Using secondary knowledge to support decision tree classification of retrospective clinical data

Retrospective clinical data presents many challenges for data mining and machine learning. The transcription of patient records from paper charts and subsequent manipulation of data often results in high volumes of noise as well as a loss of other important information. In addition, such datasets often fail to represent expert medical knowledge and reasoning in any explicit manner. In this research we describe applying data mining methods to retrospective clinical data to build a prediction model for asthma exacerbation severity for pediatric patients in the emergency department. Difficulties in building such a model forced us to investigate alternative strategies for analyzing and processing retrospective data. This paper describes this process together with an approach to mining retrospective clinical data by incorporating formalized external expert knowledge (secondary knowledge sources) into the classification task. This knowledge is used to partition the data into a number of coherent sets, where each set is explicitly described in terms of the secondary knowledge source. Instances from each set are then classified in a manner appropriate for the characteristics of the particular set. We present our methodology and outline a set of experiential results that demonstrate some advantages and some limitations of our approach. © 2008 Springer-Verlag Berlin Heidelberg.

A method for automatically eliciting node weights in a hierarchical knowledge-based structure for reasoning with uncertainty

Hierarchical knowledge structures are frequently used within clinical decision support systems as part of the model for generating intelligent advice. The nodes in the hierarchy inevitably have varying influence on the decisionmaking processes, which needs to be reflected by parameters. If the model has been elicited from human experts, it is not feasible to ask them to estimate the parameters because there will be so many in even moderately-sized structures. This paper describes how the parameters could be obtained from data instead, using only a small number of cases. The original method [1] is applied to a particular web-based clinical decision support system called GRiST, which uses its hierarchical knowledge to quantify the risks associated with mental-health problems. The knowledge was elicited from multidisciplinary mental-health practitioners but the tree has several thousand nodes, all requiring an estimation of their relative influence on the assessment process. The method described in the paper shows how they can be obtained from about 200 cases instead. It greatly reduces the experts’ elicitation tasks and has the potential for being generalised to similar knowledge-engineering domains where relative weightings of node siblings are part of the parameter space.

Designing multiple user perspectives and functionality for clinical decision support systems

Clinical Decision Support Systems (CDSSs) need to disseminate expertise in formats that suit different end users and with functionality tuned to the context of assessment. This paper reports research into a method for designing and implementing knowledge structures that facilitate the required flexibility. A psychological model of expertise is represented using a series of formally specified and linked XML trees that capture increasing elements of the model, starting with hierarchical structuring, incorporating reasoning with uncertainty, and ending with delivering the final CDSS. The method was applied to the Galatean Risk and Safety Tool, GRiST, which is a web-based clinical decision support system (www.egrist.org) for assessing mental-health risks. Results of its clinical implementation demonstrate that the method can produce a system that is able to deliver expertise targetted and formatted for specific patient groups, different clinical disciplines, and alternative assessment settings. The approach may be useful for developing other real-world systems using human expertise and is currently being applied to a logistics domain. © 2013 Polish Information Processing Society.

Classifying clinical decision making:interpreting nursing intuition, heuristics and medical diagnosis

This is the second of two linked papers exploring decision making in nursing. The first paper, 'Classifying clinical decision making: a unifying approach' investigated difficulties with applying a range of decision-making theories to nursing practice. This is due to the diversity of terminology and theoretical concepts used, which militate against nurses being able to compare the outcomes of decisions analysed within different frameworks. It is therefore problematic for nurses to assess how good their decisions are, and where improvements can be made. However, despite the range of nomenclature, it was argued that there are underlying similarities between all theories of decision processes and that these should be exposed through integration within a single explanatory framework. A proposed solution was to use a general model of psychological classification to clarify and compare terms, concepts and processes identified across the different theories. The unifying framework of classification was described and this paper operationalizes it to demonstrate how different approaches to clinical decision making can be re-interpreted as classification behaviour. Particular attention is focused on classification in nursing, and on re-evaluating heuristic reasoning, which has been particularly prone to theoretical and terminological confusion. Demonstrating similarities in how different disciplines make decisions should promote improved multidisciplinary collaboration and a weakening of clinical elitism, thereby enhancing organizational effectiveness in health care and nurses' professional status. This is particularly important as nurses' roles continue to expand to embrace elements of managerial, medical and therapeutic work. Analysing nurses' decisions as classification behaviour will also enhance clinical effectiveness, and assist in making nurses' expertise more visible. In addition, the classification framework explodes the myth that intuition, traditionally associated with nurses' decision making, is less rational and scientific than other approaches.

Cognitive-behavioural phenotype in a group of girls from 1.2 to 12 years old with the Incontinentia Pigmenti syndrome:recommendations for clinical management

Incontinentia Pigmenti (IP, OMIM#308300) is a rare X-linked genomic disorder (about 1,400 cases) that affects the neuroectodermal tissue and Central Nervous System (CNS). The objective of this study was to describe the cognitive-behavioural profile in children in order to plan a clinical intervention to improve their quality of life. A total of 14 girls (age range: from 1 year and 2 months to 12 years and 10 months) with IP and the IKBKG/NEMO gene deletion were submitted to a cognitive assessment including intelligence scales, language and visuo-spatial competence tests, learning ability tests, and a behavioural assessment. Five girls had severe to mild intellectual deficiencies and the remaining nine had a normal neurodevelopment. Four girls were of school age and two of these showed no intellectual disability, but had specific disabilities in calculation and arithmetic reasoning. This is the first description of the cognitive-behavioural profile in relation to developmental age. We stress the importance of an early assessment of learning abilities in individuals with IP without intellectual deficiencies to prevent the onset of any such deficit.