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 379 records · Page 21Linked to original sources
PubMed2002

Evaluation of computational metabolic-pathway predictions for Helicobacter pylori.

MOTIVATION: We seek to determine the accuracy of computational methods for predicting metabolic pathways in sequenced genomes, and to understand the contributions of both the prediction algorithms, and the reference pathway databases used by those algorithms, to the prediction accuracy. RESULTS: The comparisons we performed were as follows. (1) We compared two predictions of the pathway complements of Helicobacter pylori that were computed by an early version of our pathway-prediction algorithm: prediction A used the EcoCyc E. coli pathway DB as the reference database (DB) for prediction, and prediction B used the MetaCyc pathway DB (a superset of EcoCyc) as the reference pathway DB. The MetaCyc-based prediction contained 75% more pathway predictions, but we believe a significant number of those predictions were false positives. (2) We compared two predictions of the pathway complement of H. pylori that used MetaCyc as the reference pathway DB, but that used different algorithms: the original PathoLogic algorithm, and an enhanced version of the algorithm designed to eliminate false-positive pathway predictions. The improved algorithm predicted 30\% fewer metabolic pathways than the original algorithm; all of the eliminated pathways are believed to be false-positive predictions. (3) We compared the 98 pathways predicted by the enhanced algorithm with the results of a manual analysis of the pathways of H. pylori. Results: 40 of the computationally predicted pathways were consistent with the manual analysis, 13 pathways are considered false-positive predictions, and four pathways had partially overlapping topologies. Twenty-six predicted pathways were not mentioned in the manual analysis; we believe these are correct predictions by PathoLogic that were not found by the manual analysis. Five pathways from the manual analysis were not found computationally. Agreement between the computational and manual predictions was good overall, with the computational analysis inferring many pathways that the manual analysis did not identify. Ultimately the manual analysis is also partially speculative, and therefore is not an absolute measure of correctness. The algorithm is designed to err on the side of more false positives to bring more potential pathways to the user's attention. The resulting H. pylori pathway DB is freely available at http://ecocyc.org:1555/HPY/organism-summary?object=HPY. AVAILABILITY: The Pathway Tools software is freely available to academic users, and is available to commercial users for a fee. Contact pkarp@ai.sri.com for information on obtaining the software.

Algorithms↗
PubMed2006

Prediction of in vivo drug-drug interactions from in vitro data : factors affecting prototypic drug-drug interactions involving CYP2C9, CYP2D6 and CYP3A4.

BACKGROUND: Quantitative predictions of in vivo drug-drug interactions (DDIs) resulting from metabolic inhibition are commonly made based upon the inhibitor concentration at the enzyme active site [I] and the in vitro inhibition constant (K(i)). Previous studies have involved the use of various plasma inhibitor concentrations as surrogates for [I] along with K(i) values obtained from published literature. Although this approach has resulted in a high proportion of successful predictions, a number of falsely predicted interactions are also observed. OBJECTIVES: To focus on three issues that may influence the predictive value of the [I]/K(i) ratio approach: (i) the use of unbound K(i) (K(i,u)) values generated from standardised in vitro experiments compared with literature values; (ii) the selection of an appropriate [I]; and (iii) incorporation of the impact of intestinal metabolic inhibition for cytochrome P450 (CYP) 3A4 predictions. To this end we have selected eight inhibitors of CYP2C9, CYP2D6 and CYP3A4 and 18 victim drugs from a previous database analysis to allow prediction of 45 clinical DDI studies. METHODS: In vitro kinetic and inhibition studies were performed in human liver microsomes using prototypic probe substrates of CYP2C9 and CYP2D6, with various inhibitors (miconazole, sulfaphenazole, fluconazole, ketoconazole, quinidine, fluoxetine, fluvoxamine). The K(i) estimates obtained were corrected for non-specific microsomal binding, and the K(i,u) was incorporated into in vivo predictions using various [I] values. Predictions for CYP3A4 were based upon in vitro data obtained from a previous publication within our laboratory, and an assessment of the impact of the interaction in the gut wall is included. Predictions were validated against 45 in vivo studies and those within 2-fold of the in vivo ratio of area under the plasma concentration-time curve of the substrate, in the presence and absence of the inhibitor (AUC(i)/AUC) were considered successful. RESULTS: Predictions based upon the average systemic total plasma drug concentration ([I](av)) [incorporating the effects of parallel drug elimination pathways] and the K(i,u) value resulted in 91% of studies predicted to within 2-fold of the in vivo AUC(i)/AUC. This represents a 35% improvement in prediction accuracy compared with predictions based upon total K(i) values obtained from various published literature sources. A corresponding reduction in bias and an increase in precision were also observed compared with the use of other [I] surrogates (e.g. the total and new unbound maximum hepatic input plasma concentrations). No significant improvement in prediction accuracy was observed by incorporating consideration of gut wall inhibition for CYP3A4. CONCLUSION: DDI predictions based upon the use of K(i,u) data obtained under a set of optimal standardised conditions were significantly improved compared with predictions using in vitro data collated from various sources. The use of [I](av) as the [I] surrogate generated the most successful predictions as judged by several criteria. Incorporation of either plasma protein binding of inhibitor or gut wall CYP3A4 inhibition did not result in a general improvement of DDI predictions.

Algorithms↗
PubMed1994

Increasing the power of surrogate endpoint biomarkers: the aggregation of predictive factors.

A variable that predicts an outcome with sufficient accuracy is called a predictive factor. Predictive factors can be divided into three types based on the outcomes to be predicted and on the accuracy with which they can be predicted. These three types include risk factors, where the main outcome of interest is incidence and the predictive accuracy is less than 100%; diagnostic factors, where the main outcome of interest is also incidence but the predictive accuracy is almost 100%; and prognostic factors, where the main outcome of interest is death and the predictive accuracy is variable. Surrogate outcomes are predictive factors that are used for a purpose beyond the prediction of an outcome--surrogate outcomes are predictive factors that are substituted for the true outcome in order to determine the effectiveness of an intervention. Surrogate outcomes used in clinical trials are called intermediate endpoints and surrogate endpoints. Predictive factors used as surrogate outcomes have a poor accuracy rate in predicting the true outcome; aggregating risk factors increases predictive accuracy. Artificial neural networks effectively combine predictive factors. Aggregating predictive factors increases the degree of linkage of the surrogate outcome to the true outcome. The resulting increase in predictive accuracy allows enrollment of people most likely to benefit from intervention. This increases the trial's efficiency, reducing the number of people required to assess a chemopreventive agent.

Anticarcinogenic Agents↗
PubMed2003

Validation and calibration of the Kabi Pharmacia International Growth Study prediction model for children with idiopathic growth hormone deficiency.

In 1999 a model was published for prediction of growth in children with idiopathic GH deficiency (IGHD) during GH therapy, derived using data from the Kabi Pharmacia International Growth Study (KIGS) database (Pharmacia \|[amp ]\| Upjohn, Inc., International Growth Database). We validated and calibrated this KIGS model for growth in the first year of GH therapy using data from 136 Dutch children with IGHD. Observed vs. predicted outcomes were plotted, and the fitted regression line was significantly different from the line of identity (P = 0.03). It appeared that the predictions were too extreme: relatively low predictions were too low, relatively high predictions were too high. This is a well known phenomenon in the context of prediction models, called overoptimism. For valid application to other data the KIGS predictions should be calibrated. Calibrated predictions are obtained using Y(cal) = Y(orig) + (2.153 - 0.192 x Y(orig)), where Y(cal) is the calibrated prediction, and Y(orig) is the KIGS prediction. The calibrated prediction will be higher than the original KIGS prediction when the original prediction is less than 11.2 cm/yr and will be lower otherwise. The variability of the prediction errors of the calibrated predictions was positively related to the value of the prediction (P < 0.001), described by the equation SD(pred err) = -1.017 + 0.286 x Y(cal). Our calibrated model will give better predictions for children with IGHD fulfilling the same criteria.

Calibration↗
PubMed1996

Accuracy of final height prediction and effect of growth-reductive therapy in 362 constitutionally tall children.

Height reduction by means of treatment with high doses of sex steroids in constitutionally tall stature (CTS) is a well known, though still controversial, therapy. The establishment of the effect of such therapy is dependent on the height prediction method applied. We evaluated the reliability of various prediction methods together with the subjective clinician's judgment in 143 untreated children (55 boys and 88 girls) with CTS and the effect of height-reductive therapy in 249 tall children (60 boys and 159 girls) treated with high doses of sex hormones (cases). For this purpose, we compared the predicted adult height with the attained height at a mean adult age of 25 yr and adjusted the therapeutic effect for differences in bone age (BA), chronological age (CA), and height prediction between untreated and treated children. At the time of the height prediction, controls were significantly shorter, had more advanced estimated BAs (except for the BA according to Greulich and Pyle in boys), had lower target heights, and had smaller adult height predictions compared with the CTS patients (P < 0.05). At the time of the follow-up, CTS patients were significantly taller than controls for both boys and girls (P < 0.02). In controls, a large variability was found for the errors of prediction of the various prediction methods and in relation to CA. The prediction according to Bailey and Pinneau systematically overestimated adult height in CTS children, whereas the other prediction methods (Tanner-Whitehouse prediction and index of potential height) systematically underestimated final height. The mean (SD) absolute errors of the prediction methods varied from 2.3 (1.8) to 5.3 (4.3) cm in boys and from 2.0 (1.9) to 3.7 (3.5) cm in girls. They were significantly negatively correlated with CA (r = [minus 0.27 to -0.65; P < 0.05), except for the Tanner-Whitehouse prediction in boys, indicating that height prognosis is more reliable with increasing CA. In addition, experienced clinicians gave accurate height predictions by evaluating the growth chart of the child while taking into account various clinical parameters, such as CA, BA, and pubertal stage. The effect of sex hormone therapy was assessed by means of multiple regression analysis while adjusting for differences in height prediction, CA, and BA at the start of therapy between treated and untreated children. The mean (SD) adjusted effect varied from -0.5 (2.4) to 0.3 (1.4) cm in boys and from -0.6 (2.1) to 2.4 (1.4) cm in girls. The adjusted height reduction was dependent on the BA at the time of start of sex hormone therapy and was more pronounced when treatment was started at a younger BA. In boys, the treatment effect was significantly negative at BAs exceeding 14-15 yr. After cessation of therapy, additional mean (SD) growth of 2.4 (1.2) and 2.7 (1.1) cm was observed for boys and girls, respectively. The mean (SD) BA according to Greulich and Pyle at that time was 17.1 (0.7) yr for boys and 15.2 (0.6) yr for girls. These data demonstrate that height prediction in children with CTS is inaccurate in boys, but clinically acceptable in girls. With increasing age, height prognosis became more accurate. Overall, the height-reducing effect of high doses of sex hormones in children with CTS was limited, especially in boys. However, a significant effect of treatment was observed when treatment was started at BAs less than 14-15 yr, depending on the method of BA assessment. In boys, treatment appeared to be contraindicated at BAs older than 14-15 yr, because androgen administration caused extra growth instead of growth inhibition. It is recommended that referral should take place early, preferably before puberty, for careful monitoring of growth and height prediction. Moreover, it is recommended not to discontinue therapy before complete closure of the epiphyses of the hand has occurred to avoid considerable posttreatment growth.

Adolescent↗
PubMed1999

Glasgow head injury outcome prediction program: an independent assessment.

Using an independent data set, the utility of the Glasgow Head Injury Outcome Prediction Program was investigated in terms of possible frequency of use and reliability of outcome prediction in patients with severe head injury, or haematoma requiring evacuation, or coma lasting 6 hours or more, in whom outcome had been reliably assessed at 6 to 24 months after injury. Predictions were calculated on admission, before evacuation of a haematoma, or 24 hours, 3 days, and 7 days after onset of coma lasting 6 hours or more. Three hundred and twenty four patients provided 426 predictions which were possible in 76%, 97%, 19%, 34%, and 53% of patients on admission, before operation, 24 hours, 3 days, and 7 days respectively. Major reasons for non-feasible predictions were that patients were paralysed/ventilated as part of resuscitation or management. Overall, 75.8% of predictions were correct, 14.6% were pessimistic (outcome better than predicted), and 9.6% optimistic (outcome worse than predicted). Of 197 patients (267 predictions) whose eventual outcome was good or moderate, 84.3% of predictions were correct. For death or vegetative survival (96 patients with 110 predictions), 83.6% of predictions were correct but for severe disability (31 patients with 49 predictions), only 12.2% were correctly predicted. The utility of the Glasgow Head Injury Outcome Prediction Program compares favourably with other outcome prediction algorithms for patients with head injury.

Adolescent↗
PubMed2002

Validation of a birth weight prediction equation based on maternal characteristics.

OBJECTIVE: To validate the accuracy of a birth weight prediction equation based on maternal and pregnancy-specific characteristics and to assess its value in predicting fetal macrosomia. STUDY DESIGN: A previously published birth weight prediction equation based on maternal and pregnancy-specific characteristics was used to predict birth weight in 244 Caucasian gravidas with uncomplicated, singleton, term pregnancies. Results were assessed by calculating the mean absolute error, the mean absolute percentage error and the percentage of birth weights correctly predicted to within +/- 10% and +/- 15% of actual birth weight. The sensitivity, specificity and positive and negative predictive value for predicting fetal weight > 4,000 g were calculated. RESULTS: Birth weight was accurately predicted to within an average of +/- 8.1% (+/- 280 g). The percentage of weights accurately predicted to within 15% of actual birth weight was 87%, and the percentage predicted to within +/- 10% was 68%. The sensitivity for predicting fetal weight > 4,000 g was 52%, specificity 90%, positive predictive value 42% and negative predictive value 93%. The area under the receiver operating characteristic curve for predicting fetal macrosomia was 0.83. CONCLUSION: An equation using maternal and pregnancy-specific characteristics can predict term birth weight in gravidas with uncomplicated singleton pregnancies to within +/- 8.1% (+/- 280 g). The accuracy of the method for predicting birth weight > 4,000 g is comparable to that obtained using ultrasonic fetal biometry.

Adult↗
PubMed1996

Prediction of secondary structural content of proteins from their amino acid composition alone. I. New analytic vector decomposition methods.

The predictive limits of the amino acid composition for the secondary structural content (percentage of residues in the secondary structural states helix, sheet, and coil) in proteins are assessed quantitatively. For the first time, techniques for prediction of secondary structural content are presented which rely on the amino acid composition as the only information on the query protein. In our first method, the amino acid composition of an unknown protein is represented by the best (in a least square sense) linear combination of the characteristic amino acid compositions of the three secondary structural types computed from a learning set of tertiary structures. The second technique is a generalization of the first one and takes into account also possible compositional couplings between any two sorts of amino acids. Its mathematical formulation results in an eigenvalue/eigenvector problem of the second moment matrix describing the amino acid compositional fluctuations of secondary structural types in various proteins of a learning set. Possible correlations of the principal directions of the eigenspaces with physical properties of the amino acids were also checked. For example, the first two eigenvectors of the helical eigenspace correlate with the size and hydrophobicity of the residue types respectively. As learning and test sets of tertiary structures, we utilized representative, automatically generated subsets of Protein Data Bank (PDB) consisting of non-homologous protein structures at the resolution thresholds < or = 1.8A, < or = 2.0A, < or = 2.5A, and < or = 3.0 A. We show that the consideration of compositional couplings improves prediction accuracy, albeit not dramatically. Whereas in the self-consistency test (learning with the protein to be predicted), a clear decrease of prediction accuracy with worsening resolution is observed, the jackknife test (leave the predicted protein out) yielded best results for the largest dataset (< or = 3.0A, almost no difference to the self-consistency test!), i.e., only this set, with more than 400 proteins, is sufficient for stable computation of the parameters in the prediction function of the second method. The average absolute error in predicting the fraction of helix, sheet, and coil from amino acid composition of the query protein are 13.7, 12.6, and 11.4%, respectively with r.m.s. deviations in the range of 8.6 divided by 11.8% for the 3.0 A dataset in a jackknife test. The absolute precision of the average absolute errors is in the range of 1 divided by 3% as measured for other representative subsets of the PDB. Secondary structural content prediction methods found in the literature have been clustered in accordance with their prediction accuracies. To our surprise, much more complex secondary structure prediction methods utilized for the same purpose of secondary structural content prediction achieve prediction accuracies very similar to those of the present analytic techniques, implying that all the information beyond the amino acid composition is, in fact, mainly utilized for positioning the secondary structural state in the sequence but not for determination of the overall number of residues in a secondary structural type. This result implies that higher prediction accuracies cannot be achieved relying solely on the amino acid composition of an unknown query protein as prediction input. Our prediction program SSCP has been made available as a World Wide Web and E-mail service.

Amino Acids↗
PubMed2006

Predicting residue-wise contact orders in proteins by support vector regression.

BACKGROUND: The residue-wise contact order (RWCO) describes the sequence separations between the residues of interest and its contacting residues in a protein sequence. It is a new kind of one-dimensional protein structure that represents the extent of long-range contacts and is considered as a generalization of contact order. Together with secondary structure, accessible surface area, the B factor, and contact number, RWCO provides comprehensive and indispensable important information to reconstructing the protein three-dimensional structure from a set of one-dimensional structural properties. Accurately predicting RWCO values could have many important applications in protein three-dimensional structure prediction and protein folding rate prediction, and give deep insights into protein sequence-structure relationships. RESULTS: We developed a novel approach to predict residue-wise contact order values in proteins based on support vector regression (SVR), starting from primary amino acid sequences. We explored seven different sequence encoding schemes to examine their effects on the prediction performance, including local sequence in the form of PSI-BLAST profiles, local sequence plus amino acid composition, local sequence plus molecular weight, local sequence plus secondary structure predicted by PSIPRED, local sequence plus molecular weight and amino acid composition, local sequence plus molecular weight and predicted secondary structure, and local sequence plus molecular weight, amino acid composition and predicted secondary structure. When using local sequences with multiple sequence alignments in the form of PSI-BLAST profiles, we could predict the RWCO distribution with a Pearson correlation coefficient (CC) between the predicted and observed RWCO values of 0.55, and root mean square error (RMSE) of 0.82, based on a well-defined dataset with 680 protein sequences. Moreover, by incorporating global features such as molecular weight and amino acid composition we could further improve the prediction performance with the CC to 0.57 and an RMSE of 0.79. In addition, combining the predicted secondary structure by PSIPRED was found to significantly improve the prediction performance and could yield the best prediction accuracy with a CC of 0.60 and RMSE of 0.78, which provided at least comparable performance compared with the other existing methods. CONCLUSION: The SVR method shows a prediction performance competitive with or at least comparable to the previously developed linear regression-based methods for predicting RWCO values. In contrast to support vector classification (SVC), SVR is very good at estimating the raw value profiles of the samples. The successful application of the SVR approach in this study reinforces the fact that support vector regression is a powerful tool in extracting the protein sequence-structure relationship and in estimating the protein structural profiles from amino acid sequences.

Algorithms↗
PubMed2006

A community resource benchmarking predictions of peptide binding to MHC-I molecules.

Recognition of peptides bound to major histocompatibility complex (MHC) class I molecules by T lymphocytes is an essential part of immune surveillance. Each MHC allele has a characteristic peptide binding preference, which can be captured in prediction algorithms, allowing for the rapid scan of entire pathogen proteomes for peptide likely to bind MHC. Here we make public a large set of 48,828 quantitative peptide-binding affinity measurements relating to 48 different mouse, human, macaque, and chimpanzee MHC class I alleles. We use this data to establish a set of benchmark predictions with one neural network method and two matrix-based prediction methods extensively utilized in our groups. In general, the neural network outperforms the matrix-based predictions mainly due to its ability to generalize even on a small amount of data. We also retrieved predictions from tools publicly available on the internet. While differences in the data used to generate these predictions hamper direct comparisons, we do conclude that tools based on combinatorial peptide libraries perform remarkably well. The transparent prediction evaluation on this dataset provides tool developers with a benchmark for comparison of newly developed prediction methods. In addition, to generate and evaluate our own prediction methods, we have established an easily extensible web-based prediction framework that allows automated side-by-side comparisons of prediction methods implemented by experts. This is an advance over the current practice of tool developers having to generate reference predictions themselves, which can lead to underestimating the performance of prediction methods they are not as familiar with as their own. The overall goal of this effort is to provide a transparent prediction evaluation allowing bioinformaticians to identify promising features of prediction methods and providing guidance to immunologists regarding the reliability of prediction tools.

Animals↗
PubMed1989

The effect of prediction accuracy on choice reaction time.

In this study, we examined the effect of prediction accuracy on reaction time (RT). Subjects performed on three blocks of choice RT trials, all of which involved the mapping of four stimuli (red, green, 1, or 0) onto two response keys. The subjects were told that the four stimuli were equally probable and that their task was to respond to each stimulus onset by pressing the correct key. In one block (stimulus-prediction), the subjects predicted, prior to each trial, the precise stimulus that would appear. In a second block (category-prediction), the subjects predicted the category of the stimulus (i.e., color or digit) that would appear. In a third block (no-prediction), the subjects simply responded to each stimulus without making a prior prediction. In the stimulus-prediction block, RT was faster for correct predictions than for incorrect predictions. In addition, RT was faster on trials in which an incorrect prediction involved the correct category than on trials in which it involved the incorrect category: that is, a "half-wrong" prediction was better than an "all-wrong" prediction. In the category-prediction block, RT was faster when the stimulus category was predicted correctly than when it was not. There was little evidence of a response-facilitation contribution to the correct-prediction effect. These results permit inferences concerning the encoding and organization of information in memory.

Adult↗
PubMed2005

Predicting pharmacy costs and other medical costs using diagnoses and drug claims.

BACKGROUND: Predicting health care costs for individuals and populations is essential for managing care. However, the comparative power of diagnostic and drug data for predicting future costs has not been closely examined. OBJECTIVE: We sought to compare the predictive performance of claims-based models using diagnoses, drugs claims, and combined data to predict health care costs. SUBJECTS: More than 1 million commercially insured, nonelderly individuals in a national (MEDSTAT MarketScan) research database comprised our sample. MEASURES: We used 1997 and 1998 drug and diagnostic profiles to predict costs in 1998 and 1999, respectively. To assess model performance, we compared R2 values and predictive ratios (predicted costs/actual costs) for important subgroups. RESULTS: Models using both drug and diagnostic data best predicted subsequent-year total health care costs (highest R2 = 0.168 versus 0.116 and 0.146 for models based on drug or diagnostic data alone, respectively), with highly accurate predictive ratios (0.95-1.05) for subgroups of patients with major medical conditions. Models predicting pharmacy costs had substantially higher R2 values than models predicting other medical costs (highest R2 0.493 versus 0.124). Drug-based models predicted future pharmacy costs better than diagnosis-based models (highest R2 = 0.482 versus 0.243), whereas diagnosis-based models predicted total costs (highest R2 = 0.146 versus 0.116) and nonpharmacy costs (highest R2 = 0.116 versus 0.071) more effectively than drug-based models. Newer models had markedly higher R values than older ones, largely because of richer data rather than model refinements. CONCLUSIONS: Combined drug and diagnostic data predicts total health care costs better than either type of data alone. Pharmacy spending is particularly predictable from drug data, whereas diagnoses are more useful than drugs for predicting other medical costs and total costs. Using even slightly more recent data can substantially boost model performance measures; thus, model comparisons should be conducted on the same dataset.

Actuarial Analysis↗
PubMed2003

Clinical application of artificial neural network (ANN) modeling to predict pharmacokinetic parameters of severely ill patients.

Artificial neural network (ANN) modeling was used to evaluate the pharmacokinetics of aminoglycosides (arbekacin sulfate and amikacin sulfate) in severely ill patients. The plasma level was predicted by ANN modeling using parameters related to the severity of the patient's condition and the predictive performance was shown to be better than could be achieved using multiple regression analysis. These results indicate that there is a non-linear relationship between the pharmacokinetics of aminoglycosides and the severity of the patient's condition, and this should be taken into account when determining the dose for severely ill patients. Patients whose plasma levels are likely to fall below the effective level can be identified by ANN modeling with a predictive sensitivity and specificity superior to multivariate logistic regression analysis. The predictable range should be inferred from the data structure before the modeling in order to improve the predictive performance. The volume of distribution (Vd) in the normal range was weakly predicted by ANN modeling from the patients' data. Prediction of clearance by ANN modeling was poorer than that obtained from serum creatinine concentration by linear regression analysis. These results suggest that the input-output relationship (linear or non-linear) should be taken into account in selecting the modeling method. Linear modeling can give better predictive performance for linear systems and non-linear modeling can give better predictive performance for non-linear systems. In general, the performance of ANN modeling was superior to linear modeling for PK/PD prediction. For accurate modeling, a predictable range should be inferred from the data structure before the analysis. Restriction of the predictable region, as determined from the data structure, produced an increase in prediction performance. When applying ANN modeling in clinical settings, the predictive performance and predictable region should be investigated in detail to avoid the risk of harm to severely ill patients.

Critical Illness↗
PubMed2004

A mechanistic model for predicting the nutrient requirements and feed biological values for sheep.

The Cornell Net Carbohydrate and Protein System (CNCPS), a mechanistic model that predicts nutrient requirements and biological values of feeds for cattle, was modified for use with sheep. Published equations were added for predicting the energy and protein requirements of sheep, with a special emphasis on dairy sheep, whose specific needs are not considered by most sheep-feeding systems. The CNCPS for cattle equations that are used to predict the supply of nutrients from each feed were modified to include new solid and liquid ruminal passage rates for sheep, and revised equations were inserted to predict metabolic fecal N. Equations were added to predict fluxes in body energy and protein reserves from BW and condition score. When evaluated with data from seven published studies (19 treatments), for which the CNCPS for sheep predicted positive ruminal N balance, the CNCPS for sheep predicted OM digestibility, which is used to predict feed ME values, with no mean bias (1.1 g/100 g of OM; P > 0.10) and a low root mean squared prediction error (RMSPE; 3.6 g/100 g of OM). Crude protein digestibility, which is used to predict N excretion, was evaluated with eight published studies (23 treatments). The model predicted CP digestibility with no mean bias (-1.9 g/100 g of CP; P > 0.10) but with a large RMSPE (7.2 g/100 g of CP). Evaluation with a data set of published studies in which the CNCPS for sheep predicted negative ruminal N balance indicated that the model tended to underpredict OM digestibility (mean bias of -3.3 g/100 g of OM, P > 0.10; RMSPE = 6.5 g/100 g of OM; n = 12) and to overpredict CP digestibility (mean bias of 2.7 g/100 g of CP, P > 0.10; RMSPE = 12.8 g/100 g of CP; n = 7). The ability of the CNCPS for sheep to predict gains and losses in shrunk BW was evaluated using data from six studies with adult sheep (13 treatments with lactating ewes and 16 with dry ewes). It accurately predicted variations in shrunk BW when diets had positive N balance (mean bias of 5.8 g/d; P > 0.10; RMSPE of 30.0 g/d; n = 15), whereas it markedly overpredicted the variations in shrunk BW when ruminal balance was negative (mean bias of 53.4 g/d, P < 0.05; RMSPE = 84.1 g/d; n = 14). These evaluations indicated that the Cornell Net Carbohydrate and Protein System for Sheep can be used to predict energy and protein requirements, feed biological values, and BW gains and losses in adult sheep.

Animal Feed↗
PubMed2003

Activity diary method for predicting energy expenditure as evaluated by a whole-body indirect human calorimeter.

In comparison with the energy expenditure determined by a whole-body indirect human calorimeter, which provides 24-h energy expenditure (TEE) with high precision and accuracy, the accuracy of predicting energy expenditure (EE) using an activity diary (AD) method was evaluated. Observed and predicted basal metabolic rate (BMR) as well as literature values for typical physical activities were used for TEE prediction. The effect of the number of recorded items in the activity diary on the accuracy of TEE was also examined. Additionally, predicted EE was divided into sleeping, exercise, and sedentary EE to evaluate the estimation errors in the AD method. Subjects were 20- to 69-y-old Japanese women (n = 20) and men (n= 21). Predicted TEE based on the AD was derived by applying the observed or predicted BMR to literature values for physical activities; i.e., relative metabolic rate (R.M.R.), physical activity ratio (PAR), and metabolic equivalent (MET). The BMR value observed for each subject was obtained by indirect calorimetry using a Douglas bag. The BMR for the subject was also estimated from the predictive equations in the 6th revision of the Recommended Dietary Allowances for the Japanese (1999). The correlations between observed and predicted TEE appeared stronger when using observed BMR than those using predicted BMR. Although the difference of mean values between the predicted and observed TEE was small, the limits of agreement between the predicted and observed TEE were around +/- 400 kcal. Predicted EE, excluding the time periods for exercise and rest laying down when determining BMR, showed similar results to those of TEE. Furthermore, the number of recorded items in the AD was not significantly correlated to the accuracy of the predicted TEE (r = -0.03). These findings indicate that the predicted TEE of the AD using observed or predicted BMR and literature values is favorably comparable to observed TEE using a whole-body human calorimeter on a group basis; however, its use as a proxy measure of TEE or EE on an individual basis may be limited.

Adult↗
PubMed2001

CAFASP2: the second critical assessment of fully automated structure prediction methods.

The results of the second Critical Assessment of Fully Automated Structure Prediction (CAFASP2) are presented. The goals of CAFASP are to (i) assess the performance of fully automatic web servers for structure prediction, by using the same blind prediction targets as those used at CASP4, (ii) inform the community of users about the capabilities of the servers, (iii) allow human groups participating in CASP to use and analyze the results of the servers while preparing their nonautomated predictions for CASP, and (iv) compare the performance of the automated servers to that of the human-expert groups of CASP. More than 30 servers from around the world participated in CAFASP2, covering all categories of structure prediction. The category with the largest participation was fold recognition, where 24 CAFASP servers filed predictions along with 103 other CASP human groups. The CAFASP evaluation indicated that it is difficult to establish an exact ranking of the servers because the number of prediction targets was relatively small and the differences among many servers were also small. However, roughly a group of five "best" fold recognition servers could be identified. The CASP evaluation identified the same group of top servers albeit with a slightly different relative order. Both evaluations ranked a semiautomated method named CAFASP-CONSENSUS, that filed predictions using the CAFASP results of the servers, above any of the individual servers. Although the predictions of the CAFASP servers were available to human CASP predictors before the CASP submission deadline, the CASP assessment identified only 11 human groups that performed better than the best server. Furthermore, about one fourth of the top 30 performing groups corresponded to automated servers. At least half of the top 11 groups corresponded to human groups that also had a server in CAFASP or to human groups that used the CAFASP results to prepare their predictions. In particular, the CAFASP-CONSENSUS group was ranked 7. This shows that the automated predictions of the servers can be very helpful to human predictors. We conclude that as servers continue to improve, they will become increasingly important in any prediction process, especially when dealing with genome-scale prediction tasks. We expect that in the near future, the performance difference between humans and machines will continue to narrow and that fully automated structure prediction will become an effective companion and complement to experimental structural genomics.

Automation↗
PubMed1986

Predicting outcome in coronary disease. Statistical models versus expert clinicians.

To study the accuracy with which long-term prognosis can be predicted in patients with coronary artery disease, prognostic predictions from a data-based multivariable statistical model were compared with predictions from senior clinical cardiologists. Test samples of 100 patients each were selected from a large series of medically treated patients with significant coronary disease. Using detailed case summaries, five senior cardiologists each predicted one- and three-year survival and infarct-free survival probabilities for 100 patients. Fifty patients appeared in multiple samples for assessing interphysician variability. Cox regression models, developed using patients not in the test samples, predicted corresponding outcome probabilities for each test patient. Overall, model predictions correlated better with actual patient outcomes than did the doctors' predictions. For three-year survival, rank correlations were 0.61 (model) and 0.49 (doctors). For three-year infarct-free survival predictions, correlations with outcome were 0.48 (model) and 0.29 (doctors). Comparisons by individual doctor revealed Cox model three-year survival predictions were better than those of four of five doctors (model predictions added significant [p less than 0.05] prognostic information to the doctor's predictions, whereas the converse was not true). For infarct-free survival, the Cox model was superior to all five doctors. Where predictions were made by multiple doctors, the interphysician variability was substantial. In coronary artery disease, statistical models developed from carefully collected data can provide prognostic predictions that are more accurate than predictions of experienced clinicians made from detailed case summaries.

Coronary Disease↗
PubMed2004

A neural network method for prediction of beta-turn types in proteins using evolutionary information.

MOTIVATION: The prediction of beta-turns is an important element of protein secondary structure prediction. Recently, a highly accurate neural network based method Betatpred2 has been developed for predicting beta-turns in proteins using position-specific scoring matrices (PSSM) generated by PSI-BLAST and secondary structure information predicted by PSIPRED. However, the major limitation of Betatpred2 is that it predicts only beta-turn and non-beta-turn residues and does not provide any information of different beta-turn types. Thus, there is a need to predict beta-turn types using an approach based on multiple sequence alignment, which will be useful in overall tertiary structure prediction. RESULTS: In the present work, a method has been developed for the prediction of beta-turn types I, II, IV and VIII. For each turn type, two consecutive feed-forward back-propagation networks with a single hidden layer have been used where the first sequence-to-structure network has been trained on single sequences as well as on PSI-BLAST PSSM. The output from the first network along with PSIPRED predicted secondary structure has been used as input for the second-level structure-to-structure network. The networks have been trained and tested on a non-homologous dataset of 426 proteins chains by 7-fold cross-validation. It has been observed that the prediction performance for each turn type is improved significantly by using multiple sequence alignment. The performance has been further improved by using a second level structure-to-structure network and PSIPRED predicted secondary structure information. It has been observed that Type I and II beta-turns have better prediction performance than Type IV and VIII beta-turns. The final network yields an overall accuracy of 74.5, 93.5, 67.9 and 96.5% with MCC values of 0.29, 0.29, 0.23 and 0.02 for Type I, II, IV and VIII beta-turns, respectively, and is better than random prediction. AVAILABILITY: A web server for prediction of beta-turn types I, II, IV and VIII based on above approach is available at http://www.imtech.res.in/raghava/betaturns/ and http://bioinformatics.uams.edu/mirror/betaturns/ (mirror site).

Algorithms↗