Stabilizing sparse cox model using statistic and semantic structures in electronic medical records

Stability in clinical prediction models is crucial for transferability between studies, yet has received little attention. The problem is paramount in high dimensional data, which invites sparse models with feature selection capability. We introduce an effective method to stabilize sparse Cox model of time-to-events using statistical and semantic structures inherent in Electronic Medical Records (EMR). Model estimation is stabilized using three feature graphs built from (i) Jaccard similarity among features (ii) aggregation of Jaccard similarity graph and a recently introduced semantic EMR graph (iii) Jaccard similarity among features transferred from a related cohort. Our experiments are conducted on two real world hospital datasets: a heart failure cohort and a diabetes cohort. On two stability measures – the Consistency index and signal-to-noise ratio (SNR) – the use of our proposed methods significantly increased feature stability when compared with the baselines.

Lab-on-a-chip or chip-in-a-lab: challenges of commercialization lost in translation

Lab-on-a-chip technology has been long envisaged to have tremendous commercial potential, owing to the ability of such devices to encapsulate a full range of laboratory processes in a single instrument and operate in a portable manner, rapidly and at low cost. Devices are believed to have potential in fields ranging across medical diagnostics, environmental sampling and a range of consumer products, however, to date very few devices have attained commercial success. This review examines the challenges relating to the commercialization of lab-on-a-chip technology from fundamental research to mass manufacturing and aims to provide insight to both academics and product development specialists the perceived hindrances to commercialization and a strategy by which future work could be translated into commercial success.

Keeping up with innovation: a predictive framework for modeling healthcare data with evolving clinical interventions

Medical outcomes are inexorably linked to patient illness and clinical interventions. Interventions change the course of disease, crucially determining outcome. Traditional outcome prediction models build a single classifier by augmenting interventions with disease information. Interventions, however, differentially affect prognosis, thus a single prediction rule may not suffice to capture variations. Interventions also evolve over time as more advanced interventions replace older ones. To this end, we propose a Bayesian nonparametric, supervised framework that models a set of intervention groups through a mixture distribution building a separate prediction rule for each group, and allows the mixture distribution to change with time. This is achieved by using a hierarchical Dirichlet process mixture model over the interventions. The outcome is then modeled as conditional on both the latent grouping and the disease information through a Bayesian logistic regression. Experiments on synthetic and medical cohorts for 30-day readmission prediction demonstrate the superiority of the proposed model over clinical and data mining baselines.

Mass spectrometry-based proteomic data for cancer diagnosis using interval type-2 fuzzy system

An interval type-2 fuzzy logic system is introduced for cancer diagnosis using mass spectrometry-based proteomic data. The fuzzy system is incorporated with a feature extraction procedure that combines wavelet transform and Wilcoxon ranking test. The proposed feature extraction generates feature sets that serve as inputs to the type-2 fuzzy classifier. Uncertainty, noise and outliers that are common in the proteomic data motivate the use of type-2 fuzzy system. Tabu search is applied for structure learning of the fuzzy classifier. Experiments are performed using two benchmark proteomic datasets for the prediction of ovarian and pancreatic cancer. The dominance of the suggested feature extraction as well as type-2 fuzzy classifier against their competing methods is showcased through experimental results. The proposed approach therefore is helpful to clinicians and practitioners as it can be implemented as a medical decision support system in practice.

Hierarchical Bayesian nonparametric models for knowledge discovery from electronic medical records

Electronic Medical Record (EMR) has established itself as a valuable resource for large scale analysis of health data. A hospital EMR dataset typically consists of medical records of hospitalized patients. A medical record contains diagnostic information (diagnosis codes), procedures performed (procedure codes) and admission details. Traditional topic models, such as latent Dirichlet allocation (LDA) and hierarchical Dirichlet process (HDP), can be employed to discover disease topics from EMR data by treating patients as documents and diagnosis codes as words. This topic modeling helps to understand the constitution of patient diseases and offers a tool for better planning of treatment. In this paper, we propose a novel and flexible hierarchical Bayesian nonparametric model, the word distance dependent Chinese restaurant franchise (wddCRF), which incorporates word-to-word distances to discover semantically-coherent disease topics. We are motivated by the fact that diagnosis codes are connected in the form of ICD-10 tree structure which presents semantic relationships between codes. We exploit a decay function to incorporate distances between words at the bottom level of wddCRF. Efficient inference is derived for the wddCRF by using MCMC technique. Furthermore, since procedure codes are often correlated with diagnosis codes, we develop the correspondence wddCRF (Corr-wddCRF) to explore conditional relationships of procedure codes for a given disease pattern. Efficient collapsed Gibbs sampling is derived for the Corr-wddCRF. We evaluate the proposed models on two real-world medical datasets - PolyVascular disease and Acute Myocardial Infarction disease. We demonstrate that the Corr-wddCRF model discovers more coherent topics than the Corr-HDP. We also use disease topic proportions as new features and show that using features from the Corr-wddCRF outperforms the baselines on 14-days readmission prediction. Beside these, the prediction for procedure codes based on the Corr-wddCRF also shows considerable accuracy.

Multiple time series anomaly detection based on compression and correlation analysis: a medical surveillance case study

In this paper, we present a novel anomaly detection framework for multiple heterogeneous yet correlated time series, such as the medical surveillance series data. In our framework, we propose an anomaly detection algorithm from the viewpoint of trend and correlation analysis. Moreover, to efficiently process huge amount of observed time series, a new clustering-based compression method is proposed. Experimental results indicate that our framework is more effective and efficient than its peers. © 2012 Springer-Verlag Berlin Heidelberg.

Integrating internet-of-things with the power of cloud computing and the intelligence of big Data analytics- a three layered approach

This paper is written through the vision on integrating Internet-of-Things (IoT) with the power of Cloud Computing and the intelligence of Big Data analytics. But integration of all these three cutting edge technologies is complex to understand. In this research we first provide a security centric view of three layered approach for understanding the technology, gaps and security issues. Then with a series of lab experiments on different hardware, we have collected performance data from all these three layers, combined these data together and finally applied modern machine learning algorithms to distinguish 18 different activities and cyber-attacks. From our experiments we find classification algorithm RandomForest can identify 93.9% attacks and activities in this complex environment. From the existing literature, no one has ever attempted similar experiment for cyber-attack detection for IoT neither with performance data nor with a three layered approach.

Can the direct medical cost of chronic disease be transferred across different countries? Using cost-of-illness studies on type 2 diabetes, epilepsy and schizophrenia as examples

OBJECTIVES: To systematically review cost of illness studies for schizophrenia (SC), epilepsy (EP) and type 2 diabetes mellitus (T2DM) and explore the transferability of direct medical cost across countries.

METHODS: A comprehensive literature search was performed to yield studies that estimated direct medical costs. A generalized linear model (GLM) with gamma distribution and log link was utilized to explore the variation in costs that accounted by the included factors. Both parametric (Random-effects model) and non-parametric (Boot-strapping) meta-analyses were performed to pool the converted raw cost data (expressed as percentage of GDP/capita of the country where the study was conducted).

RESULTS: In total, 93 articles were included (40 studies were for T2DM, 34 studies for EP and 19 studies for SC). Significant variances were detected inter- and intra-disease classes for the direct medical costs. Multivariate analysis identified that GDP/capita (p<0.05) was a significant factor contributing to the large variance in the cost results. Bootstrapping meta-analysis generated more conservative estimations with slightly wider 95% confidence intervals (CI) than the parametric meta-analysis, yielding a mean (95%CI) of 16.43% (11.32, 21.54) for T2DM, 36.17% (22.34, 50.00) for SC and 10.49% (7.86, 13.41) for EP.

CONCLUSIONS: Converting the raw cost data into percentage of GDP/capita of individual country was demonstrated to be a feasible approach to transfer the direct medical cost across countries. The approach from our study to obtain an estimated direct cost value along with the size of specific disease population from each jurisdiction could be used for a quick check on the economic burden of particular disease for countries without such data.

Medical burden, body mass index and the outcome of psychosocial interventions for bipolar depression

Objective: Individuals with bipolar disorder experience a disproportionately high incidence of medical co-morbidity and obesity. These health-related problems are a barrier to recovery from mood episodes and have been linked with unfavorable responses to pharmacological treatment. However, little is known about whether and how these characteristics affect responses to adjunctive psychotherapy. Method: Embedded in the Systematic Treatment Enhancement Program for Bipolar Disorder was a randomized controlled trial of psychotherapy for bipolar depression comparing the efficacy of intensive psychotherapy plus pharmacotherapy with collaborative care (a three-session psycho-educational intervention) plus pharmacotherapy. We conducted a post-hoc analysis to evaluate whether medical burden and body mass index predicted and/or moderated the likelihood of recovery and time until recovery from a depressive episode among patients in the two treatments. Results: Participants who had medical co-morbidity and body mass index data constituted 199 of the 293 patients in the original Systematic Treatment Enhancement Program for Bipolar Disorder trial. Higher medical burden predicted a lower likelihood of recovery from depression in both treatment conditions (odds ratio = 0.89), but did not moderate responses to intensive psychotherapy vs collaborative care. Intensive psychotherapy yielded superior recovery rates for individuals of normal body mass index (odds ratio= 2.39) compared with collaborative care, but not among individuals who were overweight or obese. Conclusion: Medical co-morbidity and body weight impacts symptom improvement and attention to this co-morbidity may inform the development of more personalized treatments for bipolar disorder.

Risk factors for incident delirium in an acute general medical setting: a retrospective case – control study

Aims and ObjectivesTo determine predisposing and precipitating risk factors for incident delirium in medical patients during an acute hospital admission.BackgroundIncident delirium is the most common complication of hospital admission for older patients. Up to 30% of hospitalised medical patients experience incident delirium. Determining risk factors for delirium is important for identifying patients that are most susceptible to incident delirium.DesignRetrospective case-control study with two controls per case.MethodsAn audit tool was used to review medical records of patients admitted to acute medical units for data regarding potential risk factors for delirium. Data was collected between August 2013 and March 2014 at three hospital sites of a healthcare organisation in Melbourne, Australia. Cases were 161 patients admitted to an acute medical ward and diagnosed with incident delirium between 1st January 2012 and 31st December 2013. Controls were 321 patients sampled from the acute medical population admitted within the same time range, stratified for admission location and who did not develop incident delirium during hospitalisation.ResultsIdentified using logistic regression modelling, predisposing risk factors for incident delirium were: dementia, cognitive impairment, functional impairment, previous delirium, and fracture on admission. Precipitating risk factors for incident delirium were: use of an indwelling catheter, adding more than three medications during admission and having an abnormal sodium level during admission.ConclusionsMultiple risk factors for incident delirium exist; patients with a history of delirium, dementia and cognitive impairment are at greatest risk of developing delirium during hospitalisation.Relevance to clinical practiceNurses and other health care professionals should be aware of patients that have one or more risk factors for incident delirium. Knowledge of risk factors for delirium has the potential to increase the recognition and understanding of patients who are vulnerable to delirium. Early recognition and prevention of delirium can contribute to improved patients safety and reduction in harm.

Predicting risk of suicide attempt using history of physical illnesses from electronic medical records

BACKGROUND: Although physical illnesses, routinely documented in electronic medical records (EMR), have been found to be a contributing factor to suicides, no automated systems use this information to predict suicide risk.

OBJECTIVE: The aim of this study is to quantify the impact of physical illnesses on suicide risk, and develop a predictive model that captures this relationship using EMR data.

METHODS: We used history of physical illnesses (except chapter V: Mental and behavioral disorders) from EMR data over different time-periods to build a lookup table that contains the probability of suicide risk for each chapter of the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) codes. The lookup table was then used to predict the probability of suicide risk for any new assessment. Based on the different lengths of history of physical illnesses, we developed six different models to predict suicide risk. We tested the performance of developed models to predict 90-day risk using historical data over differing time-periods ranging from 3 to 48 months. A total of 16,858 assessments from 7399 mental health patients with at least one risk assessment was used for the validation of the developed model. The performance was measured using area under the receiver operating characteristic curve (AUC).

RESULTS: The best predictive results were derived (AUC=0.71) using combined data across all time-periods, which significantly outperformed the clinical baseline derived from routine risk assessment (AUC=0.56). The proposed approach thus shows potential to be incorporated in the broader risk assessment processes used by clinicians.

CONCLUSIONS: This study provides a novel approach to exploit the history of physical illnesses extracted from EMR (ICD-10 codes without chapter V-mental and behavioral disorders) to predict suicide risk, and this model outperforms existing clinical assessments of suicide risk.

Shining the light on the dark side of medical leadership - a qualitative study in Australia

Purpose The paper aims to explore the beliefs of doctors in leadership roles of the concept of "the dark side", using data collected from interviews carried out with 45 doctors in medical leadership roles across Australia. The paper looks at the beliefs from the perspectives of doctors who are already in leadership roles themselves; to identify potential barriers they might have encountered and to arrive at better-informed strategies to engage more doctors in the leadership of the Australian health system. The research question is: "What are the beliefs of medical leaders that form the key themes or dimensions of the negative perception of the 'dark side'?". Design/methodology/approach The paper analysed data from two similar qualitative studies examining medical leadership and engagement in Australia by the same author, in collaboration with other researchers, which used in-depth semi-structured interviews with 45 purposively sampled senior medical leaders in leadership roles across Australia in health services, private and public hospitals, professional associations and health departments. The data were analysed using deductive and inductive approaches through a coding framework based on the interview data and literature review, with all sections of coded data grouped into themes. Findings Medical leaders had four key beliefs about the "dark side" as perceived through the eyes of their own past clinical experience and/or their clinical colleagues. These four beliefs or dimensions of the negative perception colloquially known as "the dark side" are the belief that they lack both managerial and clinical credibility, they have confused identities, they may be in conflict with clinicians, their clinical colleagues lack insight into the complexities of medical leadership and, as a result, doctors are actively discouraged from making the transition from clinical practice to medical leadership roles in the first place. Research limitations/implications This research was conducted within the Western developed-nation setting of Australia and only involved interviews with doctors in medical leadership roles. The findings are therefore limited to the doctors' own perceptions of themselves based on their past experiences and beliefs. Future research involving doctors who have not chosen to transition to leadership roles, or other health practitioners in other settings, may provide a broader perspective. Also, this research was exploratory and descriptive in nature using qualitative methods, and quantitative research can be carried out in the future to extend this research for statistical generalisation. Practical implications The paper includes implications for health organisations, training providers, medical employers and health departments and describes a multi-prong strategy to address this important issue. Originality/value This paper fulfils an identified need to study the concept of "moving to the dark side" as a negative perception of medical leadership and contributes to the evidence in this under-researched area. This paper has used data from two similar studies, combined together for the first time, with new analysis and coding, looking at the concept of the "dark side" to discover new emergent findings.

User centric design, data analysis and performance of snowboard bindings

This study focuses on soft boot snowboard bindings by looking at how users interact with their binding and proposes a possible solution to overcome such issues. Snowboarding is a multibillion-dollar sport that has only reached mainstream in the last 30 years its levels of progression in technology have evolved in that time. However, snowboard bindings for the most part still consist of the same basic architecture in the last 20 years. This study was aimed at taking a user centric point of view and using additive manufacturing technologies to be able to generate a new snowboard binding that is completely adaptable to the user. The initial part of the study was a survey of 280 snowboarders focussing on preferences, style and habits. This survey was generated from over 15 nations with the vast majority of boarders on the snow for five to fifty days a year. Significant emphasis was placed on the relationship between boarder binding set-up and occurrence of pain and/or injury. From the detailed survey it was found that boarder's experienced pain in the front foot/toe area as a result from the toe strap being too tight. However boarders wanted tighter bindings to increase responsiveness. Survey data was compared to ankle and foot biomechanics to build a relationship to assess the problem of pain versus responsiveness. The design stage of the study was to develop a binding that overcame the over-tightening of the binding but still maintain equivalent or better responsiveness compared to traditional bindings. The resulting design integrated the snowboard boot much more into the design, by using the sole as a "semi-rigid" platform and locking it in laterally between the heel cup and the new toe strap arrangement. The new design developed using additive manufacturing techniques was tested via qualitative and quantitative measures in the snow and in the lab. It was found that using the new arrangement in a system resulted in no loss of performance or responsiveness to the user. Due to the design and manufacturing approach users have the ability to customise the design to their specific needs.

Lung cancer prediction from microarray data by gene expression programming

Lung cancer is a leading cause of cancer-related death worldwide. The early diagnosis of cancer has demonstrated to be greatly helpful for curing the disease effectively. Microarray technology provides a promising approach of exploiting gene profiles for cancer diagnosis. In this study, the authors propose a gene expression programming (GEP)-based model to predict lung cancer from microarray data. The authors use two gene selection methods to extract the significant lung cancer related genes, and accordingly propose different GEP-based prediction models. Prediction performance evaluations and comparisons between the authors' GEP models and three representative machine learning methods, support vector machine, multi-layer perceptron and radial basis function neural network, were conducted thoroughly on real microarray lung cancer datasets. Reliability was assessed by the cross-data set validation. The experimental results show that the GEP model using fewer feature genes outperformed other models in terms of accuracy, sensitivity, specificity and area under the receiver operating characteristic curve. It is concluded that GEP model is a better solution to lung cancer prediction problems.

Evaluation and comparison of 3D intervertebral disc localization and segmentation methods for 3D T2 MR data: A grand challenge.

The evaluation of changes in Intervertebral Discs (IVDs) with 3D Magnetic Resonance (MR) Imaging (MRI) can be of interest for many clinical applications. This paper presents the evaluation of both IVD localization and IVD segmentation methods submitted to the Automatic 3D MRI IVD Localization and Segmentation challenge, held at the 2015 International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI2015) with an on-site competition. With the construction of a manually annotated reference data set composed of 25 3D T2-weighted MR images acquired from two different studies and the establishment of a standard validation framework, quantitative evaluation was performed to compare the results of methods submitted to the challenge. Experimental results show that overall the best localization method achieves a mean localization distance of 0.8 mm and the best segmentation method achieves a mean Dice of 91.8%, a mean average absolute distance of 1.1 mm and a mean Hausdorff distance of 4.3 mm, respectively. The strengths and drawbacks of each method are discussed, which provides insights into the performance of different IVD localization and segmentation methods.