. When predicting that approximately 30% of the patients will survive, all the models have an error rate of less than 1.5%. The models are distinguished more by the number of variables and parameters that they contain than by their error rates; these differences suggest which models may be the most amenable to future implementation as paper-based guidelines.

Artificial Intelligence↗
PubMed1991

Usefulness of peritoneal fluid amylase levels in the differential diagnosis of peritonitis in peritoneal dialysis patients.

Peritonitis continues to be a major cause of morbidity in peritoneal dialysis patients despite recent technological advances (Y systems) that have reduced peritonitis rates to much more acceptable levels. Most of the time when a peritoneal dialysis patient presents with peritonitis, it is infectious in origin. However, these patients occasionally develop other intra-abdominal pathology that requires more intensive medical care or, at times, surgical intervention. To help in the early differential diagnosis of the cause of peritonitis in these patients, peritoneal fluid amylase levels were prospectively obtained from 50 patients presenting to the hospital with peritonitis. Thirty-nine of them had typical infectious peritonitis, and their mean peritoneal fluid amylase level was 11.1 (range, 0 to 90). Six patients had pancreatitis and a mean peritoneal fluid amylase level of 550 U/L (range, 100 to 1,140 U/L). Five patients were found to have other intra-abdominal pathology, and their mean peritoneal fluid amylase level was 816 U/L (range, 142 to 1,746 U/L). In patients who did not respond to initial therapy, sequential peritoneal fluid amylase levels did not increase in patients with typical infectious peritonitis whereas it did increase in patients with other intra-abdominal pathology. In conclusion, it was found that peritoneal fluid amylase levels were helpful in the differential diagnosis of peritonitis in these patients. An elevated level (greater than 100 U/L) differentiated those patients with other intra-abdominal causes of peritonitis from those with typical infectious peritonitis.

Amylases↗
PubMed1987

Ethylene oxide sensitivity in hemodialysis patients.

Sera from 138 patients who had experienced hypersensitivity-type reactions while on hemodialysis (reactors) were examined retrospectively by the radioallergosorbent test (RAST) for specific IgE antibody to ethylene oxide (ETO). Seventy-eight hemodialysis patients without a history of reaction were also evaluated as controls. Elevated serum RAST values (greater than 2.0) were more common in reactors (63%) than in controls (11%, p less than 0.001). In a second study, RAST assays were performed using human serum albumin conjugated to ETO (HSA-ETO) as antigen and also using a concentrate of fluid used to rinse ETO-sterilized dialyzers ("dialyzer extract") as antigen. The RAST ratios obtained with HSA-ETO were similar to those obtained using the dialyzer extract (rank order correlation coefficient = 0.829, p less than 0.001). In a third study, RAST inhibition was demonstrated both by HSA-ETO and dialyzer extract. Our results, extending previously published reports, suggest that hypersensitivity to ETO might play an important role in hemodialysis-associated hypersensitivity-type reactions.

Antigens↗