SEARCH · PubMed Health

Results for “Bayesian analysis”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 109 records · Page 6Linked to original sources

PubMed1989

Bayesian derived predictions for twice daily theophylline under outpatient conditions and an assessment of optimal sampling times.

1. The accuracy of a computerised method of pharmacokinetic interpretation of a single serum theophylline concentration, employing the statistical technique of Bayesian analysis, has been evaluated for an oral slow release form of theophylline using twice daily dosing. 2. Twenty-four hour steady state serum theophylline concentration-time profiles of one Uniphyllin Continus 400 mg tablet (Napp Laboratories) every 12 h were measured in 15 patients. These profiles demonstrated a diurnal variation of theophylline absorption which was faster during the day. 3. Revised predictions of the profiles were generated by Bayesian analysis using a single serum theophylline concentration taken during a previous outpatient appointment. Comparing the predicted and measured profiles, the accuracy of the Bayesian method is considered more than adequate for clinical purposes. 4. The predictions produced by the revised estimates were statistically less biased and more precise than those derived by a theophylline algorithm using population data. 5. The mean prediction errors of the revised estimates of the day and night-peak drug concentrations were -0.55 mg l-1 and -0.21 mg l-1 whilst those of the evening and morning troughs were 1.17 mg l-1 and 0.41 mg l-1, respectively. 6. Analysis of the predictive and relative performance of the samples drawn during the profile revealed that the sample taken prior to a morning dose produced the most accurate predictions. 7. There was no statistical difference in the relative predictive performance of samples drawn up to 4 h before or 2 h after the morning dose. It is, therefore, recommended that all serum theophylline concentrations to be used in Bayesian analysis, should be drawn within this period.

Administration, Oral↗

PubMed1984

Comparison of the multivariate analysis and CADENZA systems for determination of the probability of coronary artery disease.

The accuracy of 2 discriminate systems for diagnosis of coronary artery disease (CAD), multivariate analysis (MVA) and Bayesian analysis (CADENZA), was evaluated in 113 patients undergoing electrocardiographic stress testing and coronary angiography. MVA uses weighting factors (F values) generated from our patient data, whereas CADENZA uses probabilities gleaned from an extensive review of the American literature. Overall accuracy was similar. MVA had a higher sensitivity for 1-vessel CAD (75 versus 33%), but CADENZA was better for determining the severity of CAD. The 2 systems provided posterior probabilities for disease that were highly correlated (r = 0.56; p less than 0.001). Both systems suggest the need for further testing based on the probability generated; herein lies their major strength. The application of such systems should help the clinician reach a diagnosis or make a decision as to management in a cost-effective manner.

Bayes Theorem↗

PubMed2026

Genetic Susceptibility to Incisional Hernia Evaluation of Hernia Polygenic Risk Scores.

OBJECTIVES: Incisional hernia (IH) affects 13-30% of people after abdominal surgery, resulting in substantial morbidity and costs. While clinical risk factors have been studied extensively, genomic risk for IH is incompletely understood. We aimed to evaluate the impact of polygenic risk scores (PRS) on IH risk prediction. METHODS: We created and evaluated three PRS for abdominal hernia, ventral hernia and latent hernia susceptibility for prediction of IH in an institutional biobank. The primary outcome was defined as the diagnosis or repair of an IH based on ICD-9/10-CM/PCS and CPT codes. Clinical covariates included age, sex, body mass index (BMI), smoking status, index procedure type, and perioperative surgical site infection. A phenome-wide association study (PheWAS) was performed to assess clinical associations with increased PRS. We then tested the ability of the PRS to improve prediction for IH by modeling clinical covariates with and without PRS in patients who underwent abdominal surgery. Model performance was assessed using 10 iterations of 5-fold cross-validation to estimate Brier scores and area under the receiver operating characteristic curve (AUROC), which were compared using cross-model Bayesian analysis of variance. RESULTS: In 55,809 subjects, assessed PRS was significantly associated with incisional, umbilical, and ventral hernia on PheWAS, with 1.19 greater odds of developing IH per 1-SD increase in PRS (95% CI: 1.13-1.25, P < 0.001). Of 9,909 subjects who underwent qualifying abdominal surgery, 706 developed IH. In this cohort, the latent hernia susceptibility PRS was associated with a 16% increased hazard of developing IH per 1-SD increase (HR 1.16; 95% CI: 1.07-1.26; P < 0.001). Compared to a predictive model using clinical covariates (Brier score = 0.047, 95% CI: 0.046-0.048; AUROC = 0.660, 95% CI: 0.653-0.666), addition of the PRS showed similar Brier score and AUROC estimates (Brier score = 0.047, 95% CI: 0.046-0.048; AUROC: 0.667, 95% CI: 0.661-0.673) at five years. Cross-model Bayesian analysis demonstrated >99% probability of practical equivalence when trying to detect a difference of ≥ 0.02. CONCLUSION: All three PRS for hernia were independently associated with IH, suggesting that genomic factors contribute significantly to IH development. However, none of the three PRS meaningfully improved clinical IH risk prediction in patients who underwent abdominal surgery. This suggests that clinical comorbidities and surgical techniques may be equally as important as genomic architecture.

Bayesian analysis↗

PubMed1991

[An effective method for the estimation and comparison of the ED50 with small sample sizes].

In ED50 experiments the relationship between dose and probability of response is often modelled by the probit function. Standard statistical analysis estimates the parameters of this function by the maximum likelihood principle and derives the ED50 and its fiducial limits from these parameters. Bayesian analysis is more effective in two respects: It optionally includes prior information and in all but very few instances yields confidence intervals, whereas fiducial intervals often cannot be determined. Bayesian analysis of experiments with one substance has been treated in GRIEVE (1988). In the present article the mathematically interested reader is shown how to compare two substances. The probability of higher ED50 in the one substance as well as estimates of the ratio of the ED50's are obtained. The methods are easily extended to the effective dose for any other reasonable percentage of animals, e.g. ED90 or ED25. Experiments concerning lethal doses can be analysed by these methods as well. Both types of analysis are applied in two examples which compare new batches of vaccines with an established standard. In the first example both substances are nearly equivalent, while in the second example the new batch is considerably more efficient. An interactive FORTRAN program for a personal computer is available (cf. last section of 5.). It computes the maximum likelihood and the Bayesian solution, using approximate formulas in the latter case. Due to these approximations it was possible to develop a Bayesian program which is fast enough to run on a PC. Validation procedures have been performed. The output consists of a print file and, optionally, an ASCII file containing the coordinates of the posterior probability density and distribution functions.

Animals↗

PubMed1994

Single anticardiolipin measurement in the routine management of patients with systemic lupus erythematosus.

OBJECTIVE: To examine the usefulness of a single measurement of anticardiolipin antibodies (aCL) in systemic lupus erythematosus (SLE) in routine clinical practice. METHODS: All 127 patients with SLE currently followed by our rheumatology unit had an aCL measurement on routine clinic review. Their charts were then reviewed for specific disease manifestations. Basic statistical correlations of the aCL result and the specific disease manifestations were performed, and the clinical utility of the aCL test assessed using Bayesian analysis. RESULTS: aCL was positive (> 2 SD) in 24% and was associated with recurrent fetal loss, thrombosis, cerebrovascular disease, livedo reticularis and digital infarcts. Bayesian analysis showed that a single positive aCL test increased the relative and absolute risk of all the above complications. The criterion of aCL positivity as > 15 units (2 SD) was associated with the highest relative risk. CONCLUSION: A single positive aCL test in routine management of SLE is a useful predictor of important clinical events.

Adult↗

PubMed1995

Clinical reasoning about new symptoms despite preexisting disease: sources of error and order effects.

BACKGROUND: Previous work that studied the evaluation of new, atypical symptoms in patients with preexisting diseases indicated that physicians largely ignored the past medical history and therefore erred in their diagnoses, when compared to a Bayesian analysis. Other studies have shown that the order in which information is presented to a decision maker can affect the inferences drawn, again contrary to a Bayesian standard. OBJECTIVES: The aim of the study was to investigate the source of disparity between clinical judgment and Bayesian analysis and to investigate the effect of alternative orders of presenting information on diagnostic conclusions. METHODS: Two groups of family physicians received a written clinical scenario. One group was given the past medical history before the history of present illness, the physical exam, and the laboratory data. The second group learned about the past medical history after all other clinical information had been presented. Judgments of test accuracy and probably diagnosis were collected at several points to identify the source of any diagnostic error. RESULTS: For both groups, the major source of error was in estimating the prior probability of disease, not in estimating the accuracy of a diagnostic test or updating opinions following receipt of test results. Although both groups of physicians received the same information, they came to markedly different conclusions about the most likely diagnosis. The group given the past medical history at the beginning of the scenario considered this information much less significant than did the group who received it at the end. CONCLUSIONS: Family physicians deviate from a Bayesian standard of reasoning by wrongly specifying prior probabilities and by being influenced by the order in which clinical information is presented.

Bayes Theorem↗

PubMedDate not supplied

Designing for nonparametric Bayesian survival analysis using historical controls.

This paper gives a method for choosing the number of patients N0 out of N available patients to be randomized to current controls om a two-arm study when comparison of nonparametric survival curves is the anticipated method of data analysis. The criterion imposed is that of choosing N0 to minimize the posterior variance of the difference between the current control and experimental survival curves. A nonparametric Bayesian argument incorporating the survival curve of available historical controls establishes the criterion. Formulas and tables which facilitate this computation are presented.

Bayes Theorem↗

PubMed1989

Bayesian regression analysis of non-steady-state phenytoin concentrations: evaluation of predictive performance.

Michaelis-Menten saturable pharmacokinetics confound the determination of appropriate phenytoin maintenance doses. This study retrospectively evaluated the performance of an IBM-PC/XT computer program applying Bayesian regression to the "explicit solution to the Michaelis-Menten equation." Zero to five non-steady-state phenytoin serum concentrations were used to predict either non-steady-state concentrations at least 10 days in the future (n = 49) or steady-state concentrations (n = 20). Non-steady-state concentration prediction precision (% mean absolute error) using 0-5 non-steady-state feedbacks was 137%, 62%, 39%, 31%, 25%, and 15%, respectively, and steady-state concentration prediction precision was 446%, 47%, 50%, 44%, 21%, and 13%, respectively. Elimination of subjects receiving concurrent drugs known to induce phenytoin metabolism significantly improved predictions based on population priors; however, performance improvements were not apparent after two serum level feedbacks. The program provided clinically acceptable predictions with four or more feedbacks. Refinement of population parameters and optimal sampling times should further improve performance.

Aged↗

PubMed1995

Quantitative assessment of myocardial ultrasound tissue characterization through receiver operating characteristic analysis of Bayesian classifiers.

OBJECTIVES: This work proposes a self-consistent assessment methodology for quantitative evaluation of any combination of diagnostic features, with the immediate goal of quantitatively assessing the discriminating power in diabetic patients of features derived from ultrasound backscatter from myocardium. BACKGROUND: Four features from analysis of left ventricular myocardial ultrasound backscatter have previously been shown to be sensitive to potentially cardiomyopathic changes in patients with insulin-dependent diabetes mellitus who have no overt heart disease. The measured features were significantly different between such patients and normal control subjects, as well as among groups of such patients with and without systemic complications of the disease. The quantitative discriminating potential of the features was not assessed. METHODS: Multivariate classifier functions were constructed and analyzed by using the methodology of the receiver operating characteristic curve, which allows quantitative assessment of the discriminating power of these features, alone or in combination. The area under the receiver operating characteristic curve--the true positive rate averaged over all false positive rates--was used as a summary measure of performance. RESULTS: In distinguishing patients with insulin-dependent diabetes mellitus from normal control subjects, the most discriminating combination of ultrasound features for the detection of such changes in these patients yielded receiver operating characteristic curves with area measures of approximately 0.80; for such patients with retinopathy the measure increased to 0.90. This performance is comparable to that of many commonly used diagnostic tests. CONCLUSIONS: A self-consistent set of evaluation methodologies has quantitatively demonstrated the sensitivity of four ultrasound backscatter features to otherwise latent changes in myocardial structure that accompany the evolution of insulin-dependent diabetes mellitus. The results are remarkable in themselves and suggest the potential of the features for the general field of cardiac ultrasound tissue characterization.

Bayes Theorem↗

PubMed1988

Study and performance evaluation of statistical methods in image processing.

Two statistical image processing formalisms involving the entropy concept and Bayesian analysis are studied. Iterative imaging algorithms of the formalisms are formulated by employing, for the purpose of performance evaluation and easy implementation, the steepest descent method for the solution of entropy concept and the expectation maximization technique for the solution of Bayesian analysis. Quantitative evaluation and comparison of the convergence performance of the iterative algorithms on computer generated ideal and experimental radioisotope phantom imaging noisy data are given. The study concludes that the entropy algorithm can converge relatively fast, but it is very sensitive to noise in measured data due to the ill-posed nature of inverse problems and its lack of ability to consider the statistics of data fluctuation; while the Bayesian algorithm converges monotonically even with noisy data and has the advantage of considering both the a priori source distribution information and the statistical fluctuation of measured data.

Algorithms↗

PubMed1987

The accuracy of a pharmacokinetic theophylline predictor using once daily dosing.

1. The accuracy of a computer based pharmacokinetic prediction method based on Bayesian analysis has been evaluated for an oral show release form of theophylline. 2. In 83 patients from seven centres 24 h serum theophylline concentration-time profiles were measured under a variety of circumstances. 3. Revised predictions of 24 h serum theophylline concentration profiles were generated by Bayesian analysis using single serum drug concentrations taken before, during and after the study days in different subgroups of those patients. Comparing the predicted and measured profiles the mean prediction error (bias) was 0.05 mg l-1 for peak concentrations and 0.04 mg l-1 for trough concentrations during once daily dosing. The corresponding root mean squared prediction errors (precision) were 2.59 and 1.17 mg l-1, respectively. 4. This accuracy is considered more than adequate for clinical purposes. 5. The technique can be used with a variety of other drugs and can form a valuable part of a routine therapeutic drug monitoring service.

Adult↗

PubMedDate not supplied

[Evaluation of the activity of creatine phosphokinase for the detection of carriers of Duchenne-type muscular dystrophy in families in the city of Monterrey, Mexico].

The activity of serum creatine phosphokinase (CPK) was determined in 80 female members of 23 families with affected members of Duchenne type muscular dystrophy (DMD) and compared with the values of a control group of 100 unaffected women. The control group values exhibited a normal distribution of frequency with a mean of 21 U/L and standard deviation from the mean of 7.9 U/L. Sixty nine percent (11/16) of obligatory carriers showed CPK values higher than the mean of the control group plus two standard deviations of the mean. Thirty one percent (5/16) had false negative values. These percentages are similar to those reported in other studies. Elevated CPK activity was found in 45% (18/40) of type A possible carriers (relatives of obligatory carriers) and the group of possible carriers type B (mothers and relatives of isolated cases) 42% (10/24) exhibited high CPK values. Bayesian analysis was also used in all possible-carriers (A and B). We also report an estimation of the fertility of the DMD gene carriers and of their attitude towards family planning. It is concluded that the determination of serum CPK activity, despite its shortcomings, associated with Bayesian analysis when necessary, could be the method of choice for quick and inexpensive evaluation of the carrier status, mainly in families with members affected by DMD.

Creatine Kinase↗

PubMed1995

Comparison of methods for the estimation of carboplatin pharmacokinetics in paediatric cancer patients.

The antitumour and toxic effects of platinum drugs, in particular carboplatin, have been related to their plasma concentration and this has led to the concept of a target area under the plasma concentration-time curve (AUC) for carboplatin dosing. A formula based on renal function has been successfully applied to carboplatin dosing in adults and modified versions have also been proposed for paediatric patients. In order to monitor carboplatin AUC with maximum efficiency and minimum patient inconvenience, limited sampling strategies are desirable. A population method with Bayesian estimation is described, based on one or two samples taken following a dose of carboplatin. Population data were obtained from 22 paediatric patients treated with 200-1000 mg/m2 carboplatin as a 60-90 min infusion. Ultrafilterable carboplatin was determined by atomic absorption spectrophotometry. A two compartment model was fitted to each data set using the Maximum Likelihood estimator of the ADAPT programme. These parameter estimates provided the prior means and covariance matrix for the Bayesian estimator using a lognormal distribution. The test data sets consisted of ultrafilterable carboplatin concentrations in 23 patients (aged 1 month-18 years) who received similar treatment. The two compartment model was fitted to data sets containing one or two points, using the Bayesian maximum a posteriori (MAP) estimator and an error model derived from the population error model parameters. Results from the Bayesian analysis and other methods for the estimation of AUC, including relating clearance to surface area or to renal function, were evaluated by comparing the AUC estimate with the AUC determined by model-independent analysis. Overall, the optimal sampling strategy performed better than estimates based on renal function, which had a median bias of 5% and precision of 22%. With one data point at 60 min postinfusion, the median bias and precision were 3 and 6%, respectively. Addition of a second data point at 30 min during the infusion improved the estimate slightly (median bias -2%, precision 3%). Bayesian estimation produced more reliable estimates of AUC compared to values based on renal function, which in turn was slightly better than using surface area. A technique, developed in adult patients, for estimating AUC from a measurement of 24 h total plasma platinum was comparable to estimates based on renal function, but was less reliable. The estimation of carboplatin AUC can be performed using only one or two plasma samples and Bayesian analysis. This approach is less biased and more precise than methods based on surface area, renal function or total platinum at 24 h postdose, but is probably best used in combination with dosing based on renal function.

Adolescent↗

PubMed1992

Bayesian subset analysis in a colorectal cancer clinical trial.

Subset analysis is the examination of treatment comparisons within groups of patients with restricted levels of patient characteristics. Such analyses are vulnerable to multiplicity effects. We examine the problem in the context of a proportional hazards model with terms for treatment, each of several dichotomous covariates representing the patient characteristics of interest, and treatment-by-covariate interaction effects. Parametrically, a subset-specific treatment effect is equal to the treatment effect term plus a linear combination of the interaction terms. We present Bayesian point and interval estimates under the assumption that the interaction terms are exchangeable and the prior distributions for the other regression parameters are locally uniform. This produces a shrinking of the estimated interaction effects towards zero, thereby discounting them and dealing in a natural way with multiplicity. We illustrate the method using results of a recent North Central Cancer Treatment Group/Mayo Clinic study in advanced colorectal cancer.

Antineoplastic Combined Chemotherapy Protocols↗

PubMed1988

The accuracy and stability of Bayesian theophylline predictions.

Pharmacokinetic parameters for theophylline were determined in 33 patients (3 women), mean age 61.2 years and weight 74.6 kg using the following three methods: (a) standard one-compartmental model calculations, assuming 100% bioavailability, after a single dose of theophylline syrup (mean dose 413 mg); (b) drug nomogram; and (c) Bayesian analysis. Patients entered a randomised study of three two-monthly dosage regimens using low, medium, and high theophylline twice daily doses. These doses produced mean (+/- SE) steady-state serum theophylline concentrations of 6.3 (+/- 0.4), 12.1 (+/- 0.3) and 18.3 (+/- 0.5) mg/L, respectively. A fourth period of placebo (2-month duration) was also included. At the end of each treatment period the measured serum theophylline concentration of each patient was compared with those predicted by each of the above three methods. The revised estimates derived from Bayesian analysis produced the least biased [mean prediction error (ME)] and most precise (mean squared prediction error) predictions for all three dosage periods. Statistical analysis of relative performance demonstrated that the difference in precision between the revised estimates and those of the other two methods was significant (p less than 0.05) with the magnitude of the difference increasing with dose. The revised estimates were also found to be less biased (p less than 0.05) than those of the nomogram. The ME (+/- SE) of the revised estimates for the low, medium, and high dosage periods was 0.34 (+/- 0.30), -0.02 (+/- 0.22) and -0.48 (+/- 0.31) mg/L, respectively.

Bayes Theorem↗

PubMed1991

An assessment of population-based and Bayesian methods to individualize digoxin doses shortly after the start of therapy for atrial fibrillation.

The accuracy of population-based methods and of Bayesian analysis to predict individual digoxin pharmacokinetic variables have been evaluated by their ability to predict a measured peak and trough serum digoxin concentration. We studied 13 digitalized patients (three women) whose mean (range) age and weight was 65.8 (60-78) years and 76.6 (68-101.6) kg and who had stable renal function. The population-based methods (using a clearance of 48.87 + 0.87 x creatinine clearance in ml/h/kg and volume of distribution, in litres, of either 7.3 x weight (kg) or 269 + 3.12 x creatinine clearance) were more than adequate for clinical purposes. The mean prediction errors of a measured steady-state peak concentration from these two population methods were -0.074 and 0.013 microgram/l respectively, whilst those of a measured trough concentration were -0.058 and 0.005 microgram/l. Bayesian analysis, using a sample drawn 11 h after the dose on day five of therapy, gave overall the least biased and most precise of the revised estimates. The mean prediction errors of peak and trough values using this sample were 0.069 and -0.005 microgram/l respectively. As expected, the closer the sample was drawn to the time of the trough concentration the more precise were the Bayesian-derived predictions. The value of the Bayesian technique to individualize digoxin doses could not be validated because it was not possible to distinguish between this and the population methods.(ABSTRACT TRUNCATED AT 250 WORDS)

Aged↗

PubMed1993

Computer dosing program for the initiation of vancomycin therapy.

The predictive performance of a computer dosing program used for initiating vancomycin therapy was studied. Initial serum vancomycin concentrations in 31 adult patients receiving vancomycin were estimated by using a computer program (T.D.M.S.) incorporating a two-compartment open model. Sixty-two serum vancomycin concentrations at steady state (Css) were obtained before and after one-hour infusions and compared with estimated Css values. Bias and precision were evaluated by calculating median error (ME) and median absolute error (MAE), respectively. Population-based estimates of volume of distribution (V) and clearance (CL) were compared with those obtained by fitting each patient's data set by using Bayesian analysis (BA) and non-linear least-squares regression (NLLS). Median (mean +/- S.D.) bias and precision for peak Css were 7.7 (10.2 +/- 10.8) and 7.7 (10.6 +/- 10.5) mg/L, and for trough Css were 7.4 (7.7 +/- 7.6) and 7.4 (8.8 +/- 6.2) mg/L. The medians were significantly different from zero. Estimated median (mean +/- S.D.) V, CL, and half-life were 0.72 L/kg, 0.60 (0.67 +/- 0.21) mL/min/kg, and 11.59 (12.87 +/- 3.91) hours. Median (mean +/- S.D.) CL values determined by BA and NLLS were 0.86 (0.89 +/- 0.32) and 0.85 (0.92 +/- 0.34) mL/min/kg, respectively. Both CL values were significantly greater than the population-based estimate. However, median V values determined by BA and NLLS did not differ from the population-based estimate. A revised clearance model derived from Bayesian analysis of data for the first 21 patients was tested in the 10 other patients and appeared to improve the predictive performance of the a priori model.(ABSTRACT TRUNCATED AT 250 WORDS)

Adult↗

PubMed1991

The prediction of steady-state plasma phenobarbitone concentrations (following low-dose phenobarbitone) to refine its use as an indicator of compliance.

1. A model for predicting the steady-state plasma concentration of phenobarbitone following low-dose phenobarbitone used as an indicator of compliance was derived using data for 10 healthy volunteers. 2. Each volunteer was given a single 30 mg oral dose of phenobarbitone and the pharmacokinetics were described. Subsequently, volunteers were given phenobarbitone 2 mg daily for 28 days and a further pharmacokinetic profile determined during and after this period. 3. An initial predicted estimate of steady-state plasma drug concentration was made using each volunteer's demographic details. This estimate was revised by Bayesian analysis using single timed samples (24, 48, 72 or 96 h) following the single dose. 4. The model was tested on a further 10 healthy volunteers given a single 8 mg dose and who were subsequently given 2 mg daily for 28 days. 5. The revised estimate of peak steady-state plasma phenobarbitone concentration utilising the 96 h post-single dose concentration (356 ng ml-1) was least biased (mean prediction error +/- 95% CI = 10.6 +/- 19.8 ng ml-1) and most precise (root mean square error +/- 95% CI = 28.3 +/- 19.0 ng ml-1). In all cases the peak or trough steady-state drug concentration was within 13% of the predicted value. 6. The model reflected compliance accurately in a further eight volunteers with simulated partial (two-thirds) compliance. 7. The use of a predictive model using Bayesian analysis to estimate expected steady-state plasma phenobarbitone concentrations could increase further the usefulness of low-dose phenobarbitone as an indicator of compliance.

Adult↗