Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Predictive”

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 919 records · Page 51Linked to original sources
PubMed2004

Prediction of lymph node metastasis with use of artificial neural networks based on gene expression profiles in esophageal squamous cell carcinoma.

BACKGROUND: The aim of the study was (1) to detect candidate genes involved in lymph node metastasis in esophageal cancers and (2) to investigate whether we can estimate and predict occurrence of lymph node metastasis by analyzing artificial neural networks (ANNs) using these gene subsets. METHODS: Twenty-eight primary esophageal squamous cell carcinomas were used. Gene expression profiles of all primary tumors were obtained by cDNA microarray. Lymph node metastasis-related genes were extracted with use of Significance Analysis of Microarrays (SAM). Predictive accuracy for lymph node metastasis was calculated by evaluation of 28 cases by ANNs with leave-one-out cross-n. The results were compared with those of other analyses such as clustering or predictive scoring (LMS). RESULTS: Our ANN model could predict lymph node metastasis most accurately with 60 clones. The highest predictive accuracy for lymph node metastasis by ANN was 10 of 13 (77%) in newly added cases that were not used for gene selection by SAM and 24 of 28 (86%) in all cases (sensitivity: 15/17, 88%; specificity: 9/11, 82%). Predictive accuracy of LMS was 9 of 13 (69%) in newly added cases and 24 of 28 (86%) in all cases (sensitivity: 17/17, 100%; specificity: 7/11, 67%). It was difficult to extract useful information for the prediction of lymph node metastasis by clustering analysis. CONCLUSIONS: ANN had superior potential in comparison with other methods of analysis for the prediction of lymph node metastasis. This systematic analysis combining SAM with ANN was very useful for the prediction of lymph node metastasis in esophageal cancers and could be applied clinically in the near future.

Adult↗
PubMed1996

Prediction of rodent carcinogenicity bioassays from molecular structure using inductive logic programming.

The machine learning program Progol was applied to the problem of forming the structure-activity relationship (SAR) for a set of compounds tested for carcinogenicity in rodent bioassays by the U.S. National Toxicology Program (NTP). Progol is the first inductive logic programming (ILP) algorithm to use a fully relational method for describing chemical structure in SARs, based on using atoms and their bond connectivities. Progol is well suited to forming SARs for carcinogenicity as it is designed to produce easily understandable rules (structural alerts) for sets of noncongeneric compounds. The Progol SAR method was tested by prediction of a set of compounds that have been widely predicted by other SAR methods (the compounds used in the NTP's first round of carcinogenesis predictions). For these compounds no method (human or machine) was significantly more accurate than Progol. Progol was the most accurate method that did not use data from biological tests on rodents (however, the difference in accuracy is not significant). The Progol predictions were based solely on chemical structure and the results of tests for Salmonella mutagenicity. Using the full NTP database, the prediction accuracy of Progol was estimated to be 63% (+/- 3%) using 5-fold cross validation. A set of structural alerts for carcinogenesis was automatically generated and the chemical rationale for them investigated- these structural alerts are statistically independent of the Salmonella mutagenicity. Carcinogenicity is predicted for the compounds used in the NTP's second round of carcinogenesis predictions. The results for prediction of carcinogenesis, taken together with the previous successful applications of predicting mutagenicity in nitroaromatic compounds, and inhibition of angiogenesis by suramin analogues, show that Progol has a role to play in understanding the SARs of cancer-related compounds.

Animals↗
PubMed2001

Establishment of the prediction table of parturition day with ultrasonography in small pet dogs.

To establish a prediction table of parturition day the real-time B-mode ultrasonographic examinations were performed in the 8 pregnant Malteses and 10 Yorkshire terriers (total pups, 25 and 38 pups, respectively) from 18 days of gestation until the parturition. Ovulation was designated the first day of gestation (day 0). Extra fetal and fetal structures were measured from all conceptues. The parameters that exhibited the best correlation to parturition were used to compile a prediction table of parturition day. To testify the precision of the prediction table of parturition day, the 15 pregnant Malteses (48 pups) and 13 pregnant Yorkshire terriers (42 pups) with unknown mating time were examined using ultrasonography. Inner chorionic cavity diameter on days 18 to 37 and fetal head diameter on day 38 to parturition that showed the best correlation to gestational age were the most pertinent to the estimation of gestational age and the prediction of parturition day. The two parameters were used to compile a prediction table of parturition with averaged regression equations. In verificational examinations, with the exception of I Yorkshire terrier (3.6%) having 1 fetus, 18 of 28 bitches (64.3%) delivered exactly on the date predicted and 9 of 28 bitches (32.1%) delivered within I day of the date predicted. Therefore, the prediction table of parturition day seems to be a useful tool of the prediction of parturition day in practice.

Animals↗
PubMed2005

Evaluation of Bayesian predictability of vancomycin concentration using population pharmacokinetic parameters in pediatric patients.

The objective of this study was to evaluate the Bayesian predictability of vancomycin (VCM) pharmacokinetics in Japanese pediatric patients using one-compartment population pharmacokinetic (PPK) parameters, which we reported previously. The validity of the PPK model was evaluated by bootstrap method and cross validation method, and the Bayesian predictive performance was examined. The predictive performance of the PPK model for premature patients was also examined. The cross validation method showed the predictability to be acceptable for practical use, especially for predicting trough concentration using other trough data. However, for the external premature patient data, this PPK model did not seem to be adequate. A theoretical approach using a simulation technique was also examined to evaluate the predictive performance. The results suggested that the predictability at the peak was not necessarily good at all sampling times and the predictability at the trough was better when a later time point was used. The optimal sampling time for prediction of VCM concentration in pediatric patients is discussed.

Algorithms↗
PubMed2000

Diabetes in urban African-Americans. XIX. Prediction of the need for pharmacological therapy.

OBJECTIVE: To develop a prediction rule that will identify patients who will require pharmacological therapy within 6 months of first presentation to a diabetes clinic. RESEARCH DESIGN AND METHODS: Among the patients who came to the Grady Diabetes Clinic between 1991 and 1997, we randomized 557 frequent attenders to a development group and 520 frequent attenders to a validation group. Using multiple logistical regression, we derived a prediction rule in the development group to project whether patients would require pharmacological intervention to achieve HbA1c levels <7% after 6 months. The utility of the prediction rule was then confirmed in the validation group and tested prospectively on an additional group of 93 patients who presented from 1997 to 1998. Performance of the prediction rule was assessed using receiver operating characteristic (ROC) curves. RESULTS: The rule (-4.469 + 1.932 x sulfonylurea Rx + 1.334 x insulin Rx + 0.196 x duration + 0.468 x fasting glucose, where "Rx" indicates a prescription) predicted the need for pharmacological intervention in the development group (P < 0.0001). Use of insulin or sulfonylurea therapy at presentation, duration of diabetes, and fasting glucose levels were significant predictors of the future need for pharmacological management. The prediction rule also performed well in the validation group (positive predictive value 90%, correlation between predicted and observed need for medical management 0.99). ROC curves confirmed the value of the prediction rule (area under the curves was 0.91 for the development group, 0.85 for the validation group, and 0.81 for the prospective group). CONCLUSIONS: Early identification of individuals who will require pharmacological intervention to achieve national standards for glycemic control can be achieved with high probability, thus allowing for more efficient management of diabetes.

Black or African American↗
PubMed1993

A net carbohydrate and protein system for evaluating cattle diets: IV. Predicting amino acid adequacy.

The Cornell Net Carbohydrate and Protein System was modified to include an amino acid submodel for predicting the adequacy of absorbed essential amino acids in cattle diets. Equations for predicting the supply of and requirements for absorbed essential amino acids are described and presented. The model was evaluated for its ability to predict observed duodenal flows of nitrogen, nonammonia nitrogen, bacterial nitrogen, dietary nonammonia nitrogen, and individual essential amino acids. Model-predicted nitrogen, nonammonia nitrogen, bacterial nitrogen, and dietary nonammonia nitrogen explained 93.2, 94.6, 76.4, and 79.3% of the observed duodenal flows, respectively, based on R2 values from predicted vs observed regression analysis. Based on slopes of regression lines, model-predicted duodenal nitrogen and nonammonia nitrogen were different from observed duodenal flows (P < .05), whereas model-predicted bacterial nitrogen and dietary nonammonia nitrogen were not different from observed duodenal flows (P < .05). Model-predicted duodenal flows of individual essential amino acids explained 81 to 90% of variation in observed duodenal amino acid flows. Based on slopes of regression lines, model-predicted duodenal threonine, leucine, and arginine were the only amino acids different from observed duodenal flows (P < .05). Ideas for further model improvements and research in amino acid metabolism were also presented.

Absorption↗
PubMed1986

Psychoacoustical and audiometric prediction of auditory disability for different frequency responses at listener-adjusted presentation levels.

Audiometric prediction of word identification scores has typically used one fixed presentation level for all subjects in the sample, with presentation in quiet and a wide range of hearing impairment among the listeners; under such conditions it is hardly surprising that moderate to good predictions are found. To see if prediction is possible under clinically relevant conditions, that is, on a homogeneous clinical sample of new hearing-aid candidates and to listener-adjusted levels, as would obtain in use of a hearing aid. In addition to audiometric variables, we employed a clinical approximation to the psychoacoustic tuning curve. We tested speech identification (FAAF) performance both with a 'rising'(+9 dB/octave) and with a 'flat' frequency response. Prediction of performance in the 'flat' condition was only good when a full set of audiometric frequencies entered the multiple-regression formula, each with its own weighting. Audiometric prediction for the 'rising' frequency response was particularly poor. Thus, the fairly good predictability from thresholds found traditionally for word identification scores or other disability measures appears to be a special case, depending partly on the wide range of hearing levels employed. Within our clinical sample the predictive power of formulae based on the mean of all thresholds or of mid-frequency thresholds alone (as used in compensation schemes) or on a priori combinations of thresholds (such as slopes) was generally poor. However, a three-parameter model taking account separately of low (0.25 kHz) and high-frequency (greater than 2.0 kHz) thresholds was effective. This and other audiometric descriptions were valuably supplemented by a psychoacoustic measure of frequency resolution at 2 kHz. In particular, such supplementation here allowed a satisfactory level of prediction to be achieved for speech heard with a +9 dB/octave frequency response, which the audiogram alone did not. The limitations of the prediction paradigm are discussed and several conceptual and statistical problems not previously emphasised in the audiological literature are illustrated in relation to the data.

Adult↗
PubMed2006

Comparison of net portal absorption with predicted flow of digestible amino acids: scope for improving current models?

This study was undertaken to determine the relationship between measured net portal absorptions (NPA) and flows of digestible essential amino acids (EAA) predicted with the National Research Council model (NRC, 2001) or the Cornell Net Carbohydrate and Protein System model (CNCPS, version 5.0.34). Net portal absorption data were obtained from 33 measurements of portal-arterial plasma EAA concentration differences among 8 treatments in lactating dairy cows, with plasma flow estimated from downstream dilution of para amino-hippurate. The predicted digestible flows from NRC (2001) related better than CNCPS to NPA observed in our studies, as shown by the lower standard errors on the slopes for all EAA and lower root mean prediction errors for all EAA except Met and Phe. However, the partitioning of the prediction error indicated a systematic underprediction (mean bias) for the NRC model (2001), with the exception of Ile. It is important to note that a relationship of unity was not expected, as discussed in the paper, because of losses of EAA through portal-drained viscera metabolism. A revised set of predictive equations for digestible EAA was obtained using a subset of data from NRC (2001) limited to trials conducted with dairy cows. This increased the predicted flows of digestible EAA by only 2%. Flows of digestible EAA were also estimated using a factorial approach, assuming an AA composition for each fraction of the duodenal flow estimated by NRC (undegradable, microbial, and endogenous proteins). This resulted in a slight improvement in the slope of the regression between predicted flows and measured NPA, but still yielded predicted digestive flows that were too low to support observed NPA. Finally, on the basis of literature values, increment of the digestibility of the undegradable fraction of forages and of microbial protein is suggested to improve the relationship between predicted digestible flows and NPA. Overall, this study indirectly confirms, across EAA, smaller losses through gut metabolism for His, Met, and Lys, intermediate losses for the branched-chain AA with the higher losses for Thr.

Amino Acids, Essential↗
PubMed2006

Accounting for energy and protein reserve changes in predicting diet-allowable milk production in cattle.

Current ration formulation systems used to formulate diets on farms and to evaluate experimental data estimate metabolizable energy (ME)-allowable and metabolizable protein (MP)-allowable milk production from the intake above animal requirements for maintenance, pregnancy, and growth. The changes in body reserves, measured via the body condition score (BCS), are not accounted for in predicting ME and MP balances. This paper presents 2 empirical models developed to adjust predicted diet-allowable milk production based on changes in BCS. Empirical reserves model 1 was based on the reserves model described by the 2001 National Research Council (NRC) Nutrient Requirements of Dairy Cattle, whereas empirical reserves model 2 was developed based on published data of body weight and composition changes in lactating dairy cows. A database containing 134 individually fed lactating dairy cows from 3 trials was used to evaluate these adjustments in milk prediction based on predicted first-limiting ME or MP by the 2001 Dairy NRC and Cornell Net Carbohydrate and Protein System models. The analysis of first-limiting ME or MP milk production without adjustments for BCS changes indicated that the predictions of both models were consistent (r(2) of the regression between observed and model-predicted values of 0.90 and 0.85), had mean biases different from zero (12.3 and 5.34%), and had moderate but different roots of mean square errors of prediction (5.42 and 4.77 kg/d) for the 2001 NRC model and the Cornell Net Carbohydrate and Protein System model, respectively. The adjustment of first-limiting ME- or MP-allowable milk to BCS changes improved the precision and accuracy of both models. We further investigated 2 methods of adjustment; the first method used only the first and last BCS values, whereas the second method used the mean of weekly BCS values to adjust ME- and MP-allowable milk production. The adjustment to BCS changes based on first and last BCS values was more accurate than the adjustment to BCS based on the mean of all BCS values, suggesting that adjusting milk production for mean weekly variations in BCS added more variability to model-predicted milk production. We concluded that both models adequately predicted the first-limiting ME- or MP-allowable milk after adjusting for changes in BCS.

Adipose Tissue↗
PubMed2026

Machine Learning-Based Preoperative Predicting TERT Promoter Mutation and EGFR Gene Amplification Phenotype in IDH Wild-Type Glioblastoma Using Advanced MR Habitat Imaging.

BACKGROUND AND PURPOSE: The telomerase reverse transcriptase (TERT) gene promoter mutation is a crucial factor for identifying an isocitrate dehydrogenase (IDH) wild-type glioblastoma with poor prognosis, and the epidermal growth factor receptor (EGFR) amplification may be a potential prognostic factor. The purpose of this study was to investigate the value of the tumor habitats imaging model on advanced MRI in predicting TERT promoter mutation and EGFR gene amplification phenotype of IDH wild-type glioblastoma. MATERIALS AND METHODS: One hundred seventy-nine patients with pretreatment conventional MRI, DWI, and DSC-PWI were included. The data were divided into the training set (n=112), test set (n=29), and time-independent validation set (n=38). Based on the ADC and CBV map, the solid tumor area was split into several habitat subregions using the k-means clustering algorithm (hypovascular hypercellular area, hypervascular area, and hypovascular hypocellular area). In the training set, TERT promoter mutation and EGFR gene amplification phenotype prediction models were constructed using the random forest method. The reliability of prediction models was validated in the test and the time-independent validation sets. Receiver operating characteristic (ROC) curve analysis, calibration curve, and decision curve analysis (DCA) were used. RESULTS: The area under the curve (AUC) of the training, test, and validation sets of the TERT promoter prediction model was 0.877, 0.783, and 0.796, respectively. The accuracy of the TERT promoter prediction model was 82.1%, 75.9%, and 76.3%, respectively. The AUCs of the 3 sets for the EGFR gene amplification status prediction model were 0.877, 0.784, and 0.878, respectively. The accuracy of the EGFR gene amplification status prediction model was 79.5%, 75.9%, and 89.5%, respectively. Moreover, the prediction probability of these models was in good agreement with the actual result. CONCLUSIONS: The tumor habitat imaging model based on advanced MRI was useful for accurately predicting TERT promoter mutation and EGFR amplification status in IDH wild-type glioblastoma.

Humans↗
PubMed2002

Positive predictive value of a point-of-care testing strategy on first-draw specimens for the emergency department-based detection of acute coronary syndromes.

CONTEXT: The rapid and accurate diagnosis of the etiology of chest pain is of central importance in the triage of patients presenting to emergency departments. The "first-draw" sensitivity of serum cardiac markers is known to be low on initial presentation; however, less is understood regarding the predictive value of a positive test in this situation. OBJECTIVE: To determine the ability of a critical pathway combining medical history and physical examination, electrocardiographic findings, point-of-care testing, and central laboratory data to accurately predict the presence of acute coronary ischemia. METHODS: We investigated the positive predictive value of a testing algorithm for first-draw specimens in clinical practice, combining a qualitative, point-of-care, triple-screen testing panel for cardiac markers, including myoglobin, creatine kinase-MB, and cardiac troponin I, with confirmation of the rapid assay in the central hospital laboratory by quantitative assays for creatine kinase-MB and cardiac troponin T. RESULTS: While a positive result on any of the individual cardiac markers of the point-of-care test had a positive predictive value for the acute coronary syndrome of only 36% (creatine kinase-MB, 41%; myoglobin, 36%; and troponin I, 65%), the positive predictive value for the diagnosis of acute coronary syndrome increased to 76% if all 3 point-of-care markers were simultaneously positive. The positive predictive value for acute coronary syndrome for a positive confirmatory result in the hospital laboratory for either creatine kinase-MB or cardiac troponin T was 61%. Among those patients with a positive marker on both the point-of-care test and the laboratory test, a careful retrospective review of the clinical history (with exclusion of patients with nonischemic cardiac pathologies and renal insufficiency) increased the positive predictive value of this algorithm to 98%. CONCLUSIONS: Our data suggest that qualitative, point-of-care, triple-screen cardiac marker testing of patients with chest pain at initial presentation may exhibit relatively low positive predictive values. Positive predictive value can be significantly improved by rapid confirmation in the hospital laboratory and careful review of clinical findings.

Chest Pain↗
PubMed2001

Birth weight prediction by three-dimensional ultrasonography: fractional limb volume.

OBJECTIVE: To introduce fractional limb volume as a new ultrasonographic parameter, validate reliability of fractional limb volume measurements, develop new birth weight prediction models, and examine their practical utility for estimating fetal weight during late pregnancy. METHODS: Healthy late-third-trimester fetuses were prospectively scanned by two- and three-dimensional ultrasonography within 4 days of delivery. Volume data sets were subsequently used to extract several standard ultrasonographic measurements. Fractional limb volumes of the upper arm and thigh were based on 50% of diaphyseal bone length. Intraclass correlation was used to analyze interobserver and intraobserver reliability of fractional limb volume measurements. Several weight prediction models were developed by linear regression analysis. New prediction models were prospectively compared with the Hadlock formula in 30 healthy late-third-trimester fetuses. RESULTS: One hundred fetuses were scanned at a mean +/- SD menstrual age of 39.2 +/- 1.2 weeks. Intraclass correlation indicated a significant degree of interobserver and intraobserver reliability for fractional thigh volume. Fractional thigh volume (r = 0.86), fractional upper arm volume (r = 0.83), abdominal circumference (r = 0.83), and midthigh circumference (r = 0.82) were most highly correlated with birth weight. The best prediction model (abdominal circumference and fractional thigh volume) gave weight estimates that deviated from actual birth weight by -0.025% +/- 7.8%. For late-third-trimester fetuses, the Hadlock model yielded errors of 9.0% +/- 9.0%. Prospective testing confirmed superior performance of the new prediction model, which gave accuracy of 2.3% +/- 6.6% (Hadlock method, 8.4% +/- 8.7%). It correctly predicted 20 of 30 birth weights to within 5% of actual weight. By comparison, the Hadlock model predicted only 6 of 30 birth weights to within 5% of actual weight. CONCLUSIONS: A new birth weight prediction model, based on fractional thigh volume and abdominal circumference, is reliable during the late third trimester. It provides a means for including soft tissue evaluation for birth weight prediction. This rapid technique avoids technical limitations that currently hinder the practical implementation of three-dimensional ultrasonography for estimating birth weight.

Birth Weight↗
PubMed1999

Structure-based prediction of DNA target sites by regulatory proteins.

Regulatory proteins play a critical role in controlling complex spatial and temporal patterns of gene expression in higher organism, by recognizing multiple DNA sequences and regulating multiple target genes. Increasing amounts of structural data on the protein-DNA complex provides clues for the mechanism of target recognition by regulatory proteins. The analyses of the propensities of base-amino acid interactions observed in those structural data show that there is no one-to-one correspondence in the interaction, but clear preferences exist. On the other hand, the analysis of spatial distribution of amino acids around bases shows that even those amino acids with strong base preference such as Arg with G are distributed in a wide space around bases. Thus, amino acids with many different geometries can form a similar type of interaction with bases. The redundancy and structural flexibility in the interaction suggest that there are no simple rules in the sequence recognition, and its prediction is not straightforward. However, the spatial distributions of amino acids around bases indicate a possibility that the structural data can be used to derive empirical interaction potentials between amino acids and bases. Such information extracted from structural databases has been successfully used to predict amino acid sequences that fold into particular protein structures. We surmised that the structures of protein-DNA complexes could be used to predict DNA target sites for regulatory proteins, because determining DNA sequences that bind to a particular protein structure should be similar to finding amino acid sequences that fold into a particular structure. Here we demonstrate that the structural data can be used to predict DNA target sequences for regulatory proteins. Pairwise potentials that determine the interaction between bases and amino acids were empirically derived from the structural data. These potentials were then used to examine the compatibility between DNA sequences and the protein-DNA complex structure in a combinatorial "threading" procedure. We applied this strategy to the structures of protein-DNA complexes to predict DNA binding sites recognized by regulatory proteins. To test the applicability of this method in target-site prediction, we examined the effects of cognate and noncognate binding, cooperative binding, and DNA deformation on the binding specificity, and predicted binding sites in real promoters and compared with experimental data. These results show that target binding sites for several regulatory proteins are successfully predicted, and our data suggest that this method can serve as a powerful tool for predicting multiple target sites and target genes for regulatory proteins.

Base Sequence↗
PubMed2000

Toward more rational prediction of outcome in patients with high-grade subarachnoid hemorrhage.

OBJECTIVE: Accurate outcome prediction after high-grade subarachnoid hemorrhage remains imprecise. Several clinical grading scales are in common use, but the timing of grading and changes in grade after admission have not been carefully evaluated. We hypothesized that these latter factors could have a significant impact on outcome prediction. METHODS: Fifty-six consecutive patients with altered mental status after subarachnoid hemorrhage, who were managed at a single institution, were studied retrospectively. On the basis of prospectively assessed elements of the clinical examination, each patient was graded at admission, at best before treatment, at worst before treatment, immediately before treatment, and at best within 24 hours after treatment of the aneurysm using the Glasgow Coma Scale (GCS), the World Federation of Neurological Surgeons (WFNS) scale, and the Hunt and Hess scale. Outcome at 6 months was determined using a modification of the Glasgow Outcome Scale validated against the Karnofsky scale. All grades and clinical and radiographic data collected were compared among good and poor outcome groups. Multivariate analyses were then performed to determine which grading scale, which time of grading, and which other factors were correlated with and contributed significantly to outcome prediction. RESULTS: A good outcome was achieved in 24 (43%) of 56 patients. Our study also had a 32% mortality rate. With the Hunt and Hess scale, only the worst pretreatment grade was significantly correlated with outcome. However, with the GCS and the WFNS scale, grading at all pretreatment times was significantly correlated with outcome, although outcome was best predicted before treatment, regardless of the scale used, if grading was performed at the patient's clinical worst. Multivariate analysis revealed that the best predictor of outcome was WFNS grade at clinical worst before treatment. Used alone, a WFNS Grade 3 at worst pretreatment predicted a 75% favorable outcome, and a WFNS Grade 5 at worst pretreatment predicted an 87% poor outcome. No significant correlation was found between direction or magnitude of change in grade and outcome. Age was found to be significantly correlated with outcome, but it was only an independent factor in outcome prediction when used in conjunction with the Hunt and Hess scale and not with the WFNS scale and the GCS. CONCLUSION: Timing of grading is an important factor in outcome prediction that needs to be standardized. This study suggests that the patient's worst clinical grade is most predictive of outcome, especially when the patient is assessed using the WFNS scale or the GCS.

Adult↗
PubMed2000

Combination of symptom score, flow rate and prostate volume for predicting bladder outflow obstruction in men with lower urinary tract symptoms.

PURPOSE: The severity of lower urinary tract symptoms associated with benign prostatic enlargement correlates poorly with bladder outlet obstruction. Since urodynamic studies are presumed to be relatively complex, invasive and not cost-effective, they are not routinely performed by physicians treating men with lower urinary tract symptoms. As a result, a large number of patients are treated for bladder outlet obstruction when in fact obstruction may not be present. Since other noninvasive methods have not been effective for predicting bladder outlet obstruction, we investigated whether a combination of prostate volume, uroflowmetry and the American Urological Association (AUA) symptom index would be reliable for predicting this condition. MATERIALS AND METHODS: We prospectively evaluated 204 men with a mean age plus or minus standard deviation of 66.7 +/- 7.5 years who presented with lower urinary tract symptoms. Each patient completed an AUA symptom index questionnaire and underwent uroflowmetry, post-void residual urine volume measurement, pressure flow study and transrectal ultrasound of the prostate to estimate prostatic volume. We constructed receiver operating characteristics curves using various threshold values for maximum urine flow and prostate volume. Threshold values for maximum urine flow and prostate volume were used alone and combined with the AUA symptom index for predicting bladder outlet obstruction. We selected a cutoff value for maximum urine flow of 10 or less ml. per second and prostate volume of 40 gm. or greater, and used these values with an AUA symptom index of greater than 20 to predict bladder outlet obstruction in the group overall. RESULTS: Differences in the mean symptom index score in men with and without bladder outlet obstruction were not statistically significant. There was no obstruction in 19%, 28.9% and 35% of those with severe, moderate and mild symptoms, respectively. The selected cutoff values of maximum urine flow, prostate volume and symptom score combined correctly predicted obstruction in all 39 patients. Therefore, our combination of cutoff values proved to be highly accurate for predicting bladder outlet obstruction. Sensitivity, specificity, and positive and negative predictive values were 26%, 100%, 100% and 32%, respectively. CONCLUSIONS: Our study showed that combining the AUA symptom index, maximum urine flow and prostate volume reliably predicted bladder outlet obstruction in a small subset of patients only. Although bladder outlet obstruction was correctly predicted by our threshold values of AUA symptom index, maximum urine flow and prostate volume in only 39 men (26%) with obstruction, these patients represent a substantial group in any large urological practice treating male lower urinary tract symptoms.

Aged↗
PubMed2001

The use of neural networks and logistic regression analysis for predicting pathological stage in men undergoing radical prostatectomy: a population based study.

PURPOSE: Clinical under staging occurs in 40% to 60% of patients who undergo radical prostatectomy for prostate cancer. To decrease under staging several methods of predicting pathological stage preoperatively have been developed based on statistical logistic regression analysis and neural networks. To our knowledge none has been validated in our homogeneous regional patient population to date. We created logistic regression and neural network models, and implemented and adapted them into our practice. We also compared the 2 methods to determine their value and practicality in daily clinical practice. We present the results of our novel approach for predicting pathological staging of prostate adenocarcinoma. MATERIALS AND METHODS: Between 1986 and 1999, 600 white men from the Aragon region of Spain underwent surgery for prostate cancer; of whom 468 were selected for study. Predictive study variables included patient age, clinical stage, biopsy Gleason score and preoperative prostate specific antigen (PSA). The predicted result included in analysis was organ confined or nonorgan confined disease. Data were analyzed by multivariate logistic regression and a supervised neural network (multilayer perceptron and radial basis function). Results were compared by comparing the areas under the receiver operating characteristics curves. RESULTS: We generated 5 logistic regression models. The model created with clinical staging, Gleason biopsy score and PSA distributed in 5 categories (p <0.001) with an area under the receiver operating characteristics curve of 0.840 proved to be most predictive of pathological stage. Similarly of the 6 neural network models evaluated the radial basis function model, which included age, clinical stage, Gleason biopsy score and preoperative PSA distributed in 5 categories with an area under the curve of 0.882, proved the most predictive but not superior to the logistic regression model. The difference in the area under the curves in the 2 chosen models was 0.042 (p = 0.1). CONCLUSIONS: It is possible to generate useful predictive models of organ confined disease using logistic regression or neural networks with high indexes of clinical and statistical validity. However, using these variables neural networks did not prove to be better than logistic regression analysis. Therefore, better predictive variables must be identified, preferably nonlinear characteristics with respect to the probability of organ confined tumor, to generate better predictive models using neural networks.

Aged↗
PubMed2001

Prediction of in-hospital mortality in patients with myocardial infarction using APACHE II system.

BACKGROUND: The acute physiology and chronic health evaluation (APACHE) scoring system has been validated in many different patient populations, however, patients with myocardial infarction (MI) were not included in the original data base. To evaluate the ability of APACHE scoring system in predicting in-hospital mortality, 694 patients with MI were studied. METHODS: Data had been collected prospectively in an ICU computer database in the past 3 years. Patients admitted in coronary care unit with acute MI or acute coronary syndrome who had previous history of MI were all included. Patients were divided into survivor and non-survivor data sets. Multiple logistic regression analysis was evaluated on the variables of APACHE II score to determine which variables could predict in-hospital mortality. A logistic regression model was used to study the mortality curves. The differences of APACHE II scores between survivors and non-survivors were compared. Correlation between observed and predicted mortality was also assessed. RESULTS: According to the statistical analysis, the non-survivors tended to have significantly greater APACHE II scores than those of survivors. The APACHE II values of non-survivors and survivors were 23.64 +/- 9.41 versus 13.35 +/- 7.14 (p < 0.001), respectively. Using multiple logistic regression analysis, we found that age, creatinine, coma scale, sodium and APACHE II score were capable of predicting the in-hospital mortality (p < 0.05). With use of the logistic model, a good correlation of predicted mortality rate to observed mortality rate was found (r = 0.992). This study demonstrated that lower APACHE II scores predicted survival while high scores predicted mortality. Mortality rate increased significantly when APACHE II score was > 25. An APACHE II score greater than 28.25 predicted a more than 50% in-hospital mortality. CONCLUSIONS: This study demonstrates that the APACHE II scoring system is capable of predicting mortality in patients with MI, which makes this modality more applicable in the busy intensive care unit.

APACHE↗
PubMed2002

Using Monte Carlo techniques to judge model prediction accuracy: validation of the pesticide root zone model 3.12.

Individuals from the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) Environmental Model Validation Task Force (FEMVTF) Statistics Committee periodically met to discuss the mechanism for conducting an uncertainty analysis of Version 3.12 of the pesticide root zone model (PRZM 3.12) and to identify those model input parameters that most contribute to model prediction error. This activity was part of a larger project evaluating PRZM 3.12. The goal of the uncertainty analysis was to compare site-specific model predictions and field measurements using the variability in each as a basis of comparison. Monte Carlo analysis was used as an integral tool for judging the model's ability to predict accurately. The model was judged on how well it predicts measured values, taking into account the uncertainty in the model predictions. Monte Carlo analysis provides the tool for inferring model prediction uncertainty. We argue that this is a fairer test of the model than a simple one-to-one comparison between predictions and measurements. Because models are known to be imperfect predictors prior to running the model, the inaccuracy in model predictions should be considered when models are judged for their predictive ability. Otherwise, complex models can easily fail a validation test. Few complex models, such as PRZM 3.12, would pass a typical model validation exercise. This paper describes the approaches to the validation of PRZM 3.12 used by the committee and discusses issues in sampling distribution selection and appropriate statistics for interpreting the model validation results.

Forecasting↗