Ross Sparks

CUSUM Charts for Signalling Varying Location Shifts

The conventional cumulative sum (CUSUM) with k = 0.5 is often used as the default CUSUM statistic when future shifts are unknown. In this paper, CUSUM procedures are designed to be efficient at signalling a range of future expected but unknown location shifts. Two approaches are advocated. The first uses three simultaneous conventional CUSUM statistics with different resetting boundaries. This results in a procedure that has, on average, several levels of memory, and thus signals a broader range of location shifts more efficiently than the conventional CUSUM with k = 0.5. The second uses an adaptive CUSUM statistic that continually adjusts its form to be efficient for signalling a one-step-ahead forecast in deviation from its target value. Average run length (ARL) is used to compare the relative performance of procedures. Several applications are used to illustrate procedures.

Mining risk patterns in medical data

In this paper, we discuss a problem of finding risk patterns in medical data. We define risk patterns by a statistical metric, relative risk, which has been widely used in epidemiological research. We characterise the problem of mining risk patterns as an optimal rule discovery problem. We study an anti-monotone property for mining optimal risk pattern sets and present an algorithm to make use of the property in risk pattern discovery. The method has been applied to a real world data set to find patterns associated with an allergic event for ACE inhibitors. The algorithm has generated some useful results for medical researchers.

Remote access methods for exploratory data analysis and statistical modelling: Privacy-Preserving Analytics ®

This paper is concerned with the challenge of enabling the use of confidential or private data for research and policy analysis, while protecting confidentiality and privacy by reducing the risk of disclosure of sensitive information. Traditional solutions to the problem of reducing disclosure risk include releasing de-identified data and modifying data before release. In this paper we discuss the alternative approach of using a remote analysis server which does not enable any data release, but instead is designed to deliver useful results of user-specified statistical analyses with a low risk of disclosure. The techniques described in this paper enable a user to conduct a wide range of methods in exploratory data analysis, regression and survival analysis, while at the same time reducing the risk that the user can read or infer any individual record attribute value. We illustrate our methods with examples from biostatistics using publicly available data. We have implemented our techniques into a software demonstrator called Privacy-Preserving Analytics (PPA), via a web-based interface to the R software. We believe that PPA may provide an effective balance between the competing goals of providing useful information and reducing disclosure risk in some situations.

Understanding sources of variation in syndromic surveillance for early warning of natural or intentional disease outbreaks

Daily counts of computer records of hospital emergency department arrivals grouped according to diagnosis (called here syndrome groupings) can be monitored by epidemiologists for changes in frequency that could provide early warning of bioterrorism events or naturally occurring disease outbreaks and epidemics. This type of public health surveillance is sometimes called syndromic surveillance. We used transitional Poisson regression models to obtain one-day-ahead arrival forecasts. Regression parameter estimates and forecasts were updated for each day using the latest 365 days of data. The resulting time series of recursive estimates of parameters such as the amplitude and location of the seasonal peaks as well as the one-day-ahead forecasts and forecast errors can be monitored to understand changes in epidemiology of each syndrome grouping. The counts for each syndrome grouping were autocorrelated and non-homogeneous Poisson. As such, the main methodological contribution of the article is the adaptation of Cumulative Sum (CUSUM) and Exponentially Weighted Moving Average (EWMA) plans for monitoring non-homogeneous counts. These plans were valid for small counts where the assumption of normally distributed one-day-ahead forecasts errors, typically used in other papers, breaks down. In addition, these adaptive plans have the advantage that control limits do not have to be trained for different syndrome groupings or aggregations of emergency departments. Conventional methods for signaling increases in syndrome grouping counts, Shewhart, CUSUM, and EWMA control charts of the standardized forecast errors were also examined. Shewhart charts were, at times, insensitive to shifts of interest. CUSUM and EWMA charts were only reasonable for large counts. We illustrate our methods with respiratory, influenza, diarrhea, and abdominal pain syndrome groupings.

Patching rainfall data using regression methods.: 1. Best subset selection, EM and pseudo-EM methods: theory

Abstract In this first paper in a set of three, the problem of patching missing values in rainfall records is described, together with some possible solutions. The techniques are confined to regression methods. Two approaches for patching are explored; that based on best subset selection and that based on the expectation-maximisation (EM) algorithm. The background theory and some mathematics are used to justify six different techniques. Comparisons of efficiency and practical implementation are deferred to the second and third papers.

An overview and perspective on social network monitoring

ABSTRACT In this expository article we give an overview of some statistical methods for the monitoring of social networks. We discuss the advantages and limitations of various methods as well as some relevant issues. One of our primary contributions is to give the relationships between network monitoring methods and monitoring methods in engineering statistics and public health surveillance. We encourage researchers in the industrial process monitoring area to work on developing and comparing the performance of social network monitoring methods. We also discuss some of the issues in social network monitoring and give a number of research ideas.

Early warning CUSUM plans for surveillance of negative binomial daily disease counts

Automated public health surveillance of disease counts for rapid outbreak, epidemic or bioterrorism detection using conventional control chart methods can be hampered by over-dispersion and background (‘in-control’) mean counts that vary over time. An adaptive cumulative sum (CUSUM) plan is developed for signalling unusually high incidence in prospectively monitored time series of over-dispersed daily disease counts with a non-homogeneous mean. Negative binomial transitional regression is used to prospectively model background counts and provide ‘one-step-ahead’ forecasts of the next days count. A CUSUM plan then accumulates departures of observed counts from an offset (reference value) that is dynamically updated using the modelled forecasts. The CUSUM signals whenever the accumulated departures exceed a threshold. The amount of memory of past observations retained by the CUSUM plan is determined by the offset value; a smaller offset retains more memory and is efficient at detecting smaller shifts. Our approach optimises early outbreak detection by dynamically adjusting the offset value. We demonstrate the practical application of the ‘optimal’ CUSUM plans to daily counts of laboratory-notified influenza and Ross River virus diagnoses, with particular emphasis on the steady-state situation (i.e. changes that occur after the CUSUM statistic has run through several in-control counts).

Home Telemonitoring of Vital Signs—Technical Challenges and Future Directions

The telemonitoring of vital signs from the home is an essential element of telehealth services for the management of patients with chronic conditions, such as congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), diabetes, or poorly controlled hypertension. Telehealth is now being deployed widely in both rural and urban settings, and in this paper, we discuss the contribution made by biomedical instrumentation, user interfaces, and automated risk stratification algorithms in developing a clinical diagnostic quality longitudinal health record at home. We identify technical challenges in the acquisition of high-quality biometric signals from unsupervised patients at home, identify new technical solutions and user interfaces, and propose new measurement modalities and signal processing techniques for increasing the quality and value of vital signs monitoring at home. We also discuss use of vital signs data for the automated risk stratification of patients, so that clinical resources can be targeted to those most at risk of unscheduled admission to hospital. New research is also proposed to integrate primary care, hospital, personal genomic, and telehealth electronic health records, and apply predictive analytics and data mining for enhancing clinical decision support.

A service oriented architecture for a health research data network

This paper reports on an architecture aimed at providing a technology platform for a new research facility, called the Health Research Data Network (HRDN). The two key features - custodial control over access and use of resources; and confidentiality protection integrated into a secure end-to-end system for data sharing and analysis - distinguish HRDN from other service oriented architectures for distributed data sharing and analysis.

Representing association classification rules mined from health data

An association classification algorithm has been developed to explore adverse drug reactions in a large medical transaction dataset with unbalanced classes. Rules discovered can be used to alert medical practitioners when prescribing drugs, to certain categories of patients, to potential adverse effects. We assess the rules using survival charts and propose two kinds of probability trees to present them. Both of them represent the risk of given adverse drug reaction for certain categories of patients in terms of risk ratios, which are familiar to medical practitioners. The first approach shows risk ratios when all rule conditions apply. The second presents the risk associated with a single risk factor with other parts of the rule identifying the cohort of the patient subpopulation. Thus, the probability trees can present clearly the risk of specific adverse drug reactions to prescribers.