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 1,009 records · Page 56Linked to original sources
PubMed2002

Prediction of renal insufficiency in Pima Indians with nephropathy of type 2 diabetes mellitus.

BACKGROUND: A high prevalence and early onset of type 2 diabetes in Pima Indians is well known. Our objective is to use several statistical models to identify predictors of glomerular filtration rate (GFR) deterioration and develop an algorithm to predict GFR 4 years after the initial evaluation. METHODS: All records (n = 86) were randomly assigned to a training set (n = 60) and a testing set (n = 26). Linear regression, generalized additive, tree-based, and artificial neural network models were used to identify predictors of outcome and develop a prediction algorithm. RESULTS: Proteinuria remained the single most important predictor of long-term renal function; other predictors included baseline GFR, blood pressure, plasma renin activity, lipid profile, age, weight/body mass index, and diabetes duration. All four models achieved a good correlation (r = 0.73 to 0.78) between observed and predicted 4-year GFRs on a separate (testing) data set. Best results in predicting the value of GFR were achieved using a tree-based model with six terminal nodes (r = 0.78; root mean squared prediction error = 38.9). The tree-based and generalized additive models achieved high positive (91%) and negative (100%) predictive values in identifying subjects, who developed depressed GFRs in 4 years. An artificial neural network achieved the highest area under the receiver operating characteristic curve (0.91). CONCLUSION: GFR depression within 4 years can be predicted with a precision that suggests potential clinical utility. A tree-based model with six terminal nodes has shown the best results in predicting the actual value of GFR, whereas an artificial neural network is the model of choice to identify the group of patients that will develop renal insufficiency.

Adult↗
PubMed2002

Derived equations are not precise enough to predict the adequacy of creatinine clearance in peritoneal dialysis patients.

BACKGROUND: Most renal units assess dialysis adequacy in peritoneal dialysis (PD) patients by formal 24-hour collections of urine and effluent dialysate. We sought a reliable method of predicting dialysis adequacy that allows a decrease in the frequency of these formal and cumbersome measurements. METHODS: We created a formula for estimating total creatinine clearances, then assessed the clinical utility of this formula and other published formulae in predicting adequate and inadequate dialysis in PD patients. We collected data over a 6-month period in 2001 from 288 PD patients from 9 centers in Scotland. Four out of every 5 patients were selected at random to create a formula for estimating total creatinine clearance per week, and the fifth patient was used to form a validation group. We plotted creatinine excretion against age, and the resultant linear regression equation was transformed to produce a formula for predicting total creatinine clearance per week, based on patient sex, weight, and serum creatinine. We used the data from the validation subgroup to calculate predictive values for our derived formula and data from all of the patients to calculate predictive values for the Cockcroft and Gault, Jones, and Modification of Diet in Renal Disease Study formulae. RESULTS: Neither our derived formula nor the three published formulae were sufficiently powerful to predict accurately either adequate or inadequate PD clearance. Receiver operator characteristic curves showed that no significant improvement in these predictive values could be achieved by altering either the sensitivity or the specificity. CONCLUSION: Prediction formulae are not accurate enough to detect underdialysis in PD patients.

Adolescent↗
PubMed2002

The relative predictive ability of four different measures of hemodialysis dose.

BACKGROUND: The amount of hemodialysis that patients receive is an independent predictor of mortality. However, the relative predictive ability of four common measures of dialysis dose (urea reduction ratio, single-pool Kt/V, double-pool Kt/V, and urea index) is unclear. METHODS: Using The Renal Network Data System, we identified 14,810 incident hemodialysis patients in Indiana, Kentucky, Ohio, and Illinois from 1997 to 2000. We calculated each measure of hemodialysis dose during the first 6 months of treatment, then prospectively followed up patients for an additional 6 months. For each measure of dialysis dose, we developed a logistic regression model to examine the relationship between dose and patient mortality after adjustment for age, race, sex, cause of renal failure, comorbid conditions, and albumin level. We compared the predictive ability of the four models using the c statistic, a measure of how frequently survivors have a lower predicted probability of death compared with nonsurvivors. C statistics can vary from 0.50 (no predictive ability) to 1.00 (perfect predictive ability). RESULTS: Of all patients, 11.3% died during follow-up. Mortality was independently associated with low dialysis dose, advanced age, white race, female sex, specific comorbid conditions, and low albumin level. All four predictive models had virtually identical c statistics (range, 0.69 to 0.70). CONCLUSION: Models including hemodialysis dose and patient characteristics have a modest ability to predict mortality. Moreover, all four measures of dialysis dose have an equivalent predictive ability. Decisions to use a specific measure should be based on other considerations, such as ease of use, need to troubleshoot inadequate dialysis delivery, or research on urea kinetics or nutritional factors.

Adolescent↗
PubMed2001

How accurate are physicians' clinical predictions of survival and the available prognostic tools in estimating survival times in terminally ill cancer patients? A systematic review.

The purpose of this review was to examine the accuracy of physicians' clinical predictions of survival and the available prognostic tools in estimating survival times in terminally ill cancer patients. A MEDLINE search for English language articles published between 1966 and March 2000 was performed using the following keywords: forecasting/clinical prediction, prognosis/prognostic factors, survival and neoplasm metastasis. Searches in CancerLit, EMBASE, PubMed, the Cochrane Library and reference sections of articles were performed. Studies were included if they concerned adult patients with various cancer histological diagnoses and employed clinical prediction and the readily available clinical parameters. Biochemical and molecular markers were excluded. Grading of the evidence and recommendations was performed. Twelve articles on clinical prediction and 19 on prognostic factors met the inclusion criteria. Clinical prediction tends to be incorrect in the optimistic direction but improves with repeated measurements. Performance status has been found to be most strongly correlated with the duration of survival, followed by the 'terminal syndrome', which includes anorexia, weight loss and dysphagia. Cognitive failure and confusion have also been associated with a shorter life span. Performance status combined with clinical symptoms and the clinician's estimate helps to guide an accurate prediction, as reviewed in an Italian series. There is fair evidence to support using performance status, and clinical and biochemical parameters, in addition to clinicians' judgement to aid survival prediction. However, there is weak evidence to support that clinicians' estimates alone could be specifically employed for survival prediction.

Cognition Disorders↗
PubMed2003

A model to predict survival at one month, one year, and five years after liver transplantation based on pretransplant clinical characteristics.

Reliable models that could predict outcome of liver transplantation (LT) may guide physicians to advise their patients of immediate and late survival chances and may help them to optimize organ use. The objective of this study was to develop user-friendly models to predict short and long-term mortality after LT in adults based on pre-LT recipient characteristics. The United Network for Organ Sharing (UNOS) transplant registry (n = 38,876) from 1987 to 2001 was used to develop and validate the model. Two thirds of patients were randomized to develop the model (the modeling group), and the remaining third was randomized to cross-validate (the cross-validation group) it. Three separate models, using multivariate logistic regression analysis, were created and validated to predict survival at 1 month, 1 year, and 5 years. Using the total severity scores of patients in the modeling group, a predictive model then was created, and the predicted probability of death as a function of total score then was compared in the cross-validation group. The independent variables that were found to be very significant for 1 month and 1 year survival were age, body mass index (BMI), UNOS status 1, etiology, serum bilirubin (for 1 month and 1 year only), creatinine, and race (only for 5 years). The actual deaths in the cross-validation group followed very closely the predicted survival graph. The chi-squared goodness-of-fit test confirmed that the model could predict mortality reliably at 1 month, 1 year, and 5 years. We have developed and validated user-friendly models that could reliably predict short-term and long-term survival after LT.

Adult↗
PubMed2002

The survivin:Fas ratio is predictive of recurrent disease in neuroblastoma.

BACKGROUND/PURPOSE: Several clinical and biologic features of neuroblastoma (NB) are used to predict the risk of recurrent disease. The balance between antiapoptotic and proapoptotic factors within a tumor may affect its ability to survive. Survivin is an antiapoptotic factor expressed in highly proliferative NB, whereas Fas is a proapoptotic factor that portends a favorable prognosis. The authors determined whether the ratio of survivin to Fas (S:F ratio) is predictive of recurrent disease in patients with NB. The authors previously have shown the S:F ratio is predictive of recurrent disease in pediatric renal tumors. METHODS: The authors quantified the levels of 9 different apoptotic mRNA species using Rnase Protection assay (RPA, Riboquant, PharMingen, San Diego, CA). Twenty-eight primary tumor specimens were evaluated from patients with ganglioneuroma (n = 3), ganglioneuroblastoma (n = 2), and neuroblastoma (n = 23) from tumors of all clinical stages obtained at the time of diagnosis. mRNA levels were calculated as a percentage of L32 for each specimen assayed, and positive expression was assumed to be greater than 10% of L32. RESULTS: Survivin was expressed in 90% of tumors that went on to recur and only in 27.7% of those that were cured. The S:F ratio was significantly greater in tumors that went on to recur (n = 10) compared with those from patients that were cured (n = 18) (median S:F ratio, 3.3 v 0.75; P =.0002, Wilcoxon rank-sum test). A cutoff ratio of 2.3 was highly predictive of tumor recurrence irrespective of clinical stage of disease (area under ROC curve = 0.906). Sensitivity was 80% (CI, 44.4% to 97.5%), specificity was 94.4% (CI, 72.7% to 99.9%), positive predictive value was 88.9% (CI, 51.8% to 99.7%), and negative predictive value was 89.5% (66.9% to 98.7%). Twenty-five of 28 (89.3%) tumor ratios were correct in predicting outcome. CONCLUSIONS: The survivin:Fas ratio in primary tumors may be used to predict the risk for recurrent disease in patients with NB. The S:F ratio appears to be a more sensitive predictor of recurrent disease than survivin expression alone. Determining this ratio may not only be helpful in guiding follow-up of patients with NB, but also may aid in stratifying patients for more aggressive therapeutic strategies.

Child↗
PubMed2000

Use of cervical ultrasonography in prediction of spontaneous preterm birth in triplet gestations.

OBJECTIVE: The aim of this study was to assess the role of cervical ultrasonography in the prediction of spontaneous preterm birth in triplet gestations and to compare various ultrasonographic cervical parameters with respect to predictive ability. STUDY DESIGN: This prospective cohort study included 51 triplet gestations longitudinally evaluated between 15 and 28 weeks' gestation on 274 occasions with transvaginal cervical ultrasonography and transfundal pressure. The cervical parameters obtained were funnel width and length, cervical length, percentage of funneling, and cervical index. RESULTS: Receiver operating characteristic curve analyses showed that cervical lengths of < or =2.5 cm and < or =2.0 cm between 15 and 24 weeks' gestation and between 25 and 28 weeks' gestation, respectively, were at least as good as other ultrasonographic cervical parameters for the prediction of spontaneous preterm birth. A cervical length of < or =2.5 cm between 15 and 20 weeks' gestation had both a specificity and a positive predictive value of 100% for delivery at <28 weeks' gestation, and the sensitivities and negative predictive values ranged from 25% to 50% and from 72% to 91%, respectively, for deliveries at <28, <30, and <32 weeks' gestation. A cervical length of < or =2.5 cm between 21 and 24 weeks' gestation had an 86% sensitivity for prediction of spontaneous delivery at <28 weeks' gestation. A cervical length of < or =2.0 cm between 25 and 28 weeks' gestation had both a sensitivity and a negative predictive value of 100% for delivery at both <28 and <30 weeks' gestation. CONCLUSIONS: In triplet gestations cervical lengths of < or =2.5 cm between 15 and 24 weeks' gestation and < or =2.0 cm between 25 and 28 weeks' gestation were at least as good as other ultrasonographic cervical parameters for the prediction of spontaneous preterm birth.

Cervix Uteri↗
PubMed2000

Accuracy of regression equation prediction across the range of estimated premorbid IQ.

Linear regression is often used to predict psychological criterion variables such as premorbid IQ. The prevailing method of evaluating the accuracy of prediction indicates poor accuracy for both high and low criterion values. These results have led to the conclusion that the equations are not applicable for predicting scores beyond about one standard deviation from the mean. The apparent inaccuracy at the extremes, however, is an artifact of inappropriate analysis. An alternative analysis method is described and used to re-analyze two sets of data. Empirical results show that both high and low WAIS-R IQ scores, predicted using versions of the NART, agree with actual measured IQ scores as accurately as do predicted scores near the mean. In addition, confidence intervals were only 5% larger for more extreme predicted values than for values near the mean, which would be of minor clinical consequence. The practical constraint on prediction of extreme values arises not from the regression technique, but from the limited range of the predictor variable(s). The two reviewed IQ prediction equations would, however, have adequate range for a high percentage of individuals.

Humans↗
PubMed2000

Avoidance of dental visits: the predictive validity of three dental anxiety scales.

The purpose of this study was to determine the sensitivity, specificity, positive and negative predictive values for Corah's Dental Anxiety Scale (DAS) and two modified versions of it (MDAS; MDAS/4). A questionnaire was mailed to a simple random sample of 1,190 25-year-old residents in the west of Norway in 1997. Half the sample received DAS, the other half MDAS. The response rate after one reminder was 62%. The respondents completed the scales, gave demographic particulars and answered one question about dental visiting habits during the last 5 years plus an open-ended question about reasons for non-attendance. Using the answers to the latter question as validating criterion, it was found that, for all scales, sensitivity decreased while specificity improved when changing from a liberal to a stringent cut-off point. The scales gave low positive predictive values (< or = 0.26), but high negative predictive values (> or = 0.98). Since DAS and MDAS/4 gave almost identical findings, the two samples were combined. At a cut-off point > or = 13 sensitivity was 0.83, specificity 0.84, positive predictive value 0.18 and negative predictive value 0.99. The corresponding estimates when the cut-off point was > or = 15 were 0.67, 0.90, 0.22 and 0.98. It is concluded that, in this test, DAS and the two versions of MDAS gave acceptable, or near acceptable sensitivity, specificity and negative predictive values, but far too low positive predictive values to be useful for prediction at the individual level.

Adult↗
PubMed1997

Outcome after severe head injury: an analysis of prediction based upon comparison of neural network versus logistic regression analysis.

More reliable prediction of outcome would be helpful for clinicians who treat severely head-injured patients. To determine if neural network modeling would improve outcome prediction compared with standard logistic regression analysis and to determine if data available 24 h after severe head injury allows better prediction than data obtained within 6 h, we tested the ability of both techniques at these two times to predict outcome (dead versus alive) at 6 months. One thousand sixty-six consecutive patients with Glasgow Coma Scale scores of 8 or less during the first 24 h after injury were randomly divided into two groups. Data from the first group (n = 799) were used to develop the models; data from the second group (n = 267) were used to test the accuracy, sensitivity, and specificity of the models by comparing predicted and actual outcomes. The 6-month mortality rate was 63.5%. Our findings confirm the importance of age, Glasgow Coma Scale scores, and hypotension in predicting outcome. Using data available at 24 h improved the predictive power of both models compared with admission data; at both time points, however, the differences in the results obtained with the two models were negligible. We conclude that outcome (dead versus alive) at 6 months after severe head injury can be predicted with logistic regression or neural network models based on data available at 24 h. Critical therapeutic decisions, such as cessation of therapy, should be based on the patient's status 1 day after injury and only rarely on admission status alone.

Adolescent↗
PubMed2000

Prediction of fat and fat-free mass in male athletes using dual X-ray absorptiometry as the reference method.

The ability of bioelectrical impedance analysis and anthropometry to predict fat mass and fat-free mass was compared in a sample of 82 male athletes from a wide variety of sports, using dual-energy X-ray absorptiometry (DXA) as the reference method. The percent fat measured by DXA was 10.9+/-4.9% (mean +/- s), and fat mass was predicted with a standard error of the estimate of 1.7 kg for skinfolds and 2.8 kg for bioelectrical impedance analysis (P < 0.001). Fat-free mass was predicted with a standard error of the estimate of 1.7 kg for anthropometry and 2.6 kg for bioelectrical impedance analysis (P < 0.001). Regression of various individual skinfolds and summed skinfolds, to examine the effect of skinfold selection combinations by stepwise regression, produced an optimal fat mass prediction using the thigh and abdominal skinfold sites, and an optimal fat-free mass prediction using the thigh, abdominal and supra-ilium sites. These results suggest that anthropometry offers a better way of assessing body composition in athletes than bioelectrical impedance analysis. Applying the derived equations to a separate sample of 24 athletes predicted fat and fat-free mass with a total error of 2.3 kg (2.9%) and 2.2 kg (2.7%), respectively. Combining the samples introduced more heterogeneity into the sample (n = 106), and the optimal prediction of fat mass used six skinfolds in producing a similar standard error of the estimate (1.7 kg), although this explained a further 4% of the variation in DXA-derived fat. Fat-free mass was predicted best from four skinfolds, although the standard error of the estimate and coefficient of determination were unchanged.

Absorptiometry, Photon↗
PubMed2006

Influence of seat geometry and seating posture on NIC(max) long-term AIS 1 neck injury predictability.

OBJECTIVE: Validated injury criteria are essential when developing restraints for AIS 1 neck injuries, which should protect occupants in a variety of crash situations. Such criteria have been proposed and attempts have been made to validate or disprove these. However, no criterion has yet been fully validated. The objective of this study is to evaluate the influence of seat geometry and seating posture on the NIC(max) long-term AIS 1 neck injury predictability by making parameter analyses on reconstructed real-life rear-end crashes with known injury outcomes. METHODS: Mathematical models of the BioRID II and three car seats were used to reconstruct 79 rear-end crashes involving 110 occupants with known injury outcomes. Correlations between the NIC(max) values and the duration of AIS 1 neck injuries were evaluated for variations in seat geometry and seating posture. Sensitivities, specificities, positive predictive values, and negative predictive values were also calculated to evaluate the NIC(max) predictability. RESULTS: Correlations between the NIC(max) values and the duration of AIS 1 neck injuries were found and these relations were used to establish injury risk curves for variations in seat geometry and seating posture. Sensitivities, specificities, positive predictive values, and negative predictive values showed that the NIC(max) predicts long-term AIS 1 neck injuries also for variations in seat geometry and seating postures. CONCLUSION: The NIC(max) can be used to predict long-term AIS 1 neck injuries.

Accidents, Traffic↗
PubMed2001

Genetic adaptive neural network to predict biochemical failure after radical prostatectomy: a multi-institutional study.

BACKGROUND AND PURPOSE: Despite many new procedures, radical prostatectomy remains one of the commonest methods of treating clinically localized prostate cancer. Both from the physician's and the patient's point of view, it is important to have objective estimation of the likelihood of recurrence, which forms the foundation for treatment selection for an individual patient. Currently, it is difficult to predict the probability of biochemical recurrence (rising serum prostate specific antigen [PSA] concentration) in an individual patient, and approximately 30% of the patients do experience recurrence. Tools predicting the recurrence will be of immense practical utility in the treatment selection and planning follow up. We have utilized preoperative parameters through a computer based genetic adaptive neural network model to predict recurrence in such patients, which can help primary care physicians and urologists in making management recommendations. PATIENTS AND METHODS: Fourteen hundred patients who underwent radical prostatectomy at participating institutions form the subjects of this study. Demographic data such as age, race, preoperative PSA, systemic biopsy based staging and Gleason scores were used to construct a neural network model. This model simulated the functioning of a trained human mind and learned from the database. Once trained, it was used to predict the outcomes in new patients. RESULTS: The patients in this comprehensive database were representative of the average prostate cancer patients as seen in USA. Their mean age was 68.4 years, the mean PSA concentration before surgery was 11.6 ng/mL, and 67% patients had a Gleason sum of 5 to 7. The mean length of follow-up was 41.5 months. Eighty percent of the cancers were clinical stage T2 and 5% T3. In our series, 64% of patients had pathologically organ-confined cancer, 33% positive margins, and 14% had seminal vesicle invasion. Lymph node positive patients were not included in this series. Progression as judged by serum PSA was noted in 30.6%. With entry of a few routinely used parameters, the model could correctly predict recurrence in 76% of the patients in the validation set. The area under the curve was 0.831. The sensitivity was 85%, the specificity 74%, the positive predictive value 77%, and the negative predictive value of 83%. CONCLUSION: It was possible to predict PSA recurrence with a high accuracy (76%). Physicians desiring objective treatment counseling can use this model, and significant cost savings are anticipated because of appropriate treatment selection and patient-specific follow-up protocols. This technology can be extended to other treatments such as watchful waiting, external-beam radiation, and brachytherapy.

Aged↗
PubMed2006

Finding motifs in protein secondary structure for use in function prediction.

This paper presents a novel algorithm for the discovery of biological sequence motifs. Our motivation is the prediction of gene function. We seek to discover motifs and combinations of motifs in the secondary structure of proteins for application to the understanding and prediction of functional classes. The motifs found by our algorithm allow both flexible length structural elements and flexible length gaps and can be of arbitrary length. The algorithm is based on neither top-down nor bottom-up search, but rather is dichotomic. It is also "anytime," so that fixed termination of the search is not necessary. We have applied our algorithm to yeast sequence data to discover rules predicting function classes from secondary structure. These resultant rules are informative, consistent with known biology, and a contribution to scientific knowledge. Surprisingly, the rules also demonstrate that secondary structure prediction algorithms are effective for membrane proteins and suggest that the association between secondary structure and function is stronger in membrane proteins than globular ones. We demonstrate that our algorithm can successfully predict gene function directly from predicted secondary structure; e.g., we correctly predict the gene YGL124c to be involved in the functional class "cytoplasmic and nuclear degradation." Datasets and detailed results (generated motifs, rules, evaluation on test dataset, and predictions on unknown dataset) are available at www.aber.ac.uk/compsci/Research/bio/dss/yeast.ss.mips/, and www.genepredictions.org.

Algorithms↗
PubMed2004

Prevalence and prediction of unrecognised diabetes mellitus and impaired glucose tolerance following acute stroke.

BACKGROUND: diabetes mellitus not only increases the risk of ischaemic stroke two- to four-fold but also adversely inXuences prognosis. The prevalence of recognised diabetes mellitus in acute stroke patients is between 8 and 20%, but between 6 and 42% of patients may have undiagnosed diabetes mellitus before presentation. Post-stroke hyperglycaemia is frequent and of limited diagnostic value and the oral glucose tolerance test assumes that the patient is clinically stable and eating normally. There is a need for a simple and reliable method to predict new diabetes mellitus in acute stroke patients. OBJECTIVES: to determine the prevalence of unrecognised diabetes mellitus and impaired glucose tolerance on hospital admission and 12 weeks later in acute stroke patients with post-stroke hyperglycaemia > or = 6.1 mmol/l. To measure the accuracy of hyperglycaemia and elevated glycosylated haemoglobin concentration in predicting the presence of unrecognised diabetes mellitus at 12 weeks. DESIGN: acute (<24 hours) stroke patients (cerebral infarction and primary intracerebral haemorrhage) with admission hyperglycaemia between 6.0 and 17 mmol/l and without a previous history of insulin-treated diabetes mellitus who were randomised into the Glucose Insulin in Stroke Trial between October 1997 and May 1999 were studied. The Glucose Insulin in Stroke Trial is a randomised controlled trial investigating the benefits of maintaining euglycaemia in acute stroke patients with mild to moderate hyperglycaemia. At 12 weeks, survivors underwent a 75 g oral glucose tolerance test. The positive predictive value and negative predictive value of admission plasma glucose > or = 6.1 mmol/l and elevated glycosylated haemoglobin concentration in predicting the presence of diabetes mellitus were used to estimate the prevalence of unrecognised diabetes mellitus in a consecutive series of 582 acute stroke admissions. RESULTS: 582 consecutive acute stroke patients were assessed for eligibility for the Glucose Insulin Stroke Trial, of whom 83 (14%) had recognised diabetes mellitus. One hundred and forty-two patients were randomised and 62 underwent a 3-month oral glucose tolerance test, of whom 26 (42%) had normal glucose tolerance, 23 (37%) had impaired glucose tolerance and 13 (21%) had diabetes mellitus. Admission plasma glucose > or = 6.1 mmol/l and glycosylated haemoglobin > or = 6.2% predicted the presence of previously unrecognised diabetes mellitus at 12 weeks with a positive predictive value of 80% and negative predictive value of 96%. The estimated prevalence of unrecognised diabetes mellitus in the total series of acute stroke admissions was 16-24%. CONCLUSIONS: one-third of all acute stroke patients may have diabetes mellitus. For patients presenting with post-stroke hyperglycaemia, impaired glucose tolerance or diabetes mellitus is present in two-thirds of survivors at 12 weeks. Admission plasma glucose > or = 6.1 mmol/l combined with glycosylated haemoglobin > or = 6.2% are good predictors of the presence of diabetes mellitus following stroke.

Acute Disease↗
PubMed1997

Predicting energy needs in ventilator-dependent critically ill patients: effect of adjusting weight for edema or adiposity.

Predicting energy needs in critical illness can be difficult because of uncertainties about the influence of multiple factors on energy expenditure. Understanding these components is important to avoid limiting optimal outcome by underfeeding and to avoid complications of overfeeding. Prediction strategies often use a patient's weight to estimate needs. For overweight patients, there is controversy as to whether actual or modified weight should be used in predictions. This study was designed to evaluate a proposed technique to improve the accuracy of predicting energy needs in critically ill, overweight subjects. Subjects' energy needs were predicted [with Harris-Benedict equation (HBE) and kilojoules per kilogram (KPK) strategies] by using both actual weight and an adjusted weight developed to attempt to more accurately reflect lean mass. Results were compared with measured energy expenditure determined by indirect calorimetry. Results indicated that use of actual weights in predictions for overweight subjects may lead to overfeeding. Use of adjusted weights led to more accurate energy predictions with the KPK than with the HBE strategy. Adjusted-weight strategies could explain > 45% of the variability of resting energy expenditure in subjects 130-159% of ideal body weight. Results of this study suggest that using adjusted weights with the KPK prediction strategy may be preferable for this population, particularly for patients > or = 130% of ideal body weight. This study also indicated that multiple diagnoses may not lead to increased energy requirements.

Adolescent↗
PubMed2000

MaxSub: an automated measure for the assessment of protein structure prediction quality.

MOTIVATION: Evaluating the accuracy of predicted models is critical for assessing structure prediction methods. Because this problem is not trivial, a large number of different assessment measures have been proposed by various authors, and it has already become an active subfield of research (Moult et al. (1997,1999) and CAFASP (Fischer et al. 1999) prediction experiments have demonstrated that it has been difficult to choose one single, 'best' method to be used in the evaluation. Consequently, the CASP3 evaluation was carried out using an extensive set of especially developed numerical measures, coupled with human-expert intervention. As part of our efforts towards a higher level of automation in the structure prediction field, here we investigate the suitability of a fully automated, simple, objective, quantitative and reproducible method that can be used in the automatic assessment of models in the upcoming CAFASP2 experiment. Such a method should (a) produce one single number that measures the quality of a predicted model and (b) perform similarly to human-expert evaluations. RESULTS: MaxSub is a new and independently developed method that further builds and extends some of the evaluation methods introduced at CASP3. MaxSub aims at identifying the largest subset of C(alpha) atoms of a model that superimpose 'well' over the experimental structure, and produces a single normalized score that represents the quality of the model. Because there exists no evaluation method for assessment measures of predicted models, it is not easy to evaluate how good our new measure is. Even though an exact comparison of MaxSub and the CASP3 assessment is not straightforward, here we use a test-bed extracted from the CASP3 fold-recognition models. A rough qualitative comparison of the performance of MaxSub vis-a-vis the human-expert assessment carried out at CASP3 shows that there is a good agreement for the more accurate models and for the better predicting groups. As expected, some differences were observed among the medium to poor models and groups. Overall, the top six predicting groups ranked using the fully automated MaxSub are also the top six groups ranked at CASP3. We conclude that MaxSub is a suitable method for the automatic evaluation of models.

Algorithms↗
PubMed2025

Multimodal deep learning for immunotherapy response prediction and biomarker discovery in non-small cell lung cancer.

OBJECTIVE: Immunotherapy has emerged as a promising treatment for advanced non-small cell lung cancer (NSCLC), but accurately predicting which patients will benefit from it remains a major clinical challenge. To address this, we aim to develop a novel multimodal method, DeepAFM, that integrates histopathology, genomic features, and clinical information to predict patient responses to anti-PD-(L)1 immunotherapy. MATERIALS AND METHODS: A total of 93 patients with advanced NSCLC were included in this study. Histopathological whole-slide images were processed using a self-supervised VQVAE2 for representation learning. PCA and K-means clustering were then applied for dimensionality reduction and feature grouping. Key regions of interest were visualized through permutation importance evaluation and color-coding techniques. The extracted histopathological features, along with genomic alterations and clinical variables, were integrated into the DeepAFM multimodal prediction model. RESULTS: The DeepAFM achieved a high predictive performance with an area under the curve (AUC) of 0.77 (95% confidence interval: 0.69-1.00). Attention-based heatmaps revealed that the model could identify critical pathological patterns, genomic mutations, and clinical indicators associated with patient responses to immunotherapy. DISCUSSION: The integration of multimodal data enabled the model to capture complex interactions among pathology, genomics, and clinical characteristics, enhancing the interpretability and predictive power of immunotherapy response prediction. The visualization techniques facilitated the identification of biologically meaningful features and potential biomarkers. CONCLUSION: This study demonstrates the effectiveness of the DeepAFM in predicting responses to immunotherapy in advanced NSCLC. The approach not only improves prediction accuracy but also provides valuable insights for personalized treatment strategies and biomarker discovery.

Humans↗