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.

An artificial neural network stratifies the risks of reintervention and mortality after endovascular aneurysm repair:a retrospective observational study

Background Lifelong surveillance after endovascular repair (EVAR) of abdominal aortic aneurysms (AAA) is considered mandatory to detect potentially life-threatening endograft complications. A minority of patients require reintervention but cannot be predictively identified by existing methods. This study aimed to improve the prediction of endograft complications and mortality, through the application of machine-learning techniques. Methods Patients undergoing EVAR at 2 centres were studied from 2004-2010. Pre-operative aneurysm morphology was quantified and endograft complications were recorded up to 5 years following surgery. An artificial neural networks (ANN) approach was used to predict whether patients would be at low- or high-risk of endograft complications (aortic/limb) or mortality. Centre 1 data were used for training and centre 2 data for validation. ANN performance was assessed by Kaplan-Meier analysis to compare the incidence of aortic complications, limb complications, and mortality; in patients predicted to be low-risk, versus those predicted to be high-risk. Results 761 patients aged 75 +/- 7 years underwent EVAR. Mean follow-up was 36+/- 20 months. An ANN was created from morphological features including angulation/length/areas/diameters/ volume/tortuosity of the aneurysm neck/sac/iliac segments. ANN models predicted endograft complications and mortality with excellent discrimination between a low-risk and high-risk group. In external validation, the 5-year rates of freedom from aortic complications, limb complications and mortality were 95.9% vs 67.9%; 99.3% vs 92.0%; and 87.9% vs 79.3% respectively (p0.001) Conclusion This study presents ANN models that stratify the 5-year risk of endograft complications or mortality using routinely available pre-operative data.

The impact of bariatric surgery on estimated glomerular filtration rate in patients with type 2 diabetes:a retrospective cohort study

Background: Diabetes mellitus is the most common cause of end-stage renal disease, which is associated with increased morbidity and mortality. The impact of bariatric surgery on chronic kidney disease is unclear. Objectives: Our primary aim was to assess the impact of bariatric surgery on estimated glomerular filtration rate (eGFR) in type 2 diabetes (T2D) patients. Our secondary aim was to compare the impact of bariatric surgery versus routine care on eGFR in patients with T2D. Setting: University Hospital, United Kingdom. Methods: A retrospective cohort analysis of adults with T2D who underwent bariatric surgery at a single center between January 2005 and December 2012. Data regarding eGFR were obtained from electronic patients records. eGFR was calculated using the Modification of Diet in Renal Disease formula. Data regarding patients with T2D who did not undergo bariatric surgery ("routine care") were obtained from patients attending the diabetes clinic at the same center from 2009 to 2011. Results: One hundred sixty-three patients were included (mean age 48.5±8.8 yr; baseline body mass index 50.8±9.1 kg/m2) and were followed for 3.0±2.3 years. Bariatric surgery resulted in an improvement in eGFR (median [interquartile range] 86.0 [73.0-100.0] versus 92.0 [77.0-101.0] mL/min/1.73 m2 for baseline versus follow-up, respectively; P = .003), particularly in patients with baseline eGFR≤60 mL/min/1.73 m2 (48.0 [42.0-57.0] versus 61.0 [55.0-63.0] mL/min/1.73 m2; P = .004). After adjusting for baseline eGFR, glycated hemoglobin (HbA1C), body mass index, age, and gender, bariatric surgery was associated with higher study-end eGFR compared with routine care (B = 7.787; P< .001). Conclusion: Bariatric surgery results in significant improvements in eGFR in T2D patients, particularly those with an eGFR≤60 mL/min/1.73 m2, while routine care was associated with a decline in eGFR.

A retrospective study examining the nature, contributory factors and predictors of dispensing errors in community pharmacy

Introduction: Since 2005, the workload of community pharmacists in England has increased with a concomitant increase in stress and work pressure. However, it is unclear how these factors are impacting on the ability of community pharmacists to ensure accuracy during the dispensing process. This research seeks to extend our understanding of the nature, outcome, and predictors of dispensing errors. Methodology: A retrospective analysis of a purposive sample of incident report forms (IRFs) from the database of a pharmacist indemnity insurance provider was conducted. Data collected included; type of error, degree of harm caused, pharmacy and pharmacist demographics, and possible contributory factors. Results: In total, 339 files from UK community pharmacies were retrieved from the database. The files dated from June 2006 to November 2011. Incorrect item (45.1%, n = 153/339) followed by incorrect strength (24.5%, n = 83/339) were the most common forms of error. Almost half (41.6%, n = 147/339) of the patients suffered some form of harm ranging from minor harm (26.7%, n = 87/339) to death (0.3%, n = 1/339). Insufficient staff (51.6%, n = 175/339), similar packaging (40.7%, n = 138/339) and the pharmacy being busier than normal (39.5%, n = 134/339) were identified as key contributory factors. Cross-tabular analysis against the final accuracy check variable revealed significant association between the pharmacy location (P < 0.024), dispensary layout (P < 0.025), insufficient staff (P < 0.019), and busier than normal (P < 0.005) variables. Conclusion: The results provide an overview of some of the individual, organisational and technical factors at play at the time of a dispensing error and highlight the need to examine further the relationships between these factors and dispensing error occurrence.