Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “predictive modelling”

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 19 recordsLinked to original sources
PubMed2005

In silico ADME modelling: prediction models for blood-brain barrier permeation using a systematic variable selection method.

Quantitative Structure-Property Relationship models (QSPR) based on in vivo blood-brain permeation data (logBB) of 88 diverse compounds, 324 descriptors and a systematic variable selection method, namely 'Variable Selection and Modeling method based on the prediction (VSMP)', are reported. Of all the models developed using VSMP, the best three-descriptors model is based on Atomic type E-state index (SsssN), AlogP98 and Van der Waal's surface area (r=0.8425, q=0.8239, F=68.49 and SE=0.4165); the best four-descriptors model is based on Kappa shape index of order 1, Atomic type E-state index (SsssN), Atomic level based AI topological descriptor (AIssssC) and AlogP98 (r=0.8638, q=0.8472, F=60.982 and SE=0.3919). The performance of the models on three test sets taken from the literature is illustrated and compared with the results from other reported computational approaches. Test set III constitutes 91 compounds from the literature with known qualitative BBB indication and is used for virtual screening studies. The success rate of the reported models is 82% in the case of BBB+ compounds and a similar success rate is observed with BBB- compounds. Finally, as the models reported herein are based on computed properties, they appear as a valuable tool in virtual screening, where selection and prioritization of candidates is required.

Blood-Brain Barrier↗
PubMed2001

[Roaming through methodology. XXXIV. Limitations of predictive models].

Predictive models can be readily used within the populations in which they were developed but they often give a less than satisfactory performance when applied to another population. The main cause of this poorer performance is the fact that predictive models are often developed and evaluated within one single population. In a lot of cases, the model includes too many predictors (or a small population is used), which increases the chance of overfitting. Both external and internal validation techniques have been developed to evaluate predictive models. The most stringent test is an external validation: the application of the model to a new population. The usability of a predictive model can be evaluated using the receiver operating characteristic (ROC) curve.

Data Interpretation, Statistical↗
PubMed2000

Evidence-based growth hormone therapy prediction models.

Prediction models describing the response of various pathophysiological states to intervention can be of value in confirming a diagnosis, determining the prognosis and promoting compliance with treatment. The rigorous evaluation process of evidence-based medicine, used to assess any diagnostic test or therapeutic intervention, should be applied to studies reporting the development, validation and application of these prediction models. The models can provide only an estimate of the average effect to be expected, so the failure of an individual to exhibit an 'average' response does not necessarily imply a problem with that patient. Further development of the models is required to overcome inherent statistical problems and to allow greater applicability to the individual patient.

Body Height↗
PubMed2000

Predictive modeling of mixed microbial populations in food products: evaluation of two-species models.

Predictive microbiology is an emerging research domain in which biological and mathematical knowledge is combined to develop models for the prediction of microbial proliferation in foods. To provide accurate predictions, models must incorporate essential factors controlling microbial growth. Current models often take into account environmental conditions such as temperature, pH and water activity. One factor which has not been included in many models is the influence of a background microflora, which brings along microbial interactions. The present research explores the potential of autonomous continuous-time/two-species models to describe mixed population growth in foods. A set of four basic requirements, which a model should satisfy to be of use for this particular application, is specified. Further, a number of models originating from research fields outside predictive microbiology, but all dealing with interacting species, are evaluated with respect to the formulated model requirements by means of both graphical and analytical techniques. The analysis reveals that of the investigated models, the classical Lotka-Volterra model for two species in competition and several extensions of this model fulfill three of the four requirements. However, none of the models is in agreement with all requirements. Moreover, from the analytical approach, it is clear that the development of a model satisfying all requirements, within a framework of two autonomous differential equations, is not straightforward. Therefore, a novel prototype model structure, extending the Lotka-Volterra model with two differential equations describing two additional state variables, is proposed to describe mixed microbial populations in foods.

Evaluation Studies as Topic↗
PubMed1987

A predictive model for wound sepsis in oncologic surgery of the head and neck.

A prospective analysis of patients undergoing surgical resection of squamous cell carcinoma of the upper aerodigestive tract was performed in order to identify the patients at risk of postoperative wound infection and to develop a model predictive of wound infection. Fifty-nine patients who underwent extirpative clean-contaminated procedures--all of whom received cefazolin as the sole chemoprophylactic agent, were studied over a 1-year period. Twenty-three variables were recorded for each patient in the study. The overall rate of wound infection was 25.4%. Univariate analysis indicated that three variables were significantly related to the likelihood of postoperative wound infection. These included tumor stage (P = 0.0180), nodal stage (P = 0.0062), and duration of surgery (P = 0.0151). The Biomedical Computer Program (BMDP), a logistic regression program specifically designed for a binary dependent variable (infection vs. no infection) based on independent variables that may be continuous or categorical, was used in development of a model predictive of wound infection. T-stage, N-stage, and the presence of concomitant disease made up the combination of factors found to be most predictive of infection in our study population. Considering "success" to be the development of infection if the probability was 75% or higher, and the absence of infection if the probability was less than 25%, the multiple regression analysis model demonstrated a predictive success rate of 74.6%. Our results indicate that the risk of infection in patients undergoing clean-contaminated oncologic surgery of the head and neck is greatest for patients who have advanced disease that requires prolonged surgery in the presence of concomitant diseases.(ABSTRACT TRUNCATED AT 250 WORDS)

Adult↗
PubMed2000

Prediction models for insulin resistance in the polycystic ovary syndrome.

Women with the polycystic ovary syndrome (PCOS) have a high prevalence of insulin resistance, with consequent increased risk of metabolic diseases later in life. An early metabolic screening would therefore be of clinical relevance. By using stepwise regression analysis on several variables obtained in 72 women with PCOS, we constructed simple and reliable mathematical models predicting insulin sensitivity, as measured by the euglycaemic hyperinsulinaemic clamp. The normal ranges of insulin sensitivity were calculated from 81 non-hirsute, normally menstruating women with normal ovaries, and similar body mass index (BMI) and age as the women with PCOS. Measured variables included BMI, waist and hip circumferences, truncal-abdominal skin folds, circulating concentrations of gonadotrophins, androgens, sex hormone-binding globulin (SHBG), triglycerides, total cholesterol and cholesterol subfractions, fasting insulin, C-peptide and free fatty acids. The three best prediction models included waist circumference, together with insulin (model I: R(2) = 0.77), serum triglycerides (model II: R(2) = 0.65), and the subscapularis skin fold (model III: R(2) = 0. 64). Using reference limits for insulin sensitivity obtained in the 81 normal pre-menopausal women, the models identify insulin resistant women with PCOS. These simple and inexpensive models are potentially useful in clinical practice as an early screening in women with PCOS.

Adolescent↗
PubMed2007

Predictive modeling of Bacillus cereus spores in farm tank milk during grazing and housing periods.

The shelf life of pasteurized dairy products depends partly on the concentration of Bacillus cereus spores in raw milk. Based on a translation of contamination pathways into chains of unit-operations, 2 simulation models were developed to quantitatively identify factors that have the greatest effect on the spore concentration in milk. In addition, the models can be used to determine the reduction in concentration that could be achieved via measures at the farm level. One model predicts the concentration when soil is the source of spores, most relevant during grazing of cows. The other model predicts the concentration when feed is the main source of spores, most relevant during housing of cows. It was estimated that when teats are contaminated with soil, 33% of the farm tank milk (FTM) contains more than 3 log(10) spores/L of milk. When feed is the main source, this is only 2%. Based on the predicted spore concentrations in FTM, we calculated that the average spore concentration in raw milk stored at the dairy processor during the grazing period is 3.5 log(10) spores/L of milk and during the housing period is 2.1 log(10) spores/L. It was estimated that during the grazing period a 99% reduction could be achieved if all farms minimize the soil contamination of teats and teat cleaning is optimized. During housing, reduction of the concentration by 60% should be feasible by ensuring spore concentrations in feed below 3 log(10) spores/g and a pH of the ration offered to the cows below 5. Implementation of these measures at the farm level ensures that the concentration of B. cereus spores in raw milk never exceeds 3 log(10) spores/L.

Animals↗
PubMed2004

The differences between claim-based health risk adjustment models and cost prediction models.

There has been a significant increase in interest in using risk assessment tools with administrative claims data for provider profiling, provider payment, underwriting and disease/case management. The tools can be classified into two types: risk adjustment models and cost prediction models. The differences between the two models have not been well recognized. This paper explains the differences in terms of the objectives, the applications, and the accuracy of evaluations.

Insurance Claim Review↗
PubMed2002

Grouped-neural network modeling for model predictive control.

A group of feed-forward neural networks (NNs), each providing the prediction of an individual process output at a future step, is used as the dynamic prediction model for the model-based predictive control (MPC) scheme in the proposed work. These NNs are parallel (independent) rather than cascaded--they are trained and implemented in parallel. Therefore, the complexity and effort in the training stage is decreased and compounded error propagation is eliminated from the prediction. A new strategy of compensating for the process-model mismatch under this grouped-NN model structure is also developed. Effectiveness of the scheme as a general nonlinear MPC is demonstrated by simulation results.

Algorithms↗
PubMedDate not supplied

Predictive models for aquatic toxicity of aldehydes designed for various model chemistries.

Predictive models for the aquatic toxicity of aldehydes were designed for a set of 50 aromatic or aliphatic compounds containing at least one aldehyde group, for which the acute toxicity data for the fathead minnow (Pimephales promelas) are available (96 h test assessing 50% lethal waterborne concentration). The molecular descriptors were based on calculations with various semiempirical or ab initio model chemistries. The resulting four-parameter models were evaluated according to the correlation coefficients R(2). The best predictive model was obtained with the HF/STO-3G model chemistry (R(2) = 0.868), while the models designed for descriptors based on ab initio calculations of higher level showed a slightly worse predictivity (the HF/3-21G(d) based model R(2) = 0.800, the HF/6-31G(d) based model R(2) = 0.808, the B3LYP/6-31G(d,p) based model R(2) = 0.812). With the semiempirical methods a good predictivity was observed with the PM3 based model (R(2) = 0.811) and the AM1 based model (R(2) = 0.791), but the MNDO based model showed the worst predictivity (R(2) = 0.760). In all ab initio models and the PM3 model very similar descriptors were involved. The importance of the descriptor logarithm of the partition coefficient logP for toxicity prediction was confirmed. Additionally, the descriptors encoding the negatively charged molecular surface area, hydrogen bonding molecular surface area, and reactivity of aldehyde group were identified as essential for the toxicity prediction of aldehydes.

Aldehydes↗
PubMed1996

Predictive models of the effect of temperature, pH and acetic and lactic acids on the growth of Listeria monocytogenes.

The combined effect of temperature (1-20 degrees C), pH (4.5-7.2) and acetic acid (0-10,000 mg/l; model 1) or lactic acid (0-20,000 mg/l; model 2) on growth of Listeria monocytogenes in laboratory media was studied. Growth curves at various combinations of temperature, pH and acid concentration were fitted by the model of Baranyi and Roberts (1994), and specific growth rates derived from the curve fit were modelled. Predictions of growth from the models were compared with data in the literature, and this showed the models to be suitable for use in predicting growth of L. monocytogenes in a range of foods including meat, poultry, fish, egg and milk and dairy products. The two models are compatible, i.e. they give similar predictions for cases when no acid is present.

Acetic Acid↗
PubMed2004

Development of predictive models for the survival of Campylobacter jejuni (ATCC 43051) on cooked chicken breast patties and in broth as a function of temperature.

The objective of this study was to model the kinetics of the survival of Campylobacter jejuni on cooked chicken breast patties and in broth as a function of temperature. Both patties and broth were inoculated with 10(6) stationary-phase cells of a single strain of C. jejuni (ATCC 43051) and incubated at constant temperatures from 4 to 30 degrees C in 2 degrees C increments under aerobic conditions. In most cases, a three-phase linear model fit the primary survival curves well (r2 = 0.97 to 0.99) at all incubation temperatures regardless of model medium, indicating the presence of a resistant subpopulation of C. jejuni that would not be eliminated without thermal processing. Secondary models predicting lag time (LT) and specific death rate (SDR) as functions of temperature were also developed. The Davey and Boltzmann models were identified as appropriate secondary models for LT and SDR, respectively, on the basis of goodness of fit (Boltzmann model, r2 = 0.96; Davey model, r2 = 0.93) and prediction bias and accuracy factor tests. The results obtained indicate that C. jejuni can survive well at both refrigeration and ambient temperatures regardless of model medium. Reduced survival of C. jejuni, characterized by shorter lag times and faster death rates, was observed both on patties and in broth at ambient temperatures. In addition, the average maximum reduction of C. jejuni at 4 to 30 degrees C was 1.5 log units regardless of storage temperature or model medium. These findings suggest that C. jejuni found on contaminated poultry products has the potential to survive under conditions that are not permissive for growth and thus could cause foodborne illness if the poultry is not sufficiently cooked.

Animals↗
PubMed1997

Predictive model of the effect of CO2, pH, temperature and NaCl on the growth of Listeria monocytogenes.

The growth responses of L. monocytogenes as affected by CO2 concentration (0-100% v/v, balance nitrogen), NaCl concentration (0.5-8.0% w/v), pH (4.5-7.0) and temperature (4-20 degrees C) were studied in laboratory medium. Growth curves were fitted using the model of Baranyi and Roberts, and specific growth rates derived from the curve fit were modelled. Predictions for specific growth rate, doubling time and time to a 1000-fold increase could be made for any combination of conditions within the matrix. Predictions of growth from the model were compared with published data and this showed the model to be suitable for predicting growth of L. monocytogenes in a range of foods packaged under a modified atmosphere.

Carbon Dioxide↗
PubMed1997

Model Predictive Impedance Control: A Model for Joint Movement.

Impedance control has been suggested as the strategy employed by the central nervous system to control human postures and movements. A realization of this strategy is presented that uses a model predictive control algorithm as a higher motor controller. External disturbances are explicitly included in the model. The combination of 3 key factors-joint impedance control, model predictive controller, and external disturbance input-forms the basis for the generality of this model. The model was applied to 3 different types of joint movements: a tracking movement with an unpredicted disturbance, a rhythmic movement, and an unstable biped model of human walking. Computer simulation results showed excellent performance of the model in all 3 cases for optimal values of active joint impedances and an exact match between the musculoskeletal system and the model internal to the model predictive controller. The controller was also able to maintain acceptable performance in the presence of a 25% mismatch between the musculoskeletal system and its internal model.

equilibrium-point hypothesis↗
PubMed2002

Bayesian analysis and inference from QSAR predictive model results.

QSAR models have been under development for decades but acceptance and utilization of model results have been slow, in part, because there is no widely accepted metric for assessing their reliability. We reapply a method commonly used in quantitative epidemiology and medical decision-making for evaluating the results of screening tests to assess reliability of a QSAR model. It quantifies the accuracy (expressed as sensitivity and specificity) of QSAR models as conditional probabilities of correct and incorrect classification of chemical characteristic, given a true characteristic. Using Bayes formula, these conditional probabilities are combined with prior information to generate a posterior distribution to determine the probability a specific chemical has a particular characteristic, given a model prediction. As an example, we apply this approach to evaluate the predictive reliability of a CATABOL model and base on it a "ready" and "not ready" biodegradability classification. Finally, we show how predictive capability of the model can be improved by sequential use of two models, the first one with high sensitivity and the second with high specificity.

Bayes Theorem↗
PubMed1996

Predictive modeling of the growth of Listeria monocytogenes in CO2 environments.

The effects of pH (5.5, 6.5), temperature (4, 7 and 10 degrees C) and carbon dioxide (10, 30, 50, 70 and 90%) on the growth and/or survival of a five strain mixture of Listeria monocytogenes were examined in brain heart infusion broth. All three variables had a major influence on the growth characteristics of the organism. As expected, both the lag time and generation time increased as the CO2 level increased, and as pH and temperature decreased. Growth over a 30-day period was observed at all parameter combinations tested, except at pH 5.5, 4 degrees C in the presence of either 50, 70 or 90% carbon dioxide. Two primary models, the Gompertz and Baranyi equations, were compared for their ability to describe the growth of L. monocytogenes. In general, the Gompertz model predicted both longer lag and shorter generation times, compared to the Baranyi model. The Baranyi model appeared to fit the overall data better than the Gompertz model. However, these differences were often small. Response surface models were developed for predicting the effects and interactions of the three independent variables on the growth and/or survival of L. monocytogenes in the different modified atmospheres. Results demonstrate the importance of strict temperature control for maintaining the advantages of food shelf life extension in enriched carbon dioxide environments. The information obtained in this study could be used as a guide to manufacturers of modified-atmosphere packaged foods, especially when designing products in which this organism may be a concern.

Carbon Dioxide↗
PubMed2005

A theoretical, practical, predictive model of faculty and department research productivity.

PURPOSE: Although numerous characteristics impact faculty research productivity, and although researchers have suggested comprehensive theoretical models to explain the relationship between these characteristics and levels of faculty research productivity, few studies have assessed these models. This study tests the ability of the Bland et al. (2002) model-based on individual, institutional, and leadership variables influencing faculty research productivity-to explain individual and group (department) research productivity within the context of a large medical school. METHOD: This study used data from a University of Minnesota Medical School-Twin Cities vitality survey conducted in 2000 that had a response rate of 76% (n = 465 faculty). A statistical software package was used to conduct t tests, logistic regressions, and multiple regressions on these data. RESULTS: The validity of faculty, department, and leadership characteristics identified in the Bland et al. (2002) model were confirmed as necessary for high levels of research productivity. Faculty productivity was influenced more by individual and institutional characteristics; group productivity was more affected by institutional and leadership characteristics. CONCLUSION: The characteristics and groupings (individual, institutional, and leadership) in the Bland et al. (2002) model predict faculty research productivity. Research productivity is influenced by the interaction of the three broad groupings, and it is the dynamic interplay of individual and institutional characteristics, supplemented with effective leadership, that determines the productivity of individuals and departments.

Adult↗