Intelligent Data Engineering and Automated Learning – IDEAL 2019, 2019
This case study benchmarks a range of statistical and machine learning methods relevant to comput... more This case study benchmarks a range of statistical and machine learning methods relevant to computer-based decision support in clinical medicine, focusing on the diagnosis of knee osteoarthritis at first presentation. The methods, comprising logistic regression, Multilayer Perceptron (MLP), Chi-square Automatic Interaction Detector (CHAID) and Classification and Regression Trees (CART), are applied to a public domain database, the Osteoarthritis Initiative (OAI), a 10 year longitudinal study starting in 2002 (n = 4,796). In this real-world application, it is shown that logistic regression is comparable with the neural networks and decision trees for discrimination of positive diagnosis on this data set. This is likely because of weak non-linearities among high levels of noise. After comparing the explanations provided by the different methods, it is concluded that the interpretability of the risk score index provided by logistic regression is expressed in a form that most naturally integrates with clinical reasoning. The reason for this is that it gives a statistical assessment of the weight of evidence for making the diagnosis, so providing a direction for future research to improve explanation of generic non-linear models.
4 Department of Applied Mathematics, Liverpool John Moores University, James Parsons 5 Building, ... more 4 Department of Applied Mathematics, Liverpool John Moores University, James Parsons 5 Building, Byrom Street, Liverpool, L3 3AF, UK. 6 EPSRC Liverpool Centre for Mathematics in Healthcare, Department of Mathematical 7 Sciences, University of Liverpool, Liverpool, L69 7ZL, UK. 8 MRC Centre for Inflammation Research, Queens Medical Research Institute, University of 9 Edinburgh, 47 Little France Crescent, Edinburgh, EH16 4TJ, UK. 10 To whom correspondence should be addressed at Department of Mathematical Sciences, 11 University of Liverpool, Peach Street, Liverpool, L69 7ZL, UK. Tel: +44 151 794 4049. E12 mail: [email protected]. 13 Both authors contributed equally. 14
The most commonly prescribed painkiller worldwide, paracetamol (acetaminophen, APAP) is also the ... more The most commonly prescribed painkiller worldwide, paracetamol (acetaminophen, APAP) is also the predominant cause of acute liver failure (ALF), and therefore paracetamol-induced liver toxicity remains an important clinical problem. The standard clinical treatment framework for paracetamol overdose currently allows for antidote therapy decisions to be made based on a nomogram treatment line. This treatment threshold is lowered for patients adjudged to be highly susceptible to liver injury due to risk factors such as anorexia nervosa or bulimia. Additionally, both the original and adjusted clinical frameworks are highly dependent on knowledge from the patient regarding time since ingestion and initial dose amount, both of which are often highly unpredictable. We have recently developed a pre-clinical framework for predicting time since ingestion, initial dose amount and subsequent probability of liver injury based on novel biomarker concentrations. Here, we use identifiability analysis as a tool to increase confidence in our model parameter estimates and extend the framework to make predictions for both healthy and high-risk populations. Through pharmacokineticpharmacodynamic model refinement, we identify thresholds that determine whether necrosis or apoptosis is the dominant form of cell death, which can be essential for effective ALF interventions. Using a single blood test, rather than the multiple tests required in the current clinical frameworks, our model provides overdose identification information applicable for healthy and high-risk individuals as well as quantitative measures of estimated liver injury probability.
Introduction: Withdrawal assessment in critically ill children is complicated by the reliance on ... more Introduction: Withdrawal assessment in critically ill children is complicated by the reliance on non-specific behaviours and compounded when the child's typical behaviours are unknown. The existing approach to withdrawal assessment assumes that nurses elicit the parents' view of the child's behaviours. Objective and research methodology: This qualitative study explored parents' perspectives of their child's withdrawal and preferences for involvement and participation in withdrawal assessment. Parents of eleven children were interviewed after their child had completed sedation weaning during recovery from critical illness. Data were analysed using thematic analysis. Setting: A large children's hospital in the Northwest of England. Findings: Parents experienced varying degrees of partnership in the context of withdrawal assessment and identified information deficits which contributed to their distress of parenting a child with withdrawal syndrome. Most parents were eager to participate in withdrawal assessment and reported instances where their knowledge enabled a personalised interpretation of their child's behaviours. Reflecting on the reciprocal nature of the information deficits resulted in the development of a model for nurse-parent collaboration in withdrawal assessment. Conclusion: Facilitating nurse-parent collaboration in withdrawal assessment may have reciprocal benefits by moderating parental stress and aiding the assessment and management of withdrawal syndrome.
CPT: pharmacometrics & systems pharmacology, Jan 18, 2018
Paracetamol (acetaminophen (APAP)) is one of the most commonly used analgesics in the United King... more Paracetamol (acetaminophen (APAP)) is one of the most commonly used analgesics in the United Kingdom and the United States. However, exceeding the maximum recommended dose can cause serious liver injury and even death. Promising APAP toxicity biomarkers are thought to add value to those used currently and clarification of the functional relationships between these biomarkers and liver injury would aid clinical implementation of an improved APAP toxicity identification framework. The framework currently used to define an APAP overdose is highly dependent upon time since ingestion and initial dose; information that is often highly unpredictable. A pharmacokinetic/pharmacodynamic (PK/PD) APAP model has been built in order to understand the relationships between a panel of biomarkers and APAP dose. Visualization and statistical tools have been used to predict initial APAP dose and time since administration. Additionally, logistic regression analysis has been applied to histology data to...
The aim of the study was to evaluate registered children's nurses' approaches to the assessment a... more The aim of the study was to evaluate registered children's nurses' approaches to the assessment and management of withdrawal syndrome in children. Background Assessment of withdrawal syndrome is undertaken following critical illness when the child's condition may be unstable with competing differential diagnoses. Assessment tools aim to standardise and improve recognition of withdrawal syndrome. Making the right decisions in complex clinical situations requires a degree of mental effort and it is not known how nurses make decisions when undertaking withdrawal assessments. Design Cognitive interviews with clinical vignettes. Methods Interviews were undertaken with 12 nurses to explore the cognitive processes they utilised when assessing children using the Sedation Withdrawal Score (SWS) tool. Interviews took place in Autumn 2013. Findings Each stage of decision-making-noticing, interpreting and responding-presented cognitive challenges for nurses. When defining withdrawal behaviours nurses tended to blur the boundaries between SWS signs. Challenges in interpreting behaviours arose from not knowing if the patient's behaviour was a result of withdrawal or other co-morbidities.
Journal of Strength and Conditioning Research, 2017
The purpose of the present study was to provide a detailed analysis of the physical demands of co... more The purpose of the present study was to provide a detailed analysis of the physical demands of competitive international female soccer match-play. A total of 148 individual match observations were undertaken on 107 outfield players competing in competitive international matches during the 2011-2012 and 2012-2013 seasons, using a computerized tracking system (Prozone Sports Ltd., Leeds, England). Total distance (TD) and total high-speed running distances (THSR) were influenced by playing position, with central midfielders (CM) completing the highest (10985±706 m and 2882±500 m) and central defenders (CD) the lowest (9489±562 m and 1901±268 m) distances, respectively. Greater total very highspeed running (TVHSR) distances were completed when a team was without (399±143 m) compared to with (313±210 m) possession of the ball. The majority of sprints were over short distances with 76 % and 95 % being less than 5 m and 10 m, respectively. Between half reductions in physical performance were present for all variables, independent of playing position. The current study provides novel findings regarding the physical demands of different playing positions in competitive international female match-play and provides important insights for physical coaches preparing elite female players for competition.
The version presented here may differ from the published version or from the version of the recor... more The version presented here may differ from the published version or from the version of the record. Please see the repository URL above for details on accessing the published version and note that access may require a subscription.
The Fisher information metric defines a Riemannian space where distances reflect similarity with ... more The Fisher information metric defines a Riemannian space where distances reflect similarity with respect to a given probability distribution. This metric can be used during the process of building a relational network, resulting in a structure that is informed about the similarity criterion. Furthermore, the relational nature of this network allows for an intuitive interpretation of the data through their location within the network and the way it relates to the most representative cases or prototypes.
BackgroundThe relationship between well-being and mental ill health is complex; people may experi... more BackgroundThe relationship between well-being and mental ill health is complex; people may experience very low levels of well-being even in the absence of overt mental health problems.AimsThis study tested the hypothesis that anxiety, depression and well-being have different causal determinants and psychological mediating mechanisms.MethodThe influence of causal and mediating factors on anxiety, depression and well-being were investigated in a cross-sectional online questionnaire survey hosted on a UK national broadcasting website.ResultsMultivariate conditional independence analysis of data from 27 397 participants revealed different association pathways for the two constructs. Anxiety and depression were associated with negative life events mediated by rumination; low levels of subjective well-being were associated with material deprivation and social isolation, mediated by adaptive coping style.ConclusionsOur findings support the ‘two continua’ model of the relationship between p...
2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2007
A three stage development process for the production of a hierarchical rule based prognosis tool ... more A three stage development process for the production of a hierarchical rule based prognosis tool is described. The application for this tool is specific to breast cancer patients that have a positive expression of the HER 2 gene. The first stage is the development of a Bayesian classification neural network to classify for cancer specific mortality. Secondly, low-order Boolean rules are extracted form this model using an Orthogonal Search based Rule Extraction (OSRE) algorithm. Further to these rules additional information is gathered from the Kaplan-Meier survival estimates of the population, stratified by the categorizations of the input variables. Finally, expert knowledge is used to further simplify the rules and to rank them hierarchically in the form of a decision tree. The resulting decision tree groups all observations into specific categories by clinical profile and by event rate. The practical clinical value of this decision support tool will in future be tested by external validation with additional data from other clinical centres. CONFIDENTIAL. Limited circulation. For review only.
... Byrom Street, L3 3AF, Liverpool, UK [email protected] 2 GapInfomedia p.ramsey@ gapinfomedia... more ... Byrom Street, L3 3AF, Liverpool, UK [email protected] 2 GapInfomedia p.ramsey@ gapinfomedia.com Abstract. ... BMJ 308, 283–284 (1994) 7. Wyatt, J.: Same information, different decisions: format counts – Format as well as content matters in clinical information. ...
2007 International Joint Conference on Neural Networks, 2007
Qualitative model operation description is useful for its direct validation using expert domain k... more Qualitative model operation description is useful for its direct validation using expert domain knowledge. A framework for this purpose uses low-order Boolean rules to approximate the response surfaces generated by analytical inference models. In the case of censored data, this approach serves to characterise the allocation of patients into risk groups generated by a risk staging index. Furthermore, the low-order rules define low-dimensional sub-spaces where individual data points can be directly visualised by reference to decision boundaries for risk group allocation. The well-known ROC framework has recently been extended to a threshold independent, time-dependent performance index to quantify the predictive accuracy of censored data models, termed the C td index. Taken together, the quantitative performance index, Boolean explanatory rules and direct visualisation of the data, define a consistent and transparent validation framework based on triangulation of information. This information can be included in decision support systems.
2014 IEEE Symposium on Computational Intelligence and Data Mining (CIDM), 2014
Algorithms in the Adaptive Resonance Theory (ART) family adapt to structural changes in data as n... more Algorithms in the Adaptive Resonance Theory (ART) family adapt to structural changes in data as new information presents, making it an exciting candidate for dynamic online clustering of big health data. Its use however has largely been restricted to the signal processing field. In this paper we introduce an adaptation of the ART2-A method within a separation and concordance (SeCo) framework which has been shown to identify stable and reproducible solutions from repeated initialisations that also provides evidence for an appropriate number of initial clusters that best calibrates the algorithm with the data presented. The results show stable, reproducible solutions for a mix of real-world heath related datasets and well known benchmark datasets, selecting solutions which better represent the underlying structure of the data then using a single measure of separation. The scalability of the method and it's facility for dynamic online clustering makes it suitable for finding structure in big data.
Proceedings of the International Joint Conference on Neural Networks, 2009
Early characterization of patients with respect to their predicted response to treatment is a fun... more Early characterization of patients with respect to their predicted response to treatment is a fundamental step towards the delivery of effective, personalized care. Starting from the results of a time-to-event model with competing risks using the framework of partial logistic artificial neural networks with automatic relevance determination (PLANNCR-ARD), we discuss an effective semi-supervised approach to patient stratification with application to
Proceedings - 7th International Conference on Machine Learning and Applications, ICMLA 2008, 2008
Page 1. Missing data imputation in longitudinal cohort studies - application of PLANN-ARD in brea... more Page 1. Missing data imputation in longitudinal cohort studies - application of PLANN-ARD in breast cancer survival Ana S. Fernandes2, Ian H. Jarman1, Terence A. Etchells1 José M. Fonseca2, Elia Biganzoli3, Chris Bajdik4 and Paulo JG Lisboa1 ...
IFIP Advances in Information and Communication Technology, 2010
This paper proposes a Web clinical decision support system for clinical oncologists and for breas... more This paper proposes a Web clinical decision support system for clinical oncologists and for breast cancer patients making prognostic assessments, using the particular characteristics of the individual patient. This system comprises three different prognostic modelling methodologies: the clinically widely used Nottingham prognostic index (NPI); the Cox regression modelling and a partial logistic artificial neural network with automatic relevance determination (PLANN-ARD). All three models yield a different prognostic index that can be analysed together in order to obtain a more accurate prognostic assessment of the patient. Missing data is incorporated in the mentioned models, a common issue in medical data that was overcome using multiple imputation techniques. Risk group assignments are also provided through a methodology based on regression trees, where Boolean rules can be obtained expressed with patient characteristics.
Neutrophil to lymphocyte count ratio as an early indicator of blood stream infection in the emerg... more Neutrophil to lymphocyte count ratio as an early indicator of blood stream infection in the emergency department
The 2012 International Joint Conference on Neural Networks (IJCNN), 2012
Blind signal separation (BSS) is a powerful tool to open-up complex signals into component source... more Blind signal separation (BSS) is a powerful tool to open-up complex signals into component sources that are often interpretable. However, BSS methods are generally unsupervised, therefore the assignment of class membership from the elements of the mixing matrix may be sub-optimal. This paper proposes a three-stage approach using Fisher information metric to define a natural metric for the data, from which a Euclidean approximation can then be used to drive BSS. Results with synthetic data models of real-world high-dimensional data show that the classification accuracy of the method is good for challenging problems, while retaining interpretability.
Clustering methods and nearest neighbour classifiers typically compute distances between data poi... more Clustering methods and nearest neighbour classifiers typically compute distances between data points as a measure of similarity, with nearby pairs of points considered more like each other than remote pairs. The distance measure of choice is often Euclidean, implicitly treating all directions in space as equally relevant. This paper reviews the application of Fisher information to derive a metric in primary data space. The aim is to provide a natural coordinate space to represent pairwise distances with respect to a probability distribution p(c|x), defined by an external label c, and use it to compute more informative distances.
Intelligent Data Engineering and Automated Learning – IDEAL 2019, 2019
This case study benchmarks a range of statistical and machine learning methods relevant to comput... more This case study benchmarks a range of statistical and machine learning methods relevant to computer-based decision support in clinical medicine, focusing on the diagnosis of knee osteoarthritis at first presentation. The methods, comprising logistic regression, Multilayer Perceptron (MLP), Chi-square Automatic Interaction Detector (CHAID) and Classification and Regression Trees (CART), are applied to a public domain database, the Osteoarthritis Initiative (OAI), a 10 year longitudinal study starting in 2002 (n = 4,796). In this real-world application, it is shown that logistic regression is comparable with the neural networks and decision trees for discrimination of positive diagnosis on this data set. This is likely because of weak non-linearities among high levels of noise. After comparing the explanations provided by the different methods, it is concluded that the interpretability of the risk score index provided by logistic regression is expressed in a form that most naturally integrates with clinical reasoning. The reason for this is that it gives a statistical assessment of the weight of evidence for making the diagnosis, so providing a direction for future research to improve explanation of generic non-linear models.
4 Department of Applied Mathematics, Liverpool John Moores University, James Parsons 5 Building, ... more 4 Department of Applied Mathematics, Liverpool John Moores University, James Parsons 5 Building, Byrom Street, Liverpool, L3 3AF, UK. 6 EPSRC Liverpool Centre for Mathematics in Healthcare, Department of Mathematical 7 Sciences, University of Liverpool, Liverpool, L69 7ZL, UK. 8 MRC Centre for Inflammation Research, Queens Medical Research Institute, University of 9 Edinburgh, 47 Little France Crescent, Edinburgh, EH16 4TJ, UK. 10 To whom correspondence should be addressed at Department of Mathematical Sciences, 11 University of Liverpool, Peach Street, Liverpool, L69 7ZL, UK. Tel: +44 151 794 4049. E12 mail: [email protected]. 13 Both authors contributed equally. 14
The most commonly prescribed painkiller worldwide, paracetamol (acetaminophen, APAP) is also the ... more The most commonly prescribed painkiller worldwide, paracetamol (acetaminophen, APAP) is also the predominant cause of acute liver failure (ALF), and therefore paracetamol-induced liver toxicity remains an important clinical problem. The standard clinical treatment framework for paracetamol overdose currently allows for antidote therapy decisions to be made based on a nomogram treatment line. This treatment threshold is lowered for patients adjudged to be highly susceptible to liver injury due to risk factors such as anorexia nervosa or bulimia. Additionally, both the original and adjusted clinical frameworks are highly dependent on knowledge from the patient regarding time since ingestion and initial dose amount, both of which are often highly unpredictable. We have recently developed a pre-clinical framework for predicting time since ingestion, initial dose amount and subsequent probability of liver injury based on novel biomarker concentrations. Here, we use identifiability analysis as a tool to increase confidence in our model parameter estimates and extend the framework to make predictions for both healthy and high-risk populations. Through pharmacokineticpharmacodynamic model refinement, we identify thresholds that determine whether necrosis or apoptosis is the dominant form of cell death, which can be essential for effective ALF interventions. Using a single blood test, rather than the multiple tests required in the current clinical frameworks, our model provides overdose identification information applicable for healthy and high-risk individuals as well as quantitative measures of estimated liver injury probability.
Introduction: Withdrawal assessment in critically ill children is complicated by the reliance on ... more Introduction: Withdrawal assessment in critically ill children is complicated by the reliance on non-specific behaviours and compounded when the child's typical behaviours are unknown. The existing approach to withdrawal assessment assumes that nurses elicit the parents' view of the child's behaviours. Objective and research methodology: This qualitative study explored parents' perspectives of their child's withdrawal and preferences for involvement and participation in withdrawal assessment. Parents of eleven children were interviewed after their child had completed sedation weaning during recovery from critical illness. Data were analysed using thematic analysis. Setting: A large children's hospital in the Northwest of England. Findings: Parents experienced varying degrees of partnership in the context of withdrawal assessment and identified information deficits which contributed to their distress of parenting a child with withdrawal syndrome. Most parents were eager to participate in withdrawal assessment and reported instances where their knowledge enabled a personalised interpretation of their child's behaviours. Reflecting on the reciprocal nature of the information deficits resulted in the development of a model for nurse-parent collaboration in withdrawal assessment. Conclusion: Facilitating nurse-parent collaboration in withdrawal assessment may have reciprocal benefits by moderating parental stress and aiding the assessment and management of withdrawal syndrome.
CPT: pharmacometrics & systems pharmacology, Jan 18, 2018
Paracetamol (acetaminophen (APAP)) is one of the most commonly used analgesics in the United King... more Paracetamol (acetaminophen (APAP)) is one of the most commonly used analgesics in the United Kingdom and the United States. However, exceeding the maximum recommended dose can cause serious liver injury and even death. Promising APAP toxicity biomarkers are thought to add value to those used currently and clarification of the functional relationships between these biomarkers and liver injury would aid clinical implementation of an improved APAP toxicity identification framework. The framework currently used to define an APAP overdose is highly dependent upon time since ingestion and initial dose; information that is often highly unpredictable. A pharmacokinetic/pharmacodynamic (PK/PD) APAP model has been built in order to understand the relationships between a panel of biomarkers and APAP dose. Visualization and statistical tools have been used to predict initial APAP dose and time since administration. Additionally, logistic regression analysis has been applied to histology data to...
The aim of the study was to evaluate registered children's nurses' approaches to the assessment a... more The aim of the study was to evaluate registered children's nurses' approaches to the assessment and management of withdrawal syndrome in children. Background Assessment of withdrawal syndrome is undertaken following critical illness when the child's condition may be unstable with competing differential diagnoses. Assessment tools aim to standardise and improve recognition of withdrawal syndrome. Making the right decisions in complex clinical situations requires a degree of mental effort and it is not known how nurses make decisions when undertaking withdrawal assessments. Design Cognitive interviews with clinical vignettes. Methods Interviews were undertaken with 12 nurses to explore the cognitive processes they utilised when assessing children using the Sedation Withdrawal Score (SWS) tool. Interviews took place in Autumn 2013. Findings Each stage of decision-making-noticing, interpreting and responding-presented cognitive challenges for nurses. When defining withdrawal behaviours nurses tended to blur the boundaries between SWS signs. Challenges in interpreting behaviours arose from not knowing if the patient's behaviour was a result of withdrawal or other co-morbidities.
Journal of Strength and Conditioning Research, 2017
The purpose of the present study was to provide a detailed analysis of the physical demands of co... more The purpose of the present study was to provide a detailed analysis of the physical demands of competitive international female soccer match-play. A total of 148 individual match observations were undertaken on 107 outfield players competing in competitive international matches during the 2011-2012 and 2012-2013 seasons, using a computerized tracking system (Prozone Sports Ltd., Leeds, England). Total distance (TD) and total high-speed running distances (THSR) were influenced by playing position, with central midfielders (CM) completing the highest (10985±706 m and 2882±500 m) and central defenders (CD) the lowest (9489±562 m and 1901±268 m) distances, respectively. Greater total very highspeed running (TVHSR) distances were completed when a team was without (399±143 m) compared to with (313±210 m) possession of the ball. The majority of sprints were over short distances with 76 % and 95 % being less than 5 m and 10 m, respectively. Between half reductions in physical performance were present for all variables, independent of playing position. The current study provides novel findings regarding the physical demands of different playing positions in competitive international female match-play and provides important insights for physical coaches preparing elite female players for competition.
The version presented here may differ from the published version or from the version of the recor... more The version presented here may differ from the published version or from the version of the record. Please see the repository URL above for details on accessing the published version and note that access may require a subscription.
The Fisher information metric defines a Riemannian space where distances reflect similarity with ... more The Fisher information metric defines a Riemannian space where distances reflect similarity with respect to a given probability distribution. This metric can be used during the process of building a relational network, resulting in a structure that is informed about the similarity criterion. Furthermore, the relational nature of this network allows for an intuitive interpretation of the data through their location within the network and the way it relates to the most representative cases or prototypes.
BackgroundThe relationship between well-being and mental ill health is complex; people may experi... more BackgroundThe relationship between well-being and mental ill health is complex; people may experience very low levels of well-being even in the absence of overt mental health problems.AimsThis study tested the hypothesis that anxiety, depression and well-being have different causal determinants and psychological mediating mechanisms.MethodThe influence of causal and mediating factors on anxiety, depression and well-being were investigated in a cross-sectional online questionnaire survey hosted on a UK national broadcasting website.ResultsMultivariate conditional independence analysis of data from 27 397 participants revealed different association pathways for the two constructs. Anxiety and depression were associated with negative life events mediated by rumination; low levels of subjective well-being were associated with material deprivation and social isolation, mediated by adaptive coping style.ConclusionsOur findings support the ‘two continua’ model of the relationship between p...
2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2007
A three stage development process for the production of a hierarchical rule based prognosis tool ... more A three stage development process for the production of a hierarchical rule based prognosis tool is described. The application for this tool is specific to breast cancer patients that have a positive expression of the HER 2 gene. The first stage is the development of a Bayesian classification neural network to classify for cancer specific mortality. Secondly, low-order Boolean rules are extracted form this model using an Orthogonal Search based Rule Extraction (OSRE) algorithm. Further to these rules additional information is gathered from the Kaplan-Meier survival estimates of the population, stratified by the categorizations of the input variables. Finally, expert knowledge is used to further simplify the rules and to rank them hierarchically in the form of a decision tree. The resulting decision tree groups all observations into specific categories by clinical profile and by event rate. The practical clinical value of this decision support tool will in future be tested by external validation with additional data from other clinical centres. CONFIDENTIAL. Limited circulation. For review only.
... Byrom Street, L3 3AF, Liverpool, UK [email protected] 2 GapInfomedia p.ramsey@ gapinfomedia... more ... Byrom Street, L3 3AF, Liverpool, UK [email protected] 2 GapInfomedia p.ramsey@ gapinfomedia.com Abstract. ... BMJ 308, 283–284 (1994) 7. Wyatt, J.: Same information, different decisions: format counts – Format as well as content matters in clinical information. ...
2007 International Joint Conference on Neural Networks, 2007
Qualitative model operation description is useful for its direct validation using expert domain k... more Qualitative model operation description is useful for its direct validation using expert domain knowledge. A framework for this purpose uses low-order Boolean rules to approximate the response surfaces generated by analytical inference models. In the case of censored data, this approach serves to characterise the allocation of patients into risk groups generated by a risk staging index. Furthermore, the low-order rules define low-dimensional sub-spaces where individual data points can be directly visualised by reference to decision boundaries for risk group allocation. The well-known ROC framework has recently been extended to a threshold independent, time-dependent performance index to quantify the predictive accuracy of censored data models, termed the C td index. Taken together, the quantitative performance index, Boolean explanatory rules and direct visualisation of the data, define a consistent and transparent validation framework based on triangulation of information. This information can be included in decision support systems.
2014 IEEE Symposium on Computational Intelligence and Data Mining (CIDM), 2014
Algorithms in the Adaptive Resonance Theory (ART) family adapt to structural changes in data as n... more Algorithms in the Adaptive Resonance Theory (ART) family adapt to structural changes in data as new information presents, making it an exciting candidate for dynamic online clustering of big health data. Its use however has largely been restricted to the signal processing field. In this paper we introduce an adaptation of the ART2-A method within a separation and concordance (SeCo) framework which has been shown to identify stable and reproducible solutions from repeated initialisations that also provides evidence for an appropriate number of initial clusters that best calibrates the algorithm with the data presented. The results show stable, reproducible solutions for a mix of real-world heath related datasets and well known benchmark datasets, selecting solutions which better represent the underlying structure of the data then using a single measure of separation. The scalability of the method and it's facility for dynamic online clustering makes it suitable for finding structure in big data.
Proceedings of the International Joint Conference on Neural Networks, 2009
Early characterization of patients with respect to their predicted response to treatment is a fun... more Early characterization of patients with respect to their predicted response to treatment is a fundamental step towards the delivery of effective, personalized care. Starting from the results of a time-to-event model with competing risks using the framework of partial logistic artificial neural networks with automatic relevance determination (PLANNCR-ARD), we discuss an effective semi-supervised approach to patient stratification with application to
Proceedings - 7th International Conference on Machine Learning and Applications, ICMLA 2008, 2008
Page 1. Missing data imputation in longitudinal cohort studies - application of PLANN-ARD in brea... more Page 1. Missing data imputation in longitudinal cohort studies - application of PLANN-ARD in breast cancer survival Ana S. Fernandes2, Ian H. Jarman1, Terence A. Etchells1 José M. Fonseca2, Elia Biganzoli3, Chris Bajdik4 and Paulo JG Lisboa1 ...
IFIP Advances in Information and Communication Technology, 2010
This paper proposes a Web clinical decision support system for clinical oncologists and for breas... more This paper proposes a Web clinical decision support system for clinical oncologists and for breast cancer patients making prognostic assessments, using the particular characteristics of the individual patient. This system comprises three different prognostic modelling methodologies: the clinically widely used Nottingham prognostic index (NPI); the Cox regression modelling and a partial logistic artificial neural network with automatic relevance determination (PLANN-ARD). All three models yield a different prognostic index that can be analysed together in order to obtain a more accurate prognostic assessment of the patient. Missing data is incorporated in the mentioned models, a common issue in medical data that was overcome using multiple imputation techniques. Risk group assignments are also provided through a methodology based on regression trees, where Boolean rules can be obtained expressed with patient characteristics.
Neutrophil to lymphocyte count ratio as an early indicator of blood stream infection in the emerg... more Neutrophil to lymphocyte count ratio as an early indicator of blood stream infection in the emergency department
The 2012 International Joint Conference on Neural Networks (IJCNN), 2012
Blind signal separation (BSS) is a powerful tool to open-up complex signals into component source... more Blind signal separation (BSS) is a powerful tool to open-up complex signals into component sources that are often interpretable. However, BSS methods are generally unsupervised, therefore the assignment of class membership from the elements of the mixing matrix may be sub-optimal. This paper proposes a three-stage approach using Fisher information metric to define a natural metric for the data, from which a Euclidean approximation can then be used to drive BSS. Results with synthetic data models of real-world high-dimensional data show that the classification accuracy of the method is good for challenging problems, while retaining interpretability.
Clustering methods and nearest neighbour classifiers typically compute distances between data poi... more Clustering methods and nearest neighbour classifiers typically compute distances between data points as a measure of similarity, with nearby pairs of points considered more like each other than remote pairs. The distance measure of choice is often Euclidean, implicitly treating all directions in space as equally relevant. This paper reviews the application of Fisher information to derive a metric in primary data space. The aim is to provide a natural coordinate space to represent pairwise distances with respect to a probability distribution p(c|x), defined by an external label c, and use it to compute more informative distances.
Uploads
Papers by Ian Jarman