Artificial neural networks applied to outcome prediction for colorectal cancer patients in separate institutions

Bottaci, Leonardo; Drew, Philip; Hartley, John E.; Hadfield, M. B.; Farouk, Ridzuan; Lee, Peter Wr; Macintyre, I. M. C.; Duthie, G. S.; Monson, J. R. T.

doi:10.1016/s0140-6736(96)11196-x

Cited by 182 publications

(119 citation statements)

References 22 publications

Supporting

Mentioning

113

Contrasting

Unclassified

Order By: Relevance

“…As an alternative to regression, researchers have applied various machine learning methods, especially ANNs and support vector machines (SVMs), and the initial results have been promising. [34][35][36][37] There are also various tree-based methods that offer an appealing output that, unlike ANNs and SVMs, shows the relations between predictors. Of these, we consider random forests, as we doubted that our data would benefit from the extra complexity inherent in boosted trees.…”

Section: Machine Learning Methodsmentioning

confidence: 99%

Can valid and practical risk-prediction or casemix adjustment models, including adjustment for comorbidity, be generated from English hospital administrative data (Hospital Episode Statistics)? A national observational study

Bottle

Gaudoin

Goudie

et al. 2014

Health Services and Delivery Research

View full text Add to dashboard Cite

Background NHS hospitals collect a wealth of administrative data covering accident and emergency (A&E) department attendances, inpatient and day case activity, and outpatient appointments. Such data are increasingly being used to compare units and services, but adjusting for risk is difficult. Objectives To derive robust risk-adjustment models for various patient groups, including those admitted for heart failure (HF), acute myocardial infarction, colorectal and orthopaedic surgery, and outcomes adjusting for available patient factors such as comorbidity, using England’s Hospital Episode Statistics (HES) data. To assess if more sophisticated statistical methods based on machine learning such as artificial neural networks (ANNs) outperform traditional logistic regression (LR) for risk prediction. To update and assess for the NHS the Charlson index for comorbidity. To assess the usefulness of outpatient data for these models. Main outcome measures Mortality, readmission, return to theatre, outpatient non-attendance. For HF patients we considered various readmission measures such as diagnosis-specific and total within a year. Methods We systematically reviewed studies comparing two or more comorbidity indices. Logistic regression, ANNs, support vector machines and random forests were compared for mortality and readmission. Models were assessed using discrimination and calibration statistics. Competing risks proportional hazards regression and various count models were used for future admissions and bed-days. Results Our systematic review and empirical analysis suggested that for general purposes comorbidity is currently best described by the set of 30 Elixhauser comorbidities plus dementia. Model discrimination was often high for mortality and poor, or at best moderate, for other outcomes, for example c = 0.62 for readmission and c = 0.73 for death following stroke. Calibration was often good for procedure groups but poorer for diagnosis groups, with overprediction of low risk a common cause. The machine learning methods we investigated offered little beyond LR for their greater complexity and implementation difficulties. For HF, some patient-level predictors differed by primary diagnosis of readmission but not by length of follow-up. Prior non-attendance at outpatient appointments was a useful, strong predictor of readmission. Hospital-level readmission rates for HF did not correlate with readmission rates for non-HF; hospital performance on national audit process measures largely correlated only with HF readmission rates. Conclusions Many practical risk-prediction or casemix adjustment models can be generated from HES data using LR, though an extra step is often required for accurate calibration. Including outpatient data in readmission models is useful. The three machine learning methods we assessed added little with these data. Readmission rates for HF patients should be divided by diagnosis on readmission when used for quality improvement. Future work As HES data continue to develop and improve in scop...

show abstract

Section: Machine Learning Methodsmentioning

confidence: 99%

Can valid and practical risk-prediction or casemix adjustment models, including adjustment for comorbidity, be generated from English hospital administrative data (Hospital Episode Statistics)? A national observational study

Bottle

Gaudoin

Goudie

et al. 2014

Health Services and Delivery Research

View full text Add to dashboard Cite

show abstract

“…10 However, the size of the test set is not the whole story, as there needs to be sufficient cases that survive and sufficient that die. The study of Bottaci et al 41,67 has gained considerable publicity, yet is based on the apparent success in predicting the death of just 7 out of 92 patients, and a higher accuracy (the headline measure used) would have been obtained by predicting survival for all the patients!…”

Section: Fitting Neural Networkmentioning

confidence: 99%

Neural networks as statistical methods in survival analysis

Ripley¹,

Ripley²

2001

Clinical Applications of Artificial Neural Networks

View full text Add to dashboard Cite

“…There have been a large number of reports on the use of ANN for clinical diagnosis (18), staging and prognosis (19), and management of cancers (20). In a previous study, an ANN-based composite diagnostic index derived using a panel of four serum markers, CA 125II, CA 72-4, CA 15-3, and lipid-associated sialic acid (LASA), was evaluated for it ability to discriminate malignant from benign pelvic masses (21).…”

Section: Introductionmentioning

confidence: 99%

Combining multiple serum tumor markers improves detection of stage I epithelial ovarian cancer

Zhang¹,

Yu²,

Xu³

et al. 2007

Gynecologic Oncology

View full text Add to dashboard Cite

Objective-Currently available tumor markers for ovarian cancer are still inadequate in both sensitivity and specificity to be used for population-based screening. Artificial neural network (ANN) as a modeling tool has demonstrated its ability to assimilate information from multiple sources and to detect subtle and complex patterns. In this paper, an ANN model was evaluated for its performance in detecting early stage epithelial ovarian cancer using multiple serum markers.Methods-Serum specimens collected at four institutions in the US, the Netherlands, and the United Kingdom were analyzed for CA125II, CA72-4, CA15-3, and macrophage colony stimulating factor (M-CSF). The four tumor marker values were then used as inputs to an ANN derived using a training set from 100 apparently healthy women, 45 women with benign conditions arising from the ovary, and 55 invasive epithelial ovarian cancer patients (including 27 stage I/II cases). A separate validation set from 27 apparently healthy women, 56 women with benign conditions, and 35 women with various types of malignant pelvic masses was used to monitor the ANN's performance during training. An independent test dataset from 98 apparently healthy women and 52 early stage epithelial ovarian cancer patients (38 stage I and 4 stage II invasive cases and 10 stage I borderline ovarian tumor cases) was used to evaluate the ANN.Results-ROC analysis confirmed the overall superiority of the ANN-derived composite index over CA125II alone (p=0.0333). At a fixed specificity of 98%, the sensitivities for ANN and CA125II alone were 71% (37/52) and 46% (24/52) (p=0.047), respectively, for detecting early stage epithelial ovarian cancer, and 71% (30/42) and 43% (18/42) (p=0.040), respectively, for detecting invasive early stage epithelial ovarian cancer.Conclusions-The combined use of multiple tumor markers through an ANN improves the overall accuracy to discern healthy women from patients with early stage ovarian cancer. Analysis of multiple markers with an ANN may be a better choice than the use of CA125II alone in a two-step

show abstract

Artificial neural networks applied to outcome prediction for colorectal cancer patients in separate institutions

Cited by 182 publications

References 22 publications

Can valid and practical risk-prediction or casemix adjustment models, including adjustment for comorbidity, be generated from English hospital administrative data (Hospital Episode Statistics)? A national observational study

Can valid and practical risk-prediction or casemix adjustment models, including adjustment for comorbidity, be generated from English hospital administrative data (Hospital Episode Statistics)? A national observational study

Neural networks as statistical methods in survival analysis

Combining multiple serum tumor markers improves detection of stage I epithelial ovarian cancer

Contact Info

Product

Resources

About