Assessing the improved discriminatory power of a new biomarker in prognostic models

Although the area under the receiver operating characteristic (AUC) is the most popular measure of the performance of prediction models, it has limitations, especially when it is used to evaluate the added discrimination of a new biomarker in the model. Pencina et al. (2008) proposed two indices, the net reclassification improvement (NRI) and integrated discrimination improvement (IDI), to supplement the improvement in the AUC (IAUC). Their NRI and IDI are based on binary outcomes in case-control settings, which do not involve time-to-event outcome. However, many disease outcomes are time-dependent and the onset time can be censored. Measuring discrimination potential of a prognostic marker without considering time to event can lead to biased estimates. In this dissertation, we have extended the NRI and IDI to survival analysis settings and derived the corresponding sample estimators and asymptotic tests. Simulation studies were conducted to compare the performance of the time-dependent NRI and IDI with Pencina’s NRI and IDI. For illustration, we have applied the proposed method to a breast cancer study.^ Key words: Prognostic model, Discrimination, Time-dependent NRI and IDI ^

Prognostic modelling options for remaining useful life estimation by industry

Over recent years a significant amount of research has been undertaken to develop prognostic models that can be used to predict the remaining useful life of engineering assets. Implementations by industry have only had limited success. By design, models are subject to specific assumptions and approximations, some of which are mathematical, while others relate to practical implementation issues such as the amount of data required to validate and verify a proposed model. Therefore, appropriate model selection for successful practical implementation requires not only a mathematical understanding of each model type, but also an appreciation of how a particular business intends to utilise a model and its outputs. This paper discusses business issues that need to be considered when selecting an appropriate modelling approach for trial. It also presents classification tables and process flow diagrams to assist industry and research personnel select appropriate prognostic models for predicting the remaining useful life of engineering assets within their specific business environment. The paper then explores the strengths and weaknesses of the main prognostics model classes to establish what makes them better suited to certain applications than to others and summarises how each have been applied to engineering prognostics. Consequently, this paper should provide a starting point for young researchers first considering options for remaining useful life prediction. The models described in this paper are Knowledge-based (expert and fuzzy), Life expectancy (stochastic and statistical), Artificial Neural Networks, and Physical models.

MicroRNA Signature Obtained From the Comparison of Aggressive With Indolent Non-Hodgkin Lymphomas: Potential Prognostic Value in Mantle-Cell Lymphoma

Purpose Mantle-cell lymphoma (MCL) has a variable natural history but is incurable with current therapies. MicroRNAs (miRs) are useful in prognostic assessment of cancer. We determined an miR signature defining aggressiveness in B-cell non-Hodgkin lymphomas (NHL) and assessed whether this signature aids in MCL prognosis.MethodsWe assessed miR expression in a training set of 43 NHL cases. The miR signature was validated in 44 additional cases and examined on a training set of 119 MCL cases from four institutions in Canada. miRs significantly associated with overall survival were examined in an independent cohort of 114 MCL cases to determine association with patient outcome. miR expression was combined with current clinical prognostic factors to develop an enhanced prognostic model in patients with MCL.ResultsFourteen miRs were differentially expressed between aggressive and indolent NHL; 11 of 14 were validated in an independent set of NHL (excluding MCL). miR-127-3p and miR-615-3p were significantly associated with overall survival in the MCL training set. Their expression was validated in an independent MCL patient set. In comparison with Ki-67, expression of these miRs was more significantly associated with overall survival among patients with MCL. miR-127-3p was combined with Ki-67 to create a new prognostic model for MCL. A similar model was created with miR-615-3p and Mantle Cell Lymphoma International Prognostic Index scores.ConclusionEleven miRs are differentially expressed between aggressive and indolent NHL. Two novel miRs were associated with overall survival in MCL and were combined with clinical prognostic models to generate novel prognostic data for patients with MCL. (C) 2013 by American Society of Clinical Oncology

Structuring of Ranked Models

Prognostic procedures can be based on ranked linear models. Ranked regression type models are designed on the basis of feature vectors combined with set of relations defined on selected pairs of these vectors. Feature vectors are composed of numerical results of measurements on particular objects or events. Ranked relations defined on selected pairs of feature vectors represent additional knowledge and can reflect experts' opinion about considered objects. Ranked models have the form of linear transformations of feature vectors on a line which preserve a given set of relations in the best manner possible. Ranked models can be designed through the minimization of a special type of convex and piecewise linear (CPL) criterion functions. Some sets of ranked relations cannot be well represented by one ranked model. Decomposition of global model into a family of local ranked models could improve representation. A procedures of ranked models decomposition is described in this paper.

Intelligent prognostics of machinery health utilising suspended condition monitoring data

The ability to forecast machinery failure is vital to reducing maintenance costs, operation downtime and safety hazards. Recent advances in condition monitoring technologies have given rise to a number of prognostic models for forecasting machinery health based on condition data. Although these models have aided the advancement of the discipline, they have made only a limited contribution to developing an effective machinery health prognostic system. The literature review indicates that there is not yet a prognostic model that directly models and fully utilises suspended condition histories (which are very common in practice since organisations rarely allow their assets to run to failure); that effectively integrates population characteristics into prognostics for longer-range prediction in a probabilistic sense; which deduces the non-linear relationship between measured condition data and actual asset health; and which involves minimal assumptions and requirements. This work presents a novel approach to addressing the above-mentioned challenges. The proposed model consists of a feed-forward neural network, the training targets of which are asset survival probabilities estimated using a variation of the Kaplan-Meier estimator and a degradation-based failure probability density estimator. The adapted Kaplan-Meier estimator is able to model the actual survival status of individual failed units and estimate the survival probability of individual suspended units. The degradation-based failure probability density estimator, on the other hand, extracts population characteristics and computes conditional reliability from available condition histories instead of from reliability data. The estimated survival probability and the relevant condition histories are respectively presented as “training target” and “training input” to the neural network. The trained network is capable of estimating the future survival curve of a unit when a series of condition indices are inputted. Although the concept proposed may be applied to the prognosis of various machine components, rolling element bearings were chosen as the research object because rolling element bearing failure is one of the foremost causes of machinery breakdowns. Computer simulated and industry case study data were used to compare the prognostic performance of the proposed model and four control models, namely: two feed-forward neural networks with the same training function and structure as the proposed model, but neglected suspended histories; a time series prediction recurrent neural network; and a traditional Weibull distribution model. The results support the assertion that the proposed model performs better than the other four models and that it produces adaptive prediction outputs with useful representation of survival probabilities. This work presents a compelling concept for non-parametric data-driven prognosis, and for utilising available asset condition information more fully and accurately. It demonstrates that machinery health can indeed be forecasted. The proposed prognostic technique, together with ongoing advances in sensors and data-fusion techniques, and increasingly comprehensive databases of asset condition data, holds the promise for increased asset availability, maintenance cost effectiveness, operational safety and – ultimately – organisation competitiveness.

Utilising Reliability and Condition Monitoring Data for Asset Health Prognosis

The ability to forecast machinery health is vital to reducing maintenance costs, operation downtime and safety hazards. Recent advances in condition monitoring technologies have given rise to a number of prognostic models which attempt to forecast machinery health based on condition data such as vibration measurements. This paper demonstrates how the population characteristics and condition monitoring data (both complete and suspended) of historical items can be integrated for training an intelligent agent to predict asset health multiple steps ahead. The model consists of a feed-forward neural network whose training targets are asset survival probabilities estimated using a variation of the Kaplan–Meier estimator and a degradation-based failure probability density function estimator. The trained network is capable of estimating the future survival probabilities when a series of asset condition readings are inputted. The output survival probabilities collectively form an estimated survival curve. Pump data from a pulp and paper mill were used for model validation and comparison. The results indicate that the proposed model can predict more accurately as well as further ahead than similar models which neglect population characteristics and suspended data. This work presents a compelling concept for longer-range fault prognosis utilising available information more fully and accurately.

On the ecology of cold-water phytoplankton in the Baltic Sea

Increased anthropogenic loading of nitrogen (N) and phosphorus (P) has led to an eutrophication problem in the Baltic Sea, and the spring bloom is a key component in the biological uptake of increased nutrient concentrations. The spring bloom in the Baltic Sea is dominated by both diatoms and dinoflagellates. However, the sedimentation of these groups is different: diatoms tend to sink to the sea floor at the end of the bloom, while dinoflagellates to a large degree are been remineralized in the euphotic zone. Understanding phytoplankton competition and species specific ecological strategies is thus of importance for assessing indirect effects of phytoplankton community composition on eutrophication problems. The main objective of this thesis was to describe some basic physiological and ecological characteristics of the main cold-water diatoms and dinoflagellates in the Baltic Sea. This was achieved by specific studies of: (1) seasonal vertical positioning, (2) dinoflagellate life cycle, (3) mixotrophy, (4) primary production, respiration and growth and (5) diatom silicate uptake, using cultures of common cold-water diatoms: Chaetoceros wighamii, C. gracilis, Pauliella taeniata, Thalassiosira baltica, T. levanderi, Melosira arctica, Diatoma tenuis, Nitzschia frigida, and dinoflagellates: Peridiniella catenata, Woloszynskia halophila and Scrippsiella hangoei. The diatoms had higher primary production capacity and lower respiration rate compared with the dinoflagellates. This difference was reflected in the maximum growth rate, which for the examined diatoms range from 0.6 to 1.2 divisions d-1, compared with 0.2 to 0.3 divisions d-1 for the dinoflagellates. Among diatoms there were species specific differences in light utilization and uptake of silicate, and C. wighamii had the highest carbon assimilation capacity and maximum silicate uptake. The physiological properties of diatoms and dinoflagellates were used in a model of the onset of the spring bloom: for the diatoms the model could predict the initiation of the spring bloom; S. hangoei, on the other hand, could not compete successfully and did not obtain positive growth in the model. The other dinoflagellates did not have higher growth rates or carbon assimilation rates and would thus probably not perform better than S. hangoei in the model. The dinoflagellates do, however, have competitive advantages that were not included in the model: motility and mixotrophy. Previous investigations has revealed that the chain-forming P. catenata performs diurnal vertical migration (DVM), and the results presented here suggest that active positioning in the water column, in addition to DVM, is a key element in this species' life strategy. There was indication of mixotrophy in S. hangoei, as it produced and excreted the enzyme leucine aminopeptidase (LAP). Moreover, there was indirect evidence that W. halophila obtains carbon from other sources than photosynthesis when comparing increase in cell numbers with in situ carbon assimilation rates. The results indicate that mixotrophy is a part of the strategy of vernal dinoflagellates in the Baltic Sea. There were also indications that the seeding of the spring bloom is very important for the dinoflagellates to succeed. In mesocosm experiments dinoflagellates could not compete with diatoms when their initial numbers were low. In conclusion, this thesis has provided new information about the basic physiological and ecological properties of the main cold-water phytoplankton in the Baltic Sea. The main phytoplankton groups, diatoms and dinoflagellates, have different physiological properties, which clearly separate their life strategies. The information presented here could serve as further steps towards better prognostic models of the effects of eutrophication in the Baltic Sea.

Is there a subgroup of long-term evolution among patients with advanced lung cancer?: Hints from the analysis of survival curves from cancer registry data

Background: Recently, with the access of low toxicity biological and targeted therapies, evidence of the existence of a long-term survival subpopulation of cancer patients is appearing. We have studied an unselected population with advanced lung cancer to look for evidence of multimodality in survival distribution, and estimate the proportion of long-term survivors. Methods: We used survival data of 4944 patients with non-small-cell lung cancer (NSCLC) stages IIIb-IV at diagnostic, registered in the National Cancer Registry of Cuba (NCRC) between January 1998 and December 2006. We fitted one-component survival model and two-component mixture models to identify short-and long-term survivors. Bayesian information criterion was used for model selection. Results: For all of the selected parametric distributions the two components model presented the best fit. The population with short-term survival (almost 4 months median survival) represented 64% of patients. The population of long-term survival included 35% of patients, and showed a median survival around 12 months. None of the patients of short-term survival was still alive at month 24, while 10% of the patients of long-term survival died afterwards. Conclusions: There is a subgroup showing long-term evolution among patients with advanced lung cancer. As survival rates continue to improve with the new generation of therapies, prognostic models considering short-and long-term survival subpopulations should be considered in clinical research.

Outpatient treatment of pulmonary embolism.

Pulmonary embolism (PE) is traditionally treated in hospital. Growing evidence from non randomized prospective studies suggests that a substantial proportion of patients with non-massive PE might be safely treated in the outpatient setting using low molecular weight heparins. Based on this evidence, professional societies started to recommend outpatient care for selected patients with non-massive PE. Despite these recommendations, outpatient treatment of non-massive PE appears to be uncommon in clinical practice. The major barriers to PE outpatient care are, firstly, the uncertainty as how to identify low risk patients with PE who are candidates for outpatient care and secondly the lack of high quality evidence from randomized trials demonstrating the safety of PE outpatient care compared to traditional inpatient management. Also, although clinical prognostic models, echocardiography and cardiac biomarkers accurately identify low risk patients with PE in prospective studies, the benefit of risk stratification strategies based on these instruments should be demonstrated in prospective management studies and clinical trials before they can be implemented as decision aids to guide PE outpatient treatment. Before high quality evidence documenting the safety of an outpatient treatment approach is published, outpatient management of non-massive PE cannot be generally recommended.

Fatores de risco para óbito precoce em crianças e em adolescentes submetidos a tranplante hepático eletivo no Hospital de Clínicas de Porto Alegre

No próximo ano, completam-se 40 anos desde a primeira tentativa de transplante hepático (TxH) em seres humanos. Há quase 20 anos, o transplante (Tx) tornou-se uma opção terapêutica real para os pacientes portadores de doença hepática terminal. Atualmente, o TxH é o tratamento de escolha para diversas enfermidades hepáticas, agudas ou crônicas. Dos transplantes realizados na Europa ou nos EUA, em torno de 12% dos pacientes são crianças e adolescentes. No Brasil, 20,9% dos pacientes transplantados de fígado em 2001 tinham até 18 anos de idade e, destes, 60,7% tinham 5 anos ou menos. O objetivo do TxH é a manutenção da vida dos pacientes com doença hepática irreversível, e a principal forma de avaliação de sucesso é a sobrevida após o Tx. A primeira semana que se segue ao TxH, apesar dos excelentes progressos dos últimos anos, continua sendo o período mais crítico. A maioria dos óbitos ou das perdas do enxerto ocorrem nas primeiras semanas, em particular, nos primeiros 7 dias de TxH. Diversos fatores de risco para o resultado do TxH podem ser identificados na literatura, porém há poucos estudos específicos do Tx pediátrico. As crianças pequenas apresentam características particulares que os diferenciam do Tx nos adultos e nas crianças maiores. Com o objetivo de identificar fatores de risco para o óbito nos 7 primeiros dias após os transplantes hepáticos eletivos realizados em 45 crianças e adolescentes no Hospital de Clínicas de Porto Alegre entre março de 1995 e agosto de 2001, foi realizado um estudo de caso-controle. Entre os 6 casos (13,3%) e os 39 controles foram comparadas características relacionadas ao receptor, ao doador e ao procedimento cirúrgico e modelos prognósticos. Das variáveis relacionadas ao receptor, o gênero, o escore Z do peso e da estatura para a idade, a atresia de vias biliares, a cirurgia abdominal prévia, a cirurgia de Kasai, a história de ascite, de peritonite bacteriana espontânea, de hemorragia digestiva e de síndrome hepatopulmonar, a albuminemia, o INR, o tempo de tromboplastina parcial ativada e o fator V não foram associados com o óbito na primeira semana. A mortalidade inicial foi maior nas crianças com menor idade (p=0,0035), peso (p=0,0062) e estatura (p<0,0001), bilirrubinemia total (BT) (p=0,0083) e bilirrubinemia não conjugada (BNC) (p=0,0024) elevadas, e colesterolemia reduzida (p=0,0385). Os receptores menores de 3 anos tiveram um risco 25,5 vezes maior de óbito que as crianças maiores (IC 95%: 1,3–487,7). A chance de óbito após o Tx dos pacientes com BT superior a 20 mg/dL e BNC maior que 6 mg/dL foi 7,8 (IC95%: 1,2–50,1) e 12,7 (IC95%: 1,3–121,7) vezes maior que daqueles com níveis inferiores, respectivamente. Das características relacionadas ao doador e ao Tx, as variáveis gênero, doador de gênero e grupo sangüíneo ABO não idênticos ao do receptor, razão peso do doador/receptor, causa do óbito do doador, enxerto reduzido, tempo em lista de espera e experiência do Programa não foram associados com o óbito nos primeiros 7 dias. Transplantes com enxertos de doadores de idade até 3 anos, ou de peso até 12 Kg representaram risco para o óbito dos receptores 6,8 (IC95%: 1,1–43,5) e 19,3 (IC95%: 1,3–281,6) vezes maior, respectivamente. O tempo de isquemia total foi em média de 2 horas maior nos transplantes dos receptores não sobreviventes (p=0,0316). Os modelos prognósticos Child-Pugh, Rodeck e UNOS não foram preditivos do óbito. Os pacientes classificados como alto risco no modelo de Malatack apresentaram razão de chances para o óbito 18,0 (IC95%: 1,2–262,7) vezes maior que aqueles com baixo risco. A mortalidade na primeira semana foi associada a valores elevados do escore PELD. O risco de óbito foi de 11,3 (IC95%: 1,2–107,0) nas crianças com valor do PELD maior que 10. As crianças pequenas e com maior disfunção hepática apresentaram maior risco de óbito precoce. Doador de pequeno porte e prolongamento do tempo de isquemia também foram associados à mortalidade. Somente os modelos de Malatack e PELD foram preditivos da sobrevida.

Prognosis of patients with HIV-1 infection starting antiretroviral therapy in sub-Saharan Africa: a collaborative analysis of scale-up programmes

Background Prognostic models have been developed for patients infected with HIV-1 who start combination antiretroviral therapy (ART) in high-income countries, but not for patients in sub-Saharan Africa. We developed two prognostic models to estimate the probability of death in patients starting ART in sub-Saharan Africa. Methods We analysed data for adult patients who started ART in four scale-up programmes in Côte d'Ivoire, South Africa, and Malawi from 2004 to 2007. Patients lost to follow-up in the first year were excluded. We used Weibull survival models to construct two prognostic models: one with CD4 cell count, clinical stage, bodyweight, age, and sex (CD4 count model); and one that replaced CD4 cell count with total lymphocyte count and severity of anaemia (total lymphocyte and haemoglobin model), because CD4 cell count is not routinely measured in many African ART programmes. Death from all causes in the first year of ART was the primary outcome. Findings 912 (8·2%) of 11 153 patients died in the first year of ART. 822 patients were lost to follow-up and not included in the main analysis; 10 331 patients were analysed. Mortality was strongly associated with high baseline CD4 cell count (≥200 cells per μL vs <25; adjusted hazard ratio 0·21, 95% CI 0·17–0·27), WHO clinical stage (stages III–IV vs I–II; 3·45, 2·43–4·90), bodyweight (≥60 kg vs <45 kg; 0·23, 0·18–0·30), and anaemia status (none vs severe: 0·27, 0·20–0·36). Other independent risk factors for mortality were low total lymphocyte count, advanced age, and male sex. Probability of death at 1 year ranged from 0·9% (95% CI 0·6–1·4) to 52·5% (43·8–61·7) with the CD4 model, and from 0·9% (0·5–1·4) to 59·6% (48·2–71·4) with the total lymphocyte and haemoglobin model. Both models accurately predict early mortality in patients starting ART in sub-Saharan Africa compared with observed data. Interpretation Prognostic models should be used to counsel patients, plan health services, and predict outcomes for patients with HIV-1 infection in sub-Saharan Africa.

Risk stratification of normotensive patients with acute symptomatic pulmonary embolism

Treatment guidelines recommend strong consideration of thrombolysis in patients with acute symptomatic pulmonary embolism (PE) that present with arterial hypotension or shock because of the high risk of death in this setting. For haemodynamically stable patients with PE, the categorization of risk for subgroups may assist with decision-making regarding PE therapy. Clinical models [e.g. Pulmonary Embolism Severity Index (PESI)] may accurately identify those at low risk of overall death in the first 3 months after the diagnosis of PE, and such patients might benefit from an abbreviated hospital stay or outpatient therapy. Though some evidence suggests that a subset of high-risk normotensive patients with PE may have a reasonable risk to benefit ratio for thrombolytic therapy, single markers of right ventricular dysfunction (e.g. echocardiography, spiral computed tomography, or brain natriuretic peptide testing) and myocardial injury (e.g. cardiac troponin T or I testing) have an insufficient positive predictive value for PE-specific mortality to drive decision-making toward such therapy. Recommendations for outpatient treatment or thrombolytic therapy for patients with PE necessitate further development of prognostic models and conduct of clinical trials that assess various treatment strategies.

A strategy combining imaging and laboratory biomarkers in comparison with a simplified clinical score for risk stratification of patients with acute pulmonary embolism

This study aimed to assess the performance of two prognostic models-the European Society of Cardiology (ESC) model and the simplified Pulmonary Embolism Severity Index (sPESI)-in predicting short-term mortality in patients with pulmonary embolism (PE).