Previous Scoring Systems Research Articles

575

Carnie, Roslyn; Booth, Malcolm; Shaw, Martin; MacTavish, Pamela; Docking, Robert; Mackay, Andrew; Kinsella, John

NICeSim: an open-source simulator based on machine learning techniques to support medical research on prenatal and perinatal care decision making.

This paper describes NICeSim, an open-source simulator that uses machine learning (ML) techniques to aid health professionals to better understand the treatment and prognosis of premature newborns. The application was developed and tested using data collected in a Brazilian hospital. The available data were used to feed an ML pipeline that was designed to create a simulator capable of predicting the outcome (death probability) for newborns admitted to neonatal intensive care units. However, unlike previous scoring systems, our computational tool is not intended to be used at the patients bedside, although it is possible. Our primary goal is to deliver a computational system to aid medical research in understanding the correlation of key variables with the studied outcome so that new standards can be established for future clinical decisions. In the implemented simulation environment, the values of key attributes can be changed using a user-friendly interface, where the impact of each change on the outcome is immediately reported, allowing a quantitative analysis, in addition to a qualitative investigation, and delivering a totally interactive computational tool that facilitates hypothesis construction and testing. Our statistical experiments showed that the resulting model for death prediction could achieve an accuracy of 86.7% and an area under the receiver operating characteristic curve of 0.84 for the positive class. Using this model, three physicians and a neonatal nutritionist performed simulations with key variables correlated with chance of death. The results indicated important tendencies for the effect of each variable and the combination of variables on prognosis. We could also observe values of gestational age and birth weight for which a low Apgar score and the occurrence of respiratory distress syndrome (RDS) could be less or more severe. For instance, we have noticed that for a newborn with 2000 g or more the occurrence of RDS is far less problematic than for neonates weighing less. The significant accuracy demonstrated by our predictive model shows that NICeSim might be used for hypothesis testing to minimize in vivo experiments. We observed that the model delivers predictions that are in very good agreement with the literature, demonstrating that NICeSim might be an important tool for supporting decision making in medical practice. Other very important characteristics of NICeSim are its flexibility and dynamism. NICeSim is flexible because it allows the inclusion and deletion of variables according to the requirements of a particular study. It is also dynamic because it trains a just-in-time model. Therefore, the system is improved as data from new patients become available. Finally, NICeSim can be extended in a cooperative manner because it is an open-source system.

Massive Plasminogen Activator Inhibitor-1 (PAI-1) Upregulation and Suppressed Fibrinolysis Is the Predominant Phenotype in Severely Injured Trauma Patients

Chapman, Michael P. MD; Moore, Ernest E. MD, FACS; Moore, Hunter B. MD; Gonzalez, Eduardo MD; Chin, Theresa MD; Gamboni, Fabia PhD; Mitra, Sanchayita MS; Banerjee, Anirban PhD; Sauaia, Angela MD, PhD; Silliman, Christopher C. MD, PhD Author Information

The SOAR (Stroke Subtype, Oxford Community Stroke Project Classification, Age, Prestroke Modified Rankin) Score Strongly Predicts Early Outcomes in Acute Stroke

Previous prognostic scoring systems in predicting stroke mortality are complex, require multiple measures that vary with time and failed to produce a simple scoring system. The study aims to derive and internally validate a stroke prognostic scoring system to predict early mortality and hospital length of stay. Data from a U.K. multicenter stroke register were examined (1997-2010). Using a prior hypothesis based on our and others observations, we selected five patient-related factors (age, gender, stroke subtype, clinical classification, and prestroke disability) as candidate prognostic indicators. An 8-point score was derived based on multiple logistic regression model using four out of five variables. Performance of the model was assessed by plotting the estimated probability of in-hospital death against the actual probability by testing for overfitting (calibration) and area under the curve methods (discrimination). The total sample consisted of 12,355 acute stroke patients (ischemic stroke 91.0%). The score predicted both in-patient and seven-day mortality. The crude in-patient mortality were 1.57%, 4.02%, 10.65%, 21.41%, 46.60%, 62.72%, and 75.81% for those who scored 0, 1, 2, 3, 4, 5, and 6, respectively. The calibration of the model revealed no evidence of overfitting (estimated overfitting 0.001). The area under the curve values for both in-hospital and seven-day mortality were 0.79. The score predicted length of stay with a higher score was associated with longer median length of stay in those discharged alive and shorter median length of stay in those who died (P for both <0.001). A simple 8-point clinical score is highly predictive of acute stroke mortality and length of hospital stay. It could be used as prognostic tool in service planning and also to risk-stratify patients to use these outcomes as markers of stroke care quality across institutions.

Estimating selection on synonymous codon usage from noisy experimental data.

A key goal in molecular evolution is to extract mechanistic insights from signatures of selection. A case study is codon usage, where despite many recent advances and hypotheses, two longstanding problems remain: the relative contribution of selection and mutation in determining codon frequencies and the relative contribution of translational speed and accuracy to selection. The relevant targets of selection--the rate of translation and of mistranslation of a codon per unit time in the cell--can only be related to mechanistic properties of the translational apparatus if the number of transcripts per cell is known, requiring use of gene expression measurements. Perhaps surprisingly, different gene-expression data sets yield markedly different estimates of selection. We show that this is largely due to measurement noise, notably due to differences between studies rather than instrument error or biological variability. We develop an analytical framework that explicitly models noise in expression in the context of the population-genetic model. Estimates of mutation and selection strength in budding yeast produced by this method are robust to the expression data set used and are substantially higher than estimates using a noise-blind approach. We introduce per-gene selection estimates that correlate well with previous scoring systems, such as the codon adaptation index, while now carrying an evolutionary interpretation. On average, selection for codon usage in budding yeast is weak, yet our estimates show that genes range from virtually unselected to average per-codon selection coefficients above the inverse population size. Our analytical framework may be generally useful for distinguishing biological signals from measurement noise in other applications that depend upon measurements of gene expression.

Prognostic factors for malignant pleural mesothelioma

The prognosis for patients with malignant pleural mesothelioma (MPM) is generally regarded as poor, although rare cases of long-term survivors are recognised. Previous prognostic scoring systems have been based on clinical trial populations and are not widely utilised. Population studies consistently confirm non-epithelioid histology, advanced age, and male gender as independent risk factors for poor outcome of MPM. Genetic and immunohistochemical studies continue to provide advances in tumour biology, but no clear validated prognostic factors have been identified to date. Nuclear mitotic and atypia grading systems may provide useful prognostic knowledge; further evaluation is needed. Such biomarkers as soluble mesothelin-related protein and osteopontin provide some prognostic information, though with limitations. The baseline serum neutrophil-to-lymphocyte ratio could also provide prognostic information. Modern metabolic imaging techniques, for example PET/CT, can indicate prognosis by use of baseline total glycolytic volumes (TGV). TGV may also be useful in identifying early responders to systemic chemotherapy treatment. Research to identify clinically useful prognostic factors in MPM remains a priority.

Abstract 205: Optimizing Prediction Scores for Poor Outcome After Intra-arterial Therapy for Anterior Circulation Acute Ischemic Stroke

Background: Intra-arterial therapy (IAT) is an approach to promote recanalization of large artery occlusions (LAO) in acute ischemic stroke (AIS) but is resource intensive. Previous studies evaluated different variables that affect clinical outcome after IAT. To better identify patients who have poor outcomes despite IAT, we compared the performance of previous predictive scoring systems that relied either on clinical or imaging variables in patients undergoing IAT. We then combined imaging and clinical variables to optimize a score that would better predict poor outcome after IAT for AIS. Methods: We studied consecutive AIS patients undergoing IAT at UT-Houston for LAO (MCA or ICA) from 01/03 to 05/11. We collected demographics and clinical variables and analyzed CT head scans using the ASPECTS scores by raters blinded to outcomes. Independent predictors of poor outcome (mRS 4–6) with p values ≤0.1 were evaluated as score variables using sensitivity analysis and logistic regression. Spearman’s correlation and ROC curves were used to evaluate the final score. Houston Intra-arterial Therapy 2 (HIAT2) score ranged from 0–10 with points for: age (≤59=0, 60–79=2, ≥80 years=4), Glucose (&lt;150=0, ≥150=1), NIHSS (≤10=0), 11–20=1, ≥21=2), ASPECTS (8–10=0, ≤ 7=3). The score was created using 75% of the dataset (build group) and tested on the remaining 25% (test group). Previously published prediction scores (HIAT, THRIVE, ASPECTS) were compared against HIAT2. Results: Table 1 shows the clinical features for the 163 patients collected. Fig 1 shows the proportion of poor outcome based on HIAT-2. Patients with HIAT2 ≥ 5 were more likely to have a poor outcome at discharge (OR 6.43, 95% CI 2.75–15.02, p&lt;.001). After adjusting for reperfusion (TICI≥2b, p=.7) and time from symptom onset to recanalization (p=.4), HIAT2 remained a significant independent predictor of poor outcome (OR 5.88, 95% CI 1.96–17.64, p=.02). Fig 2 shows the performance of HIAT2 against THRIVE, HIAT and ASPECTS as well as the validation of HIAT-2. Conclusions: Clinical prediction scores for patients undergoing IAT (HIAT, THRIVE) proved superior to the ASPECTS score. The HIAT2 score, which combines clinical and imaging variables, performed better than all previous scores in predicting the chance of poor outcome after IAT for anterior circulation LAO.

The OMERACT Ultrasound Task Force — Status and Perspectives

This article reports the most recent work of the Outcome Measures in Rheumatology (OMERACT) Ultrasound Task Force, and highlights the future research priorities discussed at the OMERACT 10 meeting. Results of the following studies were presented: (1) intra- and interobserver reliability of ultrasound detecting and scoring synovitis in different joints of patients with rheumatoid arthritis (RA); (2) systematic review of previous ultrasound scoring systems of synovitis in RA; (3) enthesitis systematic review and Delphi definition exercise in spondyloarthritis enthesitis; (4) enthesitis intra- and interobserver reliability exercise; and (5) Delphi definition exercise in hand osteoarthritis, and reliability exercises. Study conclusions were discussed, and a future research agenda was approved, notably further validation of an OMERACT ultrasound global synovitis score (GLOSS) in RA, emphasizing the importance of testing feasibility, predictive value, and added value over standard clinical variables. Future research areas will include validating scoring systems for enthesitis and osteoarthritis, and testing the metric qualities of ultrasound for evaluating tenosynovitis and structural damage in RA.

Risk factors associated with hair loss, ulceration, and swelling at the hock in freestall-housed UK dairy herds

The objective of this study was to identify the risk factors associated with 3 presentations of hock lesions (hair loss, ulceration, and swelling) in freestall-housed lactating cattle. By independent identification and scoring of, and analysis of the factors associated with, hair loss, ulceration, and swelling, the aim was to identify whether risk factors were common to all 3, or differed among the presentations. A cross-sectional study was conducted on 76 herds in the UK during the winter housing period of 2007 to 2008, with a total of 3,691 cows examined for hock lesions. A randomly selected sample of approximately 50 cows in each herd were scored for body condition, lameness, cleanliness, rising behavior, and lesions at the hocks. For all cows, hair loss, ulceration, and swelling were scored separately on 4-point scales, with both left and right hocks scored. Based on a review of the literature, potential risk factors were identified and measured, collected from milk-recording data, or obtained through interviews with the farmers. Risk factors associated with hocks lesions in cattle were examined using data from the 2,982 cows housed in the 63 freestall-housed herds visited. Risk factors for each of the 3 lesion presentations were considered separately in multilevel logistic regression models, with moderate or severe hair loss, any degree of ulceration, and moderate or severe swelling as the outcome variables. Thirty risk factors were identified, none of which were common to all 3 lesion presentations. Five risk factors (locomotion score, number of days of winter housing, mean milk yield, freestall base material, and herd size) were common to both hair loss and ulceration. The stall bedding material was a common risk factor for both hair loss and swelling. A further 8, 5, and 11 risk factors were unique to hair loss, ulceration, and swelling, respectively. The existence of several differential risk factors between the 2 lesion presentations suggests that ulceration may not always be a direct extension of hair loss, as has been implied in previous scoring systems. Of the 12 risk factors associated with swelling, only 1 was common to another lesion presentation, which suggests that swelling may have a different etiology than hair loss and ulceration. The variables associated with the lesions indicate the importance of factors that affect the lying and rising behavior of the animal, including freestall structure and design, and the lying surface.

Application of a Scoring System to the Diagnosis of Acute Appendicitis

Purpose: Acute appendicitis is one of the most common diseases requiring surgical treatment. Delayed diagnosis, which causes complications like perforation of the appendix, abscess formation, or misdiagnosis, leads to unnecessary surgery. Many scoring systems have been suggested for the diagnosis of acute appendicitis. This study aims to evaluate the clinical value of previous scoring systems. Methods: This study was conducted with a total of 270 patients who had visited the National Police Hospital (NPH) Emergency Room for acute abdominal symptoms from January to June 2008. The Alvarado and the Ohmann scores were applied retrospectively based on the patients' records. We found 3 criteria which were relatively objective and clinically meaningful; then, we designed a new 10 points scoring system. Results: The sensitivity and the specificity of the Alvarado scoring system were 83.23% and 64.42%, respectively, whereas those of the Ohmann scoring system were 74.85% and 66.35%, respectively. The sensitivity and the specificity of the NPH scoring system were found to be 78.4% and 68.9%, respectively. The Ohmann scoring system showed a little lower sensitivity, and the NPH scoring system showed a little higher specificity, but the differences were not statistically significant. Conclusion: Our study indicates that the scoring systems considered are not useful diagnostic methods for primary screening and diagnosis of acute appendicitis.

Lower urinary tract symptoms (LUTS) After renal transplant in non-urologic anuric patients

Purpose To report on about LUTS after renal transplant (RTx) in pre-RTx anuric patients without associated urological malformations. Material and Methods Transplanted patients were considered for study if they had been anuric before transplant for at least 6 months, had no urological cause for their renal failure, were neurologically normal, had at least 12 months of follow-up after transplant, and were older than 5 years at follow-up. LUTS were evaluated using a questionnaire adapted from previous scoring systems in order to assess storage and voiding symptoms, bowel habit, and possible effect of symptoms on the patient's quality of life. Results Fourteen patients fulfilled the inclusion criteria. Median age at RTx was 11 (5-21) years, median duration of anuria before RTx 24 (7-46) months, and median post-RTx follow-up 2.7 (1.9-10.2) years. Thirteen cases presented a normal daytime voiding rate (5 to 7 times a day). One case presented episodes of daytime wetting and 4 suffered of urgency. None of the patients complained of a weak stream or about the need for straining during micturition. Ten cases (71%) presented nicturia and 1 occasional episodes of enuresis, nicturia was unrelated to length of follow-up. Twelve cases had a normal bowel habit. Only 1 patient reported that the symptoms affected her quality of life. Conclusions Pre-RTx anuric patients without associated urological malformation can present long-term storage symptoms. Nicturia was present in 71% of our cases irrespective of the length of follow-up.

International Prognostic Scoring System (IPSS) for Waldenström's Macroglobulinemia (WM).

The outcome of patients (pts) with WM, a rare B-cell malignancy, varies widely. The international community has convened a panel to develop a prognostically meaningful staging system for survival after treatment initiation. Toward establishing IPSS, we performed multivariate analyses with bootstrap validation in a series of 587 pts (median age 67, range: 28 to 95, M/F ratio: 1.7) diagnosed between September 1979 and December 2001. Diagnostic and treatment criteria fulfilled recommendations of the 2nd international WM workshop. Front-line treatment was initiated at diagnosis in 69% and 4 mo to 164 mo later in the remaining pts. Criteria for initiation of therapy included cytopenia, constitutional symptoms, organomegaly, hyperviscosity, IgM-related disorders, which pertained to 51%, 44%, 35%, 31% and 13%, respectively. Treatment regimens comprised alkylating agents, fludarabine and rituximab in 369, 195 and 23 subjects, respectively. Baseline parameters included age>65 yr in 57%, Hb≤11.5 g/dL in 65%, platelet count ≤100 x109/L in 9%, granulocyte count ≤1.5 x109/L in 9%, B2M >3 mg/L in 56%, albumin ≤3.5 g/dL in 36% and monoclonal protein >7.0 g/dL in 7%. With a median follow-up of 64 mo (range, 6 mo–182 mo), the median survival after treatment initiation was 87 mo (95%CI 79–103) regardless of the type of therapy applied (p=0.3). Using results of univariate survival analyses, recursive partitioning and martingale residuals analyses, 7 adverse characteristics for inclusion in multivariate analyses were identified during an expert meeting: age >65 yr, platelet count ≤ 100x109/L, B2M >3 mg/L, M-protein >7.0 g/dL, granulocytes ≤1.5 x109/L, Hb ≤11.5 g/dL and albumin ≤3.5 g/dL. The Cox proportional hazard model with stepwise selection selected the first 6 covariates. Bootstrap resampling (500 replicates) validated the selection of the first 4 covariates in at least 80% of the replicates. Selection of at least one of the last 2 covariates in more than 80% of the replicates indicated a correlation between these 2 covariates, and validated the inclusion of Hb only. Using the combination of age, Hb, platelet count, B2M and M-protein, low risk was defined by the presence of ≤1 adverse characteristic except age, high risk by the presence of >2 adverse characteristics; the remaining patients with 2 adverse characteristics or age >65yr had intermediate risk, comprising 27%, 38% and 35% of patients with 5-yr survival rates of 87%, 68% and 36% (p<0.0001), independent of treatment and age. IPSS split each subgroup identified by previous scoring systems (Morel, Merlini, Dhodapkar). Conversely, log-rank test was significant only when the latter prognostic system was assessed in low IPSS-risk pts. Thus the combination of age, B2M, M-protein and blood counts provides simple prognostic model for survival in WM, hopefully serving as an objective basis for initiation of therapy and comparison of treatment results.

Applying Ockham's razor to pancreatitis prognostication: a four-variable predictive model.

We sought to develop a simple yet accurate prognostic scoring system to determine the severity of acute pancreatitis at admission. Because acute pancreatitis has a variable and frequently unpredictable course, identifying individuals at greatest risk for significant, life-threatening complications and stratifying their care appropriately remain a concern. Previous prognostic scoring systems predict severity reasonably well but are limited by time constraints, are unwieldy to use, or both. Data from the international phase III trial of the platelet-activating factor receptor-antagonist Lexipafant were used to develop a 4-variable prognostic model. We then compared the model's ability to predict the severity of acute pancreatitis with the Ranson, Glasgow, and APACHE II systems. The model (BALI), which included BUN >or=25 mg/dL, Age >or=65 years, LDH >or=300 IU/L, and IL-6 >or=300 pg/mL, measured at admission, was similar to the Ranson, Glasgow, and APACHE II systems in its ability to identify increased mortality from acute pancreatitis. The receiver operating characteristic curve area for the BALI model was >or=0.82 +/- 0.03 (mean +/- SD) versus 0.75 +/- 0.04 (Ranson), 0.80 +/- 0.03 (Glasgow), and 0.79 +/- 0.03 (APACHE II). Furthermore, at a prevalence of 15%, the positive and negative predictive values for increased mortality were similar for all systems. The prognostic ability of the BALI 4-variable model was similar to the Ranson, Glasgow, and APACHE II systems but is unique in its simplicity and ability to accurately predict disease severity when used at admission or anytime during the first 48 hours of hospitalization.

Age and hemoglobin level emerge as most important clinical prognostic parameters in patients with osteomyelofibrosis: introduction of a simplified prognostic score

The course of disease in patients with myelofibrosis is highly variable, with survival ranging from few months to many years. Several prognostic models have been established, with the LILLE score now most commonly used. However, in recent series, the latter scoring system repeatedly failed to discriminate patients with intermediate and poor prognosis. In the present study, we re-evaluated previous prognostic models in a separate patient population. We have studied 107 patients with myelofibrosis and correlated clinical parameters at the time of diagnosis with survival. Previous scoring systems were applied to allocate patients to subgroups with distinct prognosis. Most previous scoring systems failed to clearly discriminate an intermediate and poor prognostic group. By contrast, age and hemoglobin level emerged as most significant parameters in multivariate analysis. By allocating one risk point each for hemoglobin < 10 g/dl and age >60 years, three subgroups of patients with distinct prognosis could be identified in our cohort. The overall model and the difference between each of the subgroups were statistically significant in a training group, a test group, the overall cohort of patients and the group of patients with chronic idiopathic myelofibrosis. In summary, we propose an alternative prognostic model for patients with myelofibrosis that reliably identifies three groups of patients with highly different outcome. This is in contrast to previous prognostic scores, in which the poor prognosis group was often very small and/or could not be consistently differentiated from patients with an intermediate prognosis.

A new prognostic model (MPI) for patients with myelofibrosis

6621 Background: Several prognostic models for patients with myelofibrosis with myeloid metaplasia (MMM) have been developed, with the LILLE score1 now most commonly used. The latter and other prev...

Predicting the risk of death from heart failure after coronary artery bypass graft surgery.

Heart failure is the most common cause of death among coronary artery bypass graft (CABG) patients. In addition, most variation in observed mortality rates for CABG surgery is explained by fatal heart failure. The purpose of this study was to develop a clinical risk assessment tool so that clinicians can rapidly and easily assess the risk of fatal heart failure while caring for individual patients. Using prospective data for 8,641 CABG patients, we used logistic regression analysis to predict the risk of fatal heart failure. In multivariate analysis, female sex, prior CABG surgery, ejection fraction <40%, urgent or emergency surgery, advanced age (70-79 yr and >80 yr), peripheral vascular disease, diabetes, dialysis-dependent renal failure and three-vessel coronary disease were significant predictors of fatal postoperative heart failure. A clinical risk assessment tool was developed from this logistic regression model, which had good discriminating characteristics (receiver operating characteristic clinical source = 0.75, 95% confidence interval: 0.71, 0.78). In contrast to previous cardiac surgical scoring systems that predicted total mortality, we developed a clinical risk assessment tool that evaluates risk of fatal heart failure. This distinction is relevant for quality improvement initiatives, because most of the variation in CABG mortality rates is explained by postoperative heart failure.

Wisconsin Cystic Fibrosis Chest Radiograph Scoring System

A new clinical scoring system for patients with cystic fibrosis is needed because of recent advances in diagnosis and treatment which have changed the course of this disease. Chest radiograph scoring is the best objective measure of pulmonary disease for longitudinal studies beginning with infants; however, based on pilot studies, previous scoring systems are not sensitive enough in discriminating between degrees of mild lung disease. Therefore, a new radiographic scoring system was developed with the goal of achieving both sensitivity and reproducibility. This objective was pursued by applying multiattribute utility theory, using a panel of interpreters with expertise in cystic fibrosis radiology, and employing mathematical modeling techniques to weight the various components. The system was developed and validated in three phases including comparison to the Brasfield method of quantitative radiology. The data demonstrate that the new system can be applied reliably and conveniently to generate reproducible scores of pulmonary disease severity. Evaluation of the scores by four independent raters using chest radiographs from 61 patients at an average age of 8.37 years revealed good agreement with a .714 Kendall coefficient of concordance. Assessment of serial changes over time was performed using a group of 176 chest radiographs from 25 patients ranging from 4 weeks to 6 years old; this showed that the Wisconsin system generates score differences that are greater in magnitude with disease progression compared with the Brasfield method. Therefore, the new method is more sensitive to progression of mild disease and should be superior to prior radiographic scoring systems for evaluating therapies designed to modify the early course of disease. The Wisconsin system is designed to be useful in longitudinal clinical studies involving young children with cystic fibrosis and is capable to detecting progression from normality to mild lung disease.

The Development of a Comprehensive System for Classifying Interruptions

This paper describes the development of a new system for classifying interruptions and simultaneous speech, entitled the Interruption Coding System (ICS). The system is organised in the form of a binary flow-chart, and a clear distinction is drawn between interruptive and non-interruptive speech. Interruptions are further divided into single and complex forms, and the flow-chart terminates in a total of 17 empirically discriminable events. In contrast to previous scoring systems, ICS is thus both comprehensive and easy to use.