SEARCH · PubMed Health

Results for “Prognostic model”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 181 records · Page 10Linked to original sources

PubMed2005

Proposal of a new prognostic model for hepatocellular carcinoma: an analysis of 403 patients.

BACKGROUND: The prognosis of hepatocellular carcinoma (HCC) is highly dependent on tumour extension and liver function. Recently, two new prognostic scoring systems-the CLIP score, developed by Italian investigators and the BCLC score, developed in Barcelona-have been widely used to assess prognosis in patients presenting with hepatocellular carcinoma. Each system has its own relative limitations. AIMS: To create a new prognostic scoring system which is simple, easy to calculate, and suitable for estimating prognosis during radical treatment of early HCC. METHODS: A total of 403 consecutive patients with HCC treated by percutaneous ablation at the Department of Gastroenterology, University of Tokyo Hospital, between 1990 and 1997 were used as the training sample to identify prognostic factors for our patients and used to develop the Tokyo score. As a testing sample, 203 independent patients who underwent hepatectomy at the Department of Hepato-Biliary-Pancreatic Surgery were studied. Prognostic factors were analysed by univariate and multivariate Cox proportional hazard regression. RESULTS: The Tokyo score consists of four factors: serum albumin, bilirubin, and size and number of tumours. Five year survival was 78.7%, 62.1%, 40.0%, 27.7%, and 14.3% for Tokyo scores 0, 1, 2, 3, and 4-6, respectively. The discriminatory ability of the Tokyo score was internally validated by bootstrap methods. The Tokyo score, CLIP score, and BCLC staging were compared by Akaike information criterion and Harrell's c index among training and testing samples. In the testing sample, the predictive ability of the Tokyo score was equal to CLIP and better than BCLC staging. CONCLUSIONS: The Tokyo score is a simple system which provides good prediction of prognosis for Japanese patients with HCC requiring radical therapy.

Adolescent↗

PubMed2006

MELD score as a prognostic model for listing acute liver failure patients for liver transplantation.

OBJECTIVES: The King's College Hospital (KCH) criteria are widely used for listing patients with acute liver failure (ALF) for liver transplantation (LT). Recent reports have suggested that the Model for End-Stage Liver Disease (MELD) score may be useful in assessing prognosis in ALF (nonparacetamol). This study compares prognostic accuracy of the two systems in patients with paracetamol (POD)-induced ALF treated in this unit. METHODS: Seventy-two patients (average age 38 years; F:M ratio 2:1) admitted from 1994 to 2005 with POD-related ALF were studied. Clinical and biochemical parameters were recorded. The effect of applying a MELD score of greater than 30 as listing criteria for LT was calculated and compared with the KCH criteria. Outcomes were defined as LT, death, or full recovery. RESULTS: Thirty-one patients (43%) recovered with medical therapy, 29 (40%) patients died, and 12 (17%) underwent LT. Sixty five percent of patients had a MELD > 30 and therefore could potentially be listed on admission; however, using KCH criteria only 24% patients were listed immediately. Sensitivity and negative predictive value of MELD was higher then KCH; however, we found KCH to have much higher specificity and positive predictive value. CONCLUSION: MELD has higher sensitivity and negative predictive value for POD-induced ALF than the KCH criteria. However, the high false-positive rate associated with MELD limits its clinical utility. The high negative predictive value of MELD score may allow it to be used in conjunction with KCH criteria to avoid unneeded LT in patients who will likely recover spontaneously.

Adult↗

PubMed2025

Establishment of a prognostic model based on ER stress-related cell death genes and proposing a novel combination therapy in acute myeloid leukemia.

BACKGROUND: Acute myeloid leukemia (AML) is a highly heterogeneous malignancy, presenting significant challenges in accurately predicting patient prognosis. Dysregulation of endoplasmic reticulum (ER) stress and resistance to programmed cell death (PCD) are hallmarks of AML cells. However, the prognostic significance of the interplay between ER stress and cell death pathways in AML remains largely unexplored. METHODS: We analyzed RNA sequencing and clinical data from 887 AML patients across 4 cohorts to develop an ER stress-related cell death index (ERCDI) using 10 machine-learning algorithms with 117 unique combinations. Survival and time-dependent Receiver Operating Characteristic Curve (ROC) analyses were performed to assess the model's efficacy. Clinical characteristics, the tumor immune microenvironment, and drug sensitivity differences between the high- and low-risk groups were also analyzed. The CMap database was used to identify potential therapeutic drugs. In vitro and in vivo experiments, including CCK-8, colony formation, flow cytometry, Transwell assays, and xenograft mouse models, were conducted to evaluate the effects of the target genes and candidate drugs. RESULTS: The ERCDI demonstrated strong prognostic and predictive performance for prognosis in AML patients. Furthermore, the ERCDI effectively predicted immunotherapy and chemotherapy outcomes and was associated with the immune features of the different risk groups. DNA damage-inducible transcript 4 protein (DDIT4), a key gene associated with ERCDI, is related to poor prognosis in AML patients with high expression. Additionally, the knockdown of DDIT4 significantly inhibited AML cell proliferation, induced cell apoptosis, and promoted cell cycle arrest. Chaetocin was subsequently identified as a candidate compound for AML treatment. Subsequent experiments suggested that combining chaetocin and venetoclax is a potentially promising therapeutic strategy for AML. CONCLUSION: The ERCDI provides personalized risk assessment and treatment recommendations for individual AML patients. The combined use of chaetocin and venetoclax can potentially be repurposed for AML therapy.

Humans↗

PubMed1991

Reliability of prognostic models in malignant melanoma. A 10-year follow-up study.

Certain histologic and clinical features of malignant melanoma have been shown to be indicators of prognosis, both collectively and individually. Even though the predictive value of these features is well established, long-term survival is occasionally seen in individuals with multiple poor prognostic factors. To further examine this phenomenon, histologic sections from 53 patients with malignant melanoma excised between the years 1977 and 1980 in whom reliable clinical follow-up data were obtained were evaluated for the presence of features associated with a poor prognosis (thickness greater than 1.7 mm, Clark level greater than or equal to III, vertical growth phase, high mitotic index, marked cytologic atypia, minimal tumor inflammatory infiltrate, presence of regression, presence of plasma cells, male sex, age greater than or equal to 45 years, and axial anatomic location). Sixty-eight percent of the patients survived for greater than or equal to 10 years. Of these, 26% had lesions greater than or equal to 1.7 thick. Multivariate discriminant analysis of all features provide a model that was 76% accurate in predicting outcome over a 10-year period. Although the predictive value of these prognostic variables is generally reliable, there is a significant population of long-term survivors in whom prognosis could not be accurately predicted using these features.

Adult↗

PubMed1997

A prognostic model of survival in surgically resected squamous cell carcinoma of the lung using clinical, pathologic, and biologic markers.

The biologic behavior of tumoral cells plays a significant role in the progression of the neoplasia, because 30 to 35% of patients with Stage I squamous cell carcinoma relapse. The present study was designed to determine whether age, pathologic parameters, DNA ploidy, and a cell proliferation index (the area of nucleolar organizer regions, AgNOR), could be used to predict survival in patients who undergo resection for limited squamous cell carcinoma of the lung. For histopathologic analysis, the parameters of histologic grading, pleural involvement, vascular invasion, and residual disease were considered. The cell proliferation index was evaluated by mitotic index, AgNOR quantification, and DNA ploidy by means of digital image analysis. Fifty-two patients (median age, 60 yr +/- 8.6 yr) were staged according to the TNM staging system. Cox univariate analysis showed that stage, residual disease, vascular invasion, histologic grading, DNA ploidy, and AgNOR were significant predictors of survival. Many of the univariate predictors of cancer death, however were eliminated when Cox multivariate models were computed. The variable that exhibited the most robust predictive value for overall survival was AgNOR. We conclude that measurement of cell proliferation might serve as a prognostic marker in squamous cell carcinoma of the lung.

Adult↗

PubMed2006

A prognostic model for advanced stage nonsmall cell lung cancer. Pooled analysis of North Central Cancer Treatment Group trials.

BACKGROUND: A pooled analysis was performed to examine the impact of pretreatment factors on overall survival (OS) and time to progression (TTP) in patients with advanced-stage nonsmall cell lung cancer (NSCLC) and to construct a prediction equation for OS using pretreatment factors. METHODS: A pooled data set of 1053 patients from 9 North Central Cancer Treatment Group trials was used. Age, gender, Eastern Cooperative Oncology Group performance status (PS), tumor stage (Stage IIIB vs. Stage IV), body mass index (BMI), creatinine level, hemoglobin (Hgb) level, white blood cell (WBC) count, and platelet count were evaluated for their prognostic significance in both univariate and multivariate analyses by using a Cox proportional-hazards model. RESULTS: Patients who had high WBC counts, low Hgb levels, PS >0, BMI < 18.5 kg/m2, and TNM Stage IV disease had significantly worse TTP and OS. Patients who had Stage IV disease with a high WBC count had a particularly poor prognosis. An equation to predict the OS of patients with Stage IV NSCLC based on pretreatment PS, BMI, Hgb level, and WBC count was constructed. CONCLUSIONS: In addition to the widely accepted prognostic factors of PS, BMI, and disease stage, both of the readily available laboratory parameters of Hgb level and WBC count were found to be significant prognostic factors for OS and TTP in patients with advanced-stage NSCLC. The authors' prediction equation can be used to evaluate the benefit of a treatment in Phase II trials by comparing the observed survival of a cohort with its expected survival by using the patients' own prognostic factors in place of comparisons with historic data that may have substantially different baseline patient characteristics.

Aged↗

PubMed2006

Prognostic model of stage II non-small cell lung cancer by a discriminant analysis of the immunohistochemical protein expression.

PURPOSE: We aimed to identify the key proteins that influence the prognosis of non-small cell lung cancer (NSCLC) using protein expression profiles of previously known prognostic markers. METHODS: Thirty-one cases of Stage II NSCLC with 5-year follow-up data were selected. Tissue microarrays (TMA) and immunohistochemistry were used to make protein expression profiles of 18 previously reported immunohistochemical prognostic markers and their value in NSCLC was statistically re-evaluated by a discriminant analysis. RESULTS: For the discriminant analysis using marker protein expression profiles, we selected three significant markers, TTF-1, RCAS1 and c-MET, to evaluate each patient's 5-year survival. The requested discriminant function was V = -1.08754 x (RCAS1 score) - 0.83174 x (TTF1 score) + 0.55204 x (cMET score) + 5.46972, and V = 0 served as a cut-off point. The correctness for evaluating a patient's 5-year survival by a discriminant analysis was 87.1%. CONCLUSIONS: A discriminant analysis is thus considered to be a useful statistical method for analyzing the protein expression profiles obtained by combined TMA and immunohistochemical techniques using archival NSCLC tissues. However, the sample size and selection of the marker protein depending on the histology greatly influence the results of a NSCLC study.

Antigens, Neoplasm↗

PubMed1988

Stage II malignant melanoma: presentation of a prognostic model and an assessment of specific active immunotherapy in 1,273 patients.

The ability to redefine risk factors and to predict prognosis in patients with malignant melanoma at the time they manifest nodal metastasis can be a benefit to the patient emotionally and to the physician therapeutically. A retrospective review of 1,273 patients with stage II malignant melanoma was performed at our institution. The most significant prognostic factors in a simultaneous hazard Cox multivariate analysis, predicting melanoma-related mortality among stage II patients, were the number of positive nodes (P less than 0.0001), age (P = 0.0004), site of the primary lesion (P = 0.0036), disease-free interval (P = 0.016), thickness of the primary lesion (P = 0.017), and sex of the patient (P = 0.0616). We have developed a model for predicting survival of stage II patients, designed for use in the clinic setting. Its application in a computer system makes it accessible and understandable. The most favorable risk group (18% of the population) has actuarial 5- and 10-year survival rates of 58% and 49%, respectively, from the time of the nodal metastasis. The least favorable risk group (7% of the population) has 5- and 10-year survival rates of 15% and 10%, respectively. There are three intermediate risk groups. All groups differ prognostically (P less than 0.04). The principal adjuvant therapy offered to these patients was specific active immunotherapy, which appears to have a 10-20% survival benefit in stage II patients with greater than one positive node, when compared with institutional controls. The apparent survival benefit of the immunotherapy supports continued clinical investigation of its therapeutic potential.

Actuarial Analysis↗

PubMed1996

Recurrence and death in non-small cell lung carcinomas: a prognostic model using pathological parameters, microvessel count, and gene protein products.

The 5-year survival rate of non-small cell lung carcinoma (NSCLC) has only marginally improved during the past two decades, despite advances in surgery and chemoradiotherapy. Major efforts are currently directed toward biological characterization of these tumors to define biomarkers able to add further prognostic information, thus improving new therapeutic protocols. We analyzed the predictive relevance of the microvessel count (MC), bcl-2 and p53 proteins, proliferative activity, and usual postsurgical parameters on recurrence and overall survival in a series of 70 patients with NSCLC. The expression of biological parameters (p53, bcl-2, proliferative activity, and MC) was detected using immunohistochemistry on paraffin-embedded and frozen sections from the tumors treated with surgical resection alone until relapse. In the univariate analysis, the histotype, tumor status, node status, p53, bcl-2, and MC have been shown to significantly affect progression and death. In the multiple logistic regression analysis, the MC (P < 0.000001), tumor status (P < 0.005), and node status (P < 0.0002) influenced the overall survival while prediction of relapse was strongly revealed by tumor status (P < 0.005), nodal metastatic involvement (P < 0.000001), and the assessment of the vascular count (P < 0.0004). These data have allowed the creation of a multivariate model which may add more information on risk of recurrence and death in patients with NSCLC and can form the basis for future randomized clinical trials.

Aged↗

PubMed1995

Prognostic models for subgroups of melanoma patients from the Scottish Melanoma Group database 1979-86, and their subsequent validation.

For the past 20 years thickness of the primary tumour has been accepted as the most important guide to prognosis for patients with primary cutaneous malignant melanoma. The changing epidemiology of melanoma with an increasing number of patients with thin tumours has necessitated a reappraisal of this, with particular reference to interactions among tumour thickness, the patients' sex and the presence or absence of ulceration of the primary tumour. All primary cutaneous malignant melanomas diagnosed in Scotland between 1979 and 1986 were used as the test group (1978 patients). The proportional hazards model was used on all potential risk factors in the database and their two-way interactions, and the resulting models based on stepwise procedures were subsequently validated on 289 melanoma patients first diagnosed in 1987 in the same geographic area. Four distinct subgroups of males and females with ulcerated or non-ulcerated lesions were identified. For females with ulcerated lesions, tumour thickness, mitotic count and anatomical site of primary all gave valuable prognostic information, whereas for females with non-ulcerated lesions only tumour thickness was of prognostic value. For males with ulcerated lesions, level of invasion was the only prognostic guide, while for males with non-ulcerated lesions both tumour thickness and level of invasion contributed significantly to prediction of prognosis. Prognosis markedly different across subgroups of the melanoma population, even to the extent that essential prognostic factors are not the same in the distinct subgroups. Verification of these prognostic guides derived from 1979-86 patients has been achieved for all patients diagnosed with melanoma in 1987 from the same geographic area. These data will therefore be useful aids for clinicians managing patients.

Adult↗

PubMed2005

A prognostic model for temporal courses that combines temporal abstraction and case-based reasoning.

Since clinical management of patients and clinical research are essentially time-oriented endeavours, reasoning about time has become a hot topic in medical informatics. Here we present a method for prognosis of temporal courses, which combines temporal abstractions with case-based reasoning. It is useful for application domains where neither well-known standards, nor known periodicity, nor a complete domain theory exist. We have used our method in two prognostic applications. The first one deals with prognosis of the kidney function for intensive care patients. The idea is to elicit impairments on time, especially to warn against threatening kidney failures. Our second application deals with a completely different domain, namely geographical medicine. Its intention is to compute early warnings against approaching infectious diseases, which are characterised by irregular cyclic occurrences. So far, we have applied our program on influenza and bronchitis. In this paper, we focus on influenza forecast and show first experimental results.

Computer Simulation↗

PubMed2005

Stent placement or brachytherapy for palliation of dysphagia from esophageal cancer: a prognostic model to guide treatment selection.

BACKGROUND: Brachytherapy was found to be preferable to metal stent placement for the palliation of dysphagia because of inoperable esophageal cancer in the randomized SIREC trial. The benefit of brachytherapy, however, only occurred after a relatively long survival. The objective is to develop a model that distinguishes patients with a poor prognosis from those with a relatively good prognosis. METHODS: Survival was analyzed with Cox regression analysis. Dysphagia-adjusted survival (alive with no or mild dysphagia) was studied with Kaplan-Meier analysis. Patient data is from the multicenter, randomized, controlled trial (SIREC, n = 209) and a consecutive series (n = 396). Patients received a stent or single-dose brachytherapy. RESULTS: Significant prognostic factors for survival included tumor length, World Health Organization performance score, and the presence of metastases (multivariable p < 0.001). A simple score, which also included age and gender, could satisfactorily separate patients with a poor, intermediate, and relatively good prognosis within the SIREC trial. For the poor prognosis group, the difference in dysphagia-adjusted survival was 23 days in favor of stent placement compared with brachytherapy (77 vs. 54 days, p = 0.16). For the other prognostic groups, brachytherapy resulted in a better dysphagia-adjusted survival. CONCLUSIONS: A simple prognostic score may help to identify patients with a poor prognosis in whom stent placement is at least equivalent to brachytherapy. If further validated, this score can provide an evidence-based tool for the selection of palliative treatment in esophageal cancer patients.

Aged↗

PubMed2026

AI-driven diagnostic and prognostic models for metabolic dysfunction-associated steatotic liver disease: insights from clinical, imaging, and multi-omics studies-a scoping review.

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as non-alcoholic fatty liver disease (NAFLD), is the most common chronic liver disease around the world, affecting 33.6% of the adult population (95% CI: 28.1%-39.5%; I 2 = 99.9%), or roughly one in three. The extent of the liver damage is variable, from simple steatosis to metabolic dysfunction-associated steatohepatitis (MASH, formerly NASH), cirrhosis and hepatocellular carcinoma (HCC). Early diagnosis is essential to prevent serious liver damage. Traditional diagnostic techniques such as liver biopsy, imaging, and biomarker testing are all invasive, costly, reduced sensitive to early-stage disease, and they also have variability among observers. Modern diagnostic and prognostic approaches based on the principles of Artificial Intelligence (AI) and specifically on machine learning (ML) and deep learning (DL) have enabled multimodal approaches integrating clinical, imaging and molecular data. This scoping review conducted per PRISMA-ScR guidelines, synthesizes findings from 73 studies (search window 2020-2026) across three dimensions: clinical data driven models, imaging-based classifiers (ultrasound, CT and MRI), and multi-omics (genomics, transcriptomics and proteomics) techniques. Moreover, emergence of models such as U-Net and LiverNet 2.x, classification models like DeepLiverNet and BiLSTM models, as well as transformer frameworks and the identification of biomarkers models are also described. This study also investigates challenges such as data heterogeneity, data interpretability, fairness and real-world clinical application. Finally, important areas of research opportunities and future directions are highlighted to present a developing clinically applicable, explainable and ethical AI solutions to manage MASLD.

MASLD↗

PubMed2005

VTE Risk assessment - a prognostic Model: BATER Cohort Study of young women.

BACKGROUND: Community-based cohort studies are not available that evaluated the predictive power of both clinical and genetic risk factors for venous thromboembolism (VTE). There is, however, clinical need to forecast the likelihood of future occurrence of VTE, at least qualitatively, to support decisions about intensity of diagnostic or preventive measures. MATERIALS AND METHODS: A 10-year observation period of the Bavarian Thromboembolic Risk (BATER) study, a cohort study of 4337 women (18-55 years), was used to develop a predictive model of VTE based on clinical and genetic variables at baseline (1993). The objective was to prepare a probabilistic scheme that discriminates women with virtually no VTE risk from those at higher levels of absolute VTE risk in the foreseeable future. A multivariate analysis determined which variables at baseline were the best predictors of a future VTE event, provided a ranking according to the predictive power, and permitted to design a simple graphic scheme to assess the individual VTE risk using five predictor variables. RESULTS: Thirty-four new confirmed VTEs occurred during the observation period of over 32,000 women-years (WYs). A model was developed mainly based on clinical information (personal history of previous VTE and family history of VTE, age, BMI) and one composite genetic risk markers (combining Factor V Leiden and Prothrombin G20210A Mutation). Four levels of increasing VTE risk were arbitrarily defined to map the prevalence in the study population: No/low risk of VTE (61.3%), moderate risk (21.1%), high risk (6.0%), very high risk of future VTE (0.9%). In 10.6% of the population the risk assessment was not possible due to lacking VTE cases. The average incidence rates for VTE in these four levels were: 4.1, 12.3, 47.2, and 170.5 per 104 WYs for no, moderate, high, and very high risk, respectively. CONCLUSION: Our prognostic tool - containing clinical information (and if available also genetic data) - seems to be worthwhile testing in medical practice in order to confirm or refute the positive findings of this study. Our cohort study will be continued to include more VTE cases and to increase predictive value of the model.

Journal Article↗

PubMed2005

Population-based validation of the prognostic model ADJUVANT! for early breast cancer.

PURPOSE: Adjuvant! (www.adjuvantonline.com) is a web-based tool that predicts 10-year breast cancer outcomes with and without adjuvant systemic therapy, but it has not been independently validated. METHODS: Using the British Columbia Breast Cancer Outcomes Unit (BCOU) database, demographic, pathologic, staging, and treatment data on 4,083 women diagnosed between 1989 and 1993 in British Columbia with T1-2, N0-1, M0 breast cancer were abstracted and entered into Adjuvant! to calculate predicted 10-year overall survival (OS), breast cancer-specific survival (BCSS), and event-free survival (EFS) for each patient. Individual BCOU observed outcomes at 10 years were independently determined. Predicted and observed outcomes were compared. RESULTS: Across all 4,083 patients, 10-year predicted and observed outcomes were within 1% for OS, BCSS, and EFS (all P > .05). Predicted and observed outcomes were within 2% for most demographic, pathologic, and treatment-defined subgroups. Adjuvant! overestimated OS, BCSS, and EFS in women younger than age 35 years (predicted-observed = 8.6%, 9.6%, and 13.6%, respectively; all P < .001) or with lymphatic or vascular invasion (LVI; predicted-observed = 3.6%, 3.8%, and 4.2%, respectively; all P < .05); these two prognostic factors were not automatically incorporated within the Adjuvant! algorithm. After adjusting for the distribution of LVI, using the prognostic factor impact calculator in Adjuvant!, 10-year predicted and observed outcomes were no longer significantly different. CONCLUSION: Adjuvant! performed reliably. Patients younger than age 35 or with known additional adverse prognostic factors such as LVI require adjustment of risks to derive reliable predictions of prognosis without adjuvant systemic therapy and the absolute benefits of adjuvant systemic therapy.

Adult↗

PubMed1999

Modelling prognostic power of cardiac tests using rough sets.

Rough sets (Pawlak Z. Rough Sets: Theoretical Aspects of Reasoning about Data, Dordrecht: Kluwer Academic Publishers, 1991) is a relatively new approach to representing and reasoning with incomplete and uncertain knowledge. This article introduces the basic concepts of rough sets and Boolean reasoning (Brown FM. Boolean Reasoning: The Logic of Boolean Equations, Dordrecht: Kluwer Academic Publishers, 1990). A rough set framework is then set up to investigate the prognosis of cardiac events in a set of patients with chest pain that was earlier studied by Geleijnse et al. (J Am Coll Cardiol 1996;28(2):447-454). That study used logistic regression to find that the single most important independent predictor for future hard cardiac events (cardiac death or non-fatal myocardial infarction) was an abnormal scintigraphic scan pattern. However, performing a scintigraphic scan is a relatively expensive procedure, and may for some patients not really be fully necessary as knowledge of the outcome of the scan may be redundant with respect to making a prognosis. Using an approach based on rough sets, this paper explores how a patient group in need of a scintigraphic scan can be identified for subsequent modelling. Identification of such patients may potentially contribute to lowering the cost of medical care and to improving its quality since, virtually without loss of information, fewer patients may be referred for this procedure.

Adolescent↗

PubMed1996

Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.

Multivariable regression models are powerful tools that are used frequently in studies of clinical outcomes. These models can use a mixture of categorical and continuous variables and can handle partially observed (censored) responses. However, uncritical application of modelling techniques can result in models that poorly fit the dataset at hand, or, even more likely, inaccurately predict outcomes on new subjects. One must know how to measure qualities of a model's fit in order to avoid poorly fitted or overfitted models. Measurement of predictive accuracy can be difficult for survival time data in the presence of censoring. We discuss an easily interpretable index of predictive discrimination as well as methods for assessing calibration of predicted survival probabilities. Both types of predictive accuracy should be unbiasedly validated using bootstrapping or cross-validation, before using predictions in a new data series. We discuss some of the hazards of poorly fitted and overfitted regression models and present one modelling strategy that avoids many of the problems discussed. The methods described are applicable to all regression models, but are particularly needed for binary, ordinal, and time-to-event outcomes. Methods are illustrated with a survival analysis in prostate cancer using Cox regression.

Clinical Trials as Topic↗

PubMed2003

A prognostic model and staging for metastatic uveal melanoma.

BACKGROUND: To identify factors that independently contribute to overall survival in Stage IVB uveal melanoma and to subcategorize by prognosis. METHODS: Data of 91 consecutive patients who died of metastatic uveal melanoma in 1985-2000 were analyzed by Kaplan-Meier and Cox regression analysis. Main covariates were participation in annual review, symptoms, Karnofsky index, metastatic burden, liver function tests, and age. Time on chemotherapy was modeled as a confounder. A working formulation for staging patients according to predicted survival was designed. RESULTS: Of the 91 patients, 85% underwent annual liver imaging and function tests, 63% were asymptomatic, and 73% received chemotherapy. The median survival period was 8.4 months (95% confidence interval [CI], 6.3-11.8). Karnofsky index, largest dimension of the largest metastasis, metastatic burden, serum transaminase, lactate dehydrogenase, and alkaline phosphatase (AP) levels, and time on chemotherapy were strongly (P < 0.001) associated with survival. Symptoms (P = 0.031) and regular review (P = 0.081) were weakly associated with survival. Karnofsky index (P = 0.013), the largest dimension of the largest metastasis (P = 0.003), and serum AP level (P = 0.042) retained independent significance, adjusting for time on chemotherapy. Predicted median survival calculated for relevant covariate combinations was divided into three periods (> or =12 months vs. 6-11 months vs. < 6 months). Observed median survival for Stage IVBa was 14.9 months (95% CI, 11.7-21.3), for Stage IVBb 8.9 months (95% CI, 2.7-13.7), and for Stage IVBc 2.0 months (95% CI, 1.0-3.7). CONCLUSION: The model and working formulation for categorization can be tested as an aid in patient counseling and as a tool in design and analysis of clinical trials.

Adult↗