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 1,027 records · Page 57Linked to original sources
PubMed1999

Survival prediction of terminally ill cancer patients by clinical symptoms: development of a simple indicator.

BACKGROUND: Although accurate prediction of survival is essential for palliative care, no clinical tools have been established. METHODS: Performance status and clinical symptoms were prospectively assessed on two independent series of terminally ill cancer patients (training set, n = 150; testing set, n = 95). On the training set, the cases were divided into two groups with or without a risk factor for shorter than 3 and 6 weeks survival, according to the way the classification achieved acceptable predictive value. The validity of this classification for survival prediction was examined on the test samples. RESULTS: The cases with performance status 10 or 20, dyspnea at rest or delirium were classified in the group with a predicted survival of shorter than 3 weeks. The cases with performance status 10 or 20, edema, dyspnea at rest or delirium were classified in the group with a predicted survival of shorter than 6 weeks. On the training set, this classification predicted 3 and 6 weeks survival with sensitivity 75 and 76% and specificity 84 and 78%, respectively. On the test populations, whether patients survived for 3 and 6 weeks or not was predicted with sensitivity 85 and 79% and specificity 84 and 72%, respectively. CONCLUSION: Whether or not patients live for 3 and 6 weeks can be acceptably predicted by this simple classification.

Aged↗
PubMed1994

Prediction of axillary lymph node status in breast cancer patients by use of prognostic indicators.

BACKGROUND: If axillary lymph node status of breast cancer patients could be accurately predicted from basic clinical information and from characteristics of their primary tumors, then many patients could be spared axillary lymph node dissection. Tumor size alone does not allow the identification of groups with very low or high risk of being axillary node positive. PURPOSE: Our goal was to investigate the possibility of using prognostic indicators to predict axillary node status of patients with primary breast cancer. METHODS: Data from 26,683 patients from the National Breast Cancer Tissue Resource were used in this study. Patients in this dataset were randomly assigned to a training set (patient information used to construct predictive models) or a validation set (patient information used to prospectively evaluate predictive models). The records of a total of 11,964 case patients that had complete prognostic factors and pathologic data were analyzed: 5963 patients in the training set and 6001 patients in the validation set. All of the patients studied had tumors 5 cm or less in size and at least 15 axillary lymph nodes that had been examined. Data used for construction of the predictive models were available for all patients and included tumor size, number of nodes positive, patient age, quantitative estrogen receptor levels, quantitative progesterone receptor (PgR) levels, DNA flow cytometry-derived ploidy, and S-phase fraction. Logistic regression models were used to predict nodal status. RESULTS: Multivariate predictive models were produced that used tumor size, patient age, S phase, and PgR as independent predictors. These models allowed identification of patient risks of being node positive ranging from 6%-79% and as having 10 or more positive nodes ranging from less than 1% to slightly more than 30%. CONCLUSION: Addition of prognostic indicator information to tumor size can refine estimates of whether a patient is likely to be node positive. However, no patient subsets could be identified as having greater than 95% chance of being node negative or node positive. IMPLICATIONS: These predictive models cannot alleviate the necessity of axillary node dissection for staging of breast cancer patients in situations in which nodal status would affect therapeutic decisions. Subsets of patients could be identified who had a less than 5% chance of having 10 or more positive nodes. Thus, some patients could be spared axillary dissection if it was being performed solely to identify patients with this high-risk feature.

Axilla↗
PubMed1997

Evaluation of logistic versus linear regression models for predicting pulmonary hypertension syndrome (ascites) using cold exposure or pulmonary artery clamp models in broilers.

Syndromes such as ascites (pulmonary hypertension syndrome) present difficulties both in the interpretation of associated physiological observations and in their analyses. The ability to predict which physiological variables have the greatest influence on survival or, more importantly, which individuals are most susceptible or resistant to ascites would be very useful selection tools. When addressed in this manner, ascites data become binary data sets (healthy or affected). Binary data can be problematic in that they do not meet all of the assumptions necessary for more traditional analyses such as ANOVA and linear regression. Binary data are discrete and do not have normally distributed errors, which violates a fundamental assumption of linear models. The predictive abilities of linear and logistic regression were evaluated in two replicated experiments using two methods to induce ascites, cold exposure (COLD) and surgical clamping of one pulmonary artery (PAC). The logistic and linear predictive models were derived using the same data and variables. The first data set from PAC and COLD were used to develop the predictive models and the replicate data sets of PAC and COLD were used as "test data sets" for the prediction of ascites. The linear models developed were complex, using four or five variables and requiring up to seven different measurements. On average, the linear models predicted ascites correctly 87.6% of the time. The logistic models were simple (single variable) models that predicted ascites correctly 92.0% of the time. The variables used in the logistic models were derivations of the ratio of right ventricular weight to total ventricular weight, either corrected for age or the body weight of the bird. Although linear regression predicted the incidence of ascites almost as well as logistic regression did, logistic regression is the more appropriate test statistic to use.

Analysis of Variance↗
PubMed2003

The predictive value of modified computerized thromboelastography and platelet function analysis for postoperative blood loss in routine cardiac surgery.

UNLABELLED: Hemorrhage after cardiopulmonary bypass (CPB) remains a clinical problem. Point-of-care tests to identify hemostatic disturbances at the bedside are desirable. In the present study, we evaluated the predictive value of two point-of-care tests on postoperative bleeding after routine cardiac surgery. Prospectively, 255 consecutive patients were studied to compare the ability of modified thromboelastography (ROTEG) as well as a platelet function analyzer (PFA-100) to predict postoperative blood loss. Measurements were performed at three time points: preoperatively, during CPB, and after protamine administration with three modified thromboelastography and PFA tests. The best predictors of increased bleeding tendency were the tests performed after CPB. The angle alpha is the best predictor (area under the receiver operating characteristic curve 0.69) and, in combination with the adenosine diphosphate-PFA test, the predictive accuracy is enhanced (area under the receiver operating characteristic curve 0.73). The negative predictive value for the angle alpha is 82%, although the positive predictive value is small (41%). Thromboelastography is a better predictor than PFA. In routine cardiac surgery, impaired hemostasis as identified by point-of-care tests does not inevitably lead to hemorrhage postoperatively. However, patients with normal test results are unlikely to bleed for hemostatic reasons. Bleeding in these patients is probably caused surgically. The high negative predictive value supports early identification and targeted treatment of surgical bleeding by distinguishing it from a significant coagulopathy. IMPLICATIONS: Thrombelastography and platelet function analysis in routine cardiac surgery demonstrate high negative predictive values for postoperative bleeding, which supports early identification and targeted treatment of surgical bleeding by distinguishing it from a significant coagulopathy. The positive predictive values are small. The best predictors are thrombelastography values obtained after cardiopulmonary bypass.

Aged↗
PubMed1997

AIDS prognosis based on HIV-1 RNA, CD4+ T-cell count and function: markers with reciprocal predictive value over time after seroconversion.

OBJECTIVE: HIV-1 RNA levels in peripheral blood are strongly associated with progression to AIDS, CD4+ T-cell decline, or death. Their predictive value is reportedly independent of the predictive value of CD4+ T-cell counts. Because the interrelations between these parameters of HIV-1 infection are poorly understood, we studied the kinetics and predictive value of serum HIV-1 RNA levels, CD4+ T-cell counts, and T-cell function. DESIGN AND METHODS: HIV RNA levels, CD4+ T-cell counts, and T-cell function were measured from seroconversion to AIDS in 123 homosexual men who seroconverted during a prospective study and were followed over 10 years. RESULTS: Two patterns of median HIV-1 RNA levels were found during infection: a steady-state and a 'U-shaped' curve. Steady-state high RNA levels were related to rapid disease progression. For the U-shaped curve, there were groups with high and low RNA levels related to disease progression. At 1 year after seroconversion, RNA level was the only marker that was strongly predictive. Furthermore, decreasing RNA levels in the first year following seroconversion were related to better prognosis than stable low levels. Low CD4+ T-cell count and T-cell function became predictive of progression to AIDS at 2 and 5 years after seroconversion, respectively. CONCLUSIONS: With ongoing infection, the predictive value of low CD4+ T-cell count and T-cell function increases, whereas the predictive value of high HIV-1 RNA level decreases. These findings reflect the observation that infection with HIV progressively leads towards immune deficiency, which in later stages is most predictive of disease progression.

Acquired Immunodeficiency Syndrome↗
PubMedDate not supplied

Prediction of equilibrated urea in children on chronic hemodialysis.

Urea rebound (UR) after hemodialysis (HD) requires the use of equilibrated urea (Ceq) instead of immediate end-dialysis urea (Ct) for correct quantification of HD, which is impractical. A new formula for predicting Ceq in children is suggested in our study. Thirty eight standard pediatric HD sessions (single pool Kt/V = 1.70 +/- 0.35, K = 4.65 +/- 1.14 ml/min/kg, UF coeff. = 3.2-6.2 ml/h/mm Hg, t = 3.80 +/- 0.46 h) in 15 children (M: 6, F: 9), ages 14.5 +/- 3.28 years were analyzed. Blood samples were taken: before, 70 min from the start, at the end, and 60 min after the end of HD sessions. After correlating UR (20.32 +/- 7.74%) to various HD parameters, we found that it was mainly determined by HD efficiency parameters. Therefore we correlated Ceq to HD efficiency parameters (Ct, urea reduction ratio, Kt/V, and K/V) and found a very high correlation between Ct and Ceq (r = 0.973). Linear regression analysis was used to further investigate this relationship, and a new formula to predict Ceq from Ct was obtained (Ceq = 1.085 Ct + 0.729, R2 = 0.946, SE = 0.49, absolute residuals = 0.38 +/- 0.29 mmol/L). In a validation study (10 HD sessions with new set of urea blood samples) the results obtained by the new formula were compared with measured values of Ceq and those obtained by the Smye formulae. Values predicted by the new formula (9.91 +/- 2.92 mmol/L) were not significantly different from the measured values (10.33 +/- 3.44 mmol/L). Absolute error of the new formula was 0.78 +/- 0.73 mmol/L, median 0.65; ie., 6.93 +/- 5.3%, median 7.7%. Ceq predicted by the Smye formulae (10.95 +/- 4.18 mmol/L) also did not significantly differ from the measured values, but absolute error of predicted values was markedly higher (1.21 +/- 0.90 mmol/L, median 0.89; 11.73 +/- 7.72%, median 10.11%; p < 0.05). When predicted Ceq was used for calculating equilibrated Kt/V (eKt/V), the new formula resulted in lower absolute error (0.09 +/- 0.07, median 0.08) than the Smye method (0.14 +/- 0.08, median 0.12). We conclude that our simple formula is sufficiently accurate in predicting Ceq in standard pediatric HD and that it is more accurate than the existing Smye formulae, while requiring only pre- and post-HD urea samples. We suggest the use of the new formula for predicting Ceq, which can then be used instead of Ct for a more accurate estimation of double pool Kt/V, URR, V, and PCR.

Adolescent↗
PubMed2003

Pulse pressure responses to psychological tasks improve the prediction of left ventricular mass: 10 years of follow-up.

OBJECTIVES: To examine the role of casual blood pressure measurements and blood pressure responses to psychological tasks in the prediction of future left ventricular mass index (LVMI), and to determine the importance of different components of blood pressure, and the predictive value of an individual's personal characteristics and antihypertensive medication on future LVMI. METHODS: At baseline, blood pressure was recorded by casual measurements; during tests it was recorded by intra-arterial monitoring. The participants were healthy, untreated 35-45-year old men. Echocardiography data both at baseline and after 10 years of follow-up were available from 65 individuals, of whom 49 (75%) were not taking antihypertensive medication at follow-up. Those not taking antihypertensive medication were included in the prediction of LVMI (g/m2). RESULTS: Baseline LVMI correlated significantly with future LVMI only among the 49 unmedicated individuals (r = 0.52, P < 0.0001). The predictive value of baseline LVMI on future LVMI among them (adjusted coefficient of determination = 0.26) was not improved by the inclusion of casual blood pressure. In contrast, blood pressure responses to the psychological tasks improved the prediction of future LVMI by 4-13%. Pulse pressure was the blood pressure variable that entered the final prediction models; the correlations with future LVMI were best for pulse pressure response to habituation task (r = 0.43, P < 0.05) and to relaxation (r = 0.37, P < 0.05). CONCLUSIONS: To our knowledge, this is the longest prospective follow-up to show that blood pressure responses to psychological tasks improve the prediction of LVMI compared with casual blood pressure measurements. The pulse pressure, which reflects the properties of the arterial wall, is the most significant blood pressure variable in predicting future LVMI.

Adult↗
PubMed2003

Does a routinely measured blood pressure in young adolescence accurately predict hypertension and total cardiovascular risk in young adulthood?

BACKGROUND: It is insufficiently known if routine blood pressure (BP) measurement by school doctors has added predictive value for later hypertension and cardiovascular risk. OBJECTIVE: To assess whether screening of BP in adolescence has additional predictive value to already routinely collected indicators of later hypertension and cardiovascular risk. METHODS: In the Dutch city of Utrecht, routine BPs and anthropometry were collected from school health records of 750 adolescents. In The Hague, standardized repeated BP measurements and anthropometry were available for 262 adolescents. Of both cohorts, 998 now young adults were recently re-examined. Predictors of adult hypertension, systolic blood pressure (SBP) > or = 140 mmHg and/or diastolic blood pressure (DBP) > or = 90 mmHg) and 10-year cardiovascular risk were analysed by logistic regression and area under receiver operator characteristics curve (AUC). RESULTS: A total of 167 young adults had hypertension. Single adolescent SBP and DBP predicted hypertension: odds ratio (OR) 1.04 per mmHg [95% confidence interval (CI): 1.03-1.06], OR 1.02 (1.00-1.04), respectively, but with little discriminative power. Gender, adolescent body mass index (BMI) and age combined predicted hypertension: AUC 0.71 (0.67-0.75), which slightly improved by adding SBP: AUC 0.74 (0.70-0.77); difference in AUC 0.03 (0.002-0.06). SBP exclusively predicted hypertension within men: OR 1.03 (1.01-1.04), AUC: 0.59 (0.53-0.65), and within women: OR 1.08 (1.05-1.11), AUC 0.74 (0.67-0.82). However, an adolescent BP of > or = 120 mmHg did not efficiently detect hypertensive men, while it detected 57.9% of hypertensive women. Only young adult men had meaningful 10-year cardiovascular risks, which only SBP predicted: OR risk score > 95th percentile 1.04 (1.02-1.07), AUC 0.67 (0.60-0.75). CONCLUSION: A single routine BP measurement in adolescent girls efficiently predicts young adult hypertension. In adolescent boys, BP predicts young adult 10-year cardiovascular risk.

Adolescent↗
PubMed1992

Prospective comparison of clinical judgment and APACHE II score in predicting the outcome in critically ill surgical patients.

Prospective identification of patients who will not survive has been proposed as a means of limiting utilization of medical resources including critical care. This study prospectively compared prediction of outcome for surgical ICU patients by clinical assessment and the APACHE II score. Five hundred seventy-eight patients were assessed within 24 hours of admission by the ICU attending physician and predicted to live or die. An APACHE II score was calculated in that same time period. All data were stored in a data base and compared with actual SICU outcome. There were 40 deaths in 578 patients (6.9%). The clinical assessment had an overall accuracy of 95.2% vs. 90.9% for APACHE II. The Pearson correlation coefficients for the two methods of prediction were 0.59 for clinical assessment and 0.44 for APACHE II. Predictive power was not greatly improved by combining both prediction methods. Over 40% of patients predicted to die by both methods actually survived. This study demonstrates that clinical assessment is superior to APACHE II in predicting outcome in this group of surgical patients, although the difference is small. In addition, this study suggests that neither clinical assessment nor the APACHE II score, when obtained within 24 hours of admission, is very reliable at predicting which surgical ICU patients will die.

Adolescent↗
PubMed2001

Predictors of Gleason pattern 4/5 prostate cancer on prostatectomy specimens: can high grade tumor be predicted preoperatively?

PURPOSE: Radical prostatectomy provides excellent cancer control in men with clinically localized prostate carcinoma. However, to our knowledge preoperative parameters for distinguishing indolent from clinically significant cancer are not well characterized. In fact, recent evidence suggests that the percent of Gleason pattern 4/5 carcinoma in the complete radical prostatectomy specimen is one of the strongest predictors of prostate cancer progression and a valid measure of cancer severity. However, it is unclear whether preoperative parameters, including biopsy Gleason pattern 4/5 carcinoma, may predict radical prostatectomy Gleason pattern 4/5 disease and, thereby, distinguish indolent from clinically significant cancer. MATERIALS AND METHODS: We prospectively obtained 101 consecutive radical prostatectomy specimens and processed them in whole mount fashion. In addition to total tumor volume, we determined tumor volume for each Gleason pattern. Biopsy tumor area was measured in a similar fashion. Univariate and multivariate analyses were performed to identify preoperative clinical and pathology parameters for predicting Gleason pattern 4/5 carcinoma on prostatectomy specimens. RESULTS: Biopsy Gleason score 7 or greater, Gleason pattern 4/5 carcinoma, perineural invasion and biopsy tumor area had statistically significant associations for identifying Gleason pattern 4/5 carcinoma on prostatectomy specimens. Logistic regression models for predicting any or greater than 10% Gleason pattern 4/5 carcinoma on prostatectomy specimens revealed that an area of pattern 4/5 disease of greater than 0.01 cm.2 on biopsy was the best single predictor with odds ratios of 15.0 (95% confidence interval 3.3 to 69.0, p = 0.0005) and 3.9 (95% confidence interval 1. 4 to 10.9, p = 0.009), respectively. For predicting any pattern 4/5 carcinoma on prostatectomy specimens a biopsy area of pattern 4/5 disease of greater than 0.01 cm.2 had only 38% sensitivity but 96% specificity. Similarly for predicting significant pattern 4/5 disease on prostatectomy specimens, defined as 10% or greater pattern 4/5, sensitivity and specificity for a biopsy area of greater than 0.01 cm.2 were 34% and 88%, respectively. Therefore, due to high false-negative rates these models had limited predictive value on an individual basis. CONCLUSIONS: Biopsy parameters such as Gleason pattern 4/5 carcinoma may provide adequate specificity for predicting clinically significant cancer, as defined by high grade Gleason patterns in the corresponding radical prostatectomy specimen. However, the accuracy of these parameters for predicting indolent cancer is limited by a prohibitive rate of false-negative findings.

Adenocarcinoma↗
PubMed2001

Relevance of 99mTc-MIBI rest uptake, ejection fraction and location of contractile abnormality in predicting myocardial recovery after revascularization.

The aim of this study was to analyse the influence of rest technetium-99m-methoxy-isobutyl-isonitrile (99mTc-MIBI) uptake, left ventricular ejection fraction (EF) and dysfunctional location in the prediction of myocardial viability. Rest 99mTc-MIBI single photon emission computed tomography (SPECT) was analysed in 82 patients (59+/-9 years, 70 men, 12 women) with one or more segments showing severe hypokinesia, akinesia or dyskinesia who had undergone coronary revascularization. Before and within 3-6 months after the revascularization, gated blood pool scintigraphy was performed. In the post-revascularization control, contractile recovery was observed in 48.7% (155/318) of the segments with severe hypokinesia, akinesia or dyskinesia. Significant increases in sensitivity (53%, 72% and 91%, P<0.0001) and negative predictive value (62%, 68% and 79%, P = 0.01) were observed with decreasing rest uptake 99mTc-MIBI levels of 50%, 40% and 30%, respectively. The decrease in specificity was also significant (67%, 53% and 32%, P<0.0001). The negative predictive value was higher than the positive predictive value mainly in patients with EF < or = 0.35 and with anterior dysfunction. In logistic regression analysis, uptake levels and EF were independent variables that influenced sensitivity and specificity. The negative predictive value was influenced by EF and the positive predictive value only by dysfunctional location. This study suggests that the negative predictive value of 99mTc-MIBI SPECT is higher than the positive predictive value, mainly in patients with EF < or = 0.35, and that the rest uptake level, EF and dysfunctional location are factors that must be considered when results of 99mTc-MIBI SPECT are analysed.

Aged↗
PubMed1994

Use of the potential acuity meter and laser interferometer to predict visual acuity after macular hole surgery.

PURPOSE: As reported anatomic success rates for macular hole surgery increase, a method of accurately predicting postoperative visual acuity has increased clinical utility. METHODS: A series of 18 patients undergoing vitrectomy for idiopathic macular holes was evaluated prospectively for best refracted preoperative visual acuity, potential acuity meter (PAM) reading, and laser interferometer (LI) reading. Best refracted visual acuity after surgery was recorded and analyzed to assess the predictive value of the PAM and LI readings. RESULTS: The LI correctly predicted the final visual result in 6 of the 10 patients who achieved a final visual acuity of 20/50 or better (P = < 0.011). The PAM did not accurately predict postoperative visual acuity for any of the eyes with a final visual acuity of 20/50 or better. Both correctly predicted outcome in the 7 patients with final visual acuity worse than 20/50. The LI prediction was within 2 lines of final visual acuity in 12 (70%) of 17 anatomically successful cases, and the PAM was within 2 lines of final visual acuity in 11 (64%) of 17 cases. CONCLUSION: Both the LI and PAM were modestly accurate in predicting final visual acuity after macular hole surgery, but the LI was more accurate in predicting a visual acuity of 20/50 or better.

Aged↗
PubMed1991

Inpatient cardiopulmonary resuscitation: is survival prediction possible?

We retrospectively reviewed 443 patients who had cardiopulmonary resuscitation (CPR). The focus of the study was to discover what preexisting factors should be assessed to determine the probability of survival. There were 88 successes out of 340 cases (25.9%). The absence of a previous myocardial infarction (MI), shock, partial pressure of oxygen (PaO2) less than 60 mm Hg, blood urea nitrogen (BUN) level greater than 20 mg/dL, pneumonia, pulmonary edema, and oliguria were found to predict a successful outcome. Logistic regression was used to predict percentage of successes in the various groups of patients with various clinical characteristics. The observed and predicted numbers of successes were in close agreement in most cases. We also constructed a classification function to predict whether an individual subject would survive the event for which CPR was required. Sixty-seven of the 88 observed successes would have been predicted, for an estimated sensitivity of 76%, and 164 of the 252 failures would have been predicted, for an estimated specificity of 65%. A large percentage (24%) of cases in which the patient actually survived CPR would have been predicted to be failures. We conclude that preexisting factors before a cardiopulmonary arrest do not accurately predict survival after CPR.

Aged↗
PubMed2004

An artificial neural network ensemble to predict disposition and length of stay in children presenting with bronchiolitis.

BACKGROUND: Artificial neural networks apply complex non-linear functions to pattern recognition problems. An ensemble is a 'committee' of neural networks that usually outperforms single neural networks. Bronchiolitis is a common manifestation of viral lower respiratory tract infection in infants and toddlers. OBJECTIVE: To train artificial neural network ensembles to predict the disposition and length of stay in children presenting to the Emergency Department with bronchiolitis. METHODS: A specifically constructed database of 119 episodes of bronchiolitis was used to train, validate, and test a neural network ensemble. We used EasyNN 7.0 on a 200 Mhz pentium PC with a maths co-processor. The ensemble of neural networks constructed was subjected to fivefold validation. Comparison with actual and predicted dispositions was measured using the kappa statistic for disposition and the Kaplan-Meier estimations and log rank test for predictions of length of stay. RESULTS: The neural network ensembles correctly predicted disposition in 81% (range 75-90%) of test cases. When compared with actual disposition the neural network performed similarly to a logistic regression model and significantly better than various 'dumb machine' strategies with which we compared it. The prediction of length of stay was poorer, 65% (range 60-80%), but the difference between observed and predicted lengths of stay were not significantly different. CONCLUSION: Artificial neural network ensembles can predict disposition for infants and toddlers with bronchiolitis; however, the prediction of length of hospital stay is not as good.

Bronchiolitis↗
PubMed2003

Audibility-index predictions of normal-hearing and hearing-impaired listeners' performance on the connected speech test.

OBJECTIVE: In a previous study (Sherbecoe & Studebaker, 2002), we derived a frequency-importance function and a transfer function for the audio compact disc version of the Connected Speech Test (CST). The current investigation evaluated the validity of these audibility-index (AI) functions based on how well they predicted data from four published studies that presented the CST to normal-hearing and hearing-impaired subjects. DESIGN: AI values were calculated for the test conditions received by 78 normal-hearing and 72 hearing-impaired subjects from the selected studies. The observed CST scores and AI values for these conditions/subjects were then plotted and the dispersion of the data compared to the expected range based on critical differences. The AI values for the conditions/subjects were also converted into expected CST scores and subtracted from their corresponding observed scores to determine the distribution of the resulting difference scores and the relationship between the difference scores and subject age. RESULTS: Good predictions were obtained for normal-hearing subjects who had been tested under audio-only conditions but not those who had received audiovisual tests. The expected scores for the latter subjects were too low when the AI accounted only for audibility and too high when it included the correction for visual cues from ANSI S3.5-1997. All of the hearing-impaired subjects had been tested under audio-only conditions. In their case, the mean difference between the observed and the expected scores was comparable with the audio-only mean for the normal-hearing subjects when the AI included corrections for speech level distortion and hearing loss desensitization. However, the hearing-impaired subject data had greater variability. The predictions for these subjects also decreased in accuracy when subject age increased beyond 70 yr despite the application of an AI correction for age. CONCLUSIONS: The results of this study suggest that the AI functions derived for the CST satisfactorily predict the scores of normal-hearing subjects when they listen in speech babble under audio-only conditions but not when they receive visual cues. To obtain accurate predictions for the audiovisual form of the CST, it will be necessary to develop new ANSI-style AI correction equations for visual cues or new AI functions based on audiovisual test scores. If the current AI functions are used to predict the scores of hearing-impaired listeners tested under audio-only conditions, the AI should include corrections for the effects of speech level and hearing loss. A correction for subject age also could be applied, if it seems appropriate to do so. In either case, however, the predictions are still likely to be less accurate than the predictions for normal-hearing subjects. This may be because speech recognition deficits in people with hearing loss are not due solely to diminished audibility. Hearing-impaired subjects, particularly if they are elderly, also may be more susceptible to masking effects or other factors not accounted for by the AI.

Adolescent↗
PubMed2005

Validation of a prediction rule for renal artery stenosis.

OBJECTIVES: We previously developed a prediction rule to estimate the probability of renal artery stenosis. This rule should be validated before it can be used reliably to select hypertensive patients for renal angiography. We determined the validity of the prediction rule in recent patients and in other settings. DESIGN: We studied three aspects of validity (agreement between predicted and observed probability of stenosis, discriminative ability, and clinical usefulness) in 180 consecutive patients with drug-resistant hypertension and normal or mildly impaired renal function, who visited six hypertension clinics of academic and community hospitals in the Netherlands. Thirty-five patients (19%) had a significant stenosis. RESULTS: The clinical characteristics in the rule (age, sex, vascular disease, recent onset of hypertension, smoking, body mass index, abdominal bruit, serum creatinine concentration, and hypercholesterolemia) had similar predictive value in the validation sample and development sample. The predicted probabilities of stenosis agreed well with the observed frequencies (Hosmer-Lemeshow goodness-of-fit test, P = 0.87). The prediction rule discriminated reasonably between patients with and without stenosis in the validation sample with an area under the receiver operating characteristic curve of 0.71. If only patients with predicted probabilities of stenosis of 5% or more were referred for renal angiography, the number of referrals was reduced by 20%, while 9% of patients with a stenosis were missed. CONCLUSIONS: The prediction rule was valid in more recently treated patients in other settings. If used conservatively, the rule can reliably exclude a small proportion of patients from angiography.

Age Distribution↗
PubMed2004

Predicting the presence and side of extracapsular extension: a nomogram for staging prostate cancer.

PURPOSE: We developed a model to predict the side specific probability of extracapsular extension (ECE) in radical prostatectomy (RP) specimens based on the clinical features of the cancer. MATERIALS AND METHODS: We studied 763 patients with clinical stage T1c-T3 prostate cancer who were diagnosed by systematic needle biopsy and subsequently treated with RP. Candidate predictor variables associated with ECE were clinical T stage, the highest Gleason sum in any core, percent positive cores, percent cancer in the cores from each side and serum prostate specific antigen (PSA). Receiver operating characteristic (ROC) analyses were performed to assess the predictive value of each variable alone and in combination. We constructed and internally validated nomograms to predict the side specific probability of ECE based on logistic regression analysis. RESULTS: Overall 30% of the patients and 17% of 1,526 prostate lobes (left or right) had ECE. The areas under the ROC curves (AUC) of the standard features in predicting side specific probability of ECE were 0.627 for PSA, 0.695 for clinical T stage on each side and 0.727 for Gleason sum on each side. When these features were combined predictive accuracy increased to 0.788. The highest value (0.806) was achieved by adding the percent positive cores and the percent cancer in the biopsy specimen to the standard features. The resulting nomograms were internally validated and had excellent calibration and discrimination accuracy. CONCLUSIONS: Standard clinical features of prostate cancer in each lobe-PSA, palpable induration and biopsy Gleason sum-can be used to predict the side specific probability of ECE in RP specimens. The predictive accuracy is increased by adding information from systematic biopsy results. The predictive nomograms are sufficiently accurate for use in clinical practice in decisions such as wide versus close dissection of the cavernous nerves from the prostate.

Adult↗
PubMed2002

Comparison of Artificial Neural Network (ANN) Model Development Methods for Prediction of Macroinvertebrate Communities in the Zwalm River Basin in Flanders, Belgium.

Modelling has become an interesting tool to support decision making in water management. River ecosystem modelling methods have improved substantially during recent years. New concepts, such as artificial neural networks, fuzzy logic, evolutionary algorithms, chaos and fractals, cellular automata, etc., are being more commonly used to analyse ecosystem databases and to make predictions for river management purposes. In this context, artificial neural networks were applied to predict macroinvertebrate communities in the Zwalm River basin (Flanders, Belgium). Structural characteristics (meandering, substrate type, flow velocity) and physical and chemical variables (dissolved oxygen, pH) were used as predictive variables to predict the presence or absence of macroinvertebrate taxa in the headwaters and brooks of the Zwalm River basin. Special interest was paid to the frequency of occurrence of the taxa as well as the selection of the predictors and variables to be predicted on the prediction reliability of the developed models. Sensitivity analyses allowed us to study the impact of the predictive variables on the prediction of presence or absence of macroinvertebrate taxa and to define which variables are the most influential in determining the neural network outputs.

Amphipoda↗