Cristiana Larizza

Dynamic conditional independence models and Markov chain Monte Carlo methods

Abstract In dynamic statistical modeling situations, observations arise sequentially, causing the model to expand by progressive incorporation of new data items and new unknown parameters. For example, in clinical monitoring, patients and data arrive sequentially, and new patient-specific parameters are introduced with each new patient. Markov chain Monte Carlo (MCMC) might be used for continuous updating of the evolving posterior distribution, but would need to be restarted from scratch at each expansion stage. Thus MCMC methods are often too slow for real-time inference in dynamic contexts. By combining MCMC with importance resampling, we show how real-time sequential updating of posterior distributions can be effected. The proposed dynamic sampling algorithms use posterior samples from previous updating stages and exploit conditional independence between groups of parameters to allow samples of parameters no longer of interest to be discarded, such as when a patient dies or is discharged. We apply the ...

A telemedicine support for diabetes management: the T-IDDM project

In the context of the EU funded Telematic Management of Insulin-Dependent Diabetes Mellitus (T-IDDM) project, we have designed, developed and evaluated a telemedicine system for insulin dependent diabetic patients management. The system relies on the integration of two modules, a Patient Unit (PU) and a Medical Unit (MU), able to communicate over the Internet and the Public Switched Telephone Network. Using the PU, patients are allowed to automatically download their monitoring data from the blood glucose monitoring device, and to send them to the hospital data-base; moreover, they are supported in their every day self monitoring activity. The MU provides physicians with a set of tools for data visualization, data analysis and decision support, and allows them to send messages and/or therapeutic advice to the patients. The T-IDDM service has been evaluated through the application of a formal methodology, and has been used by European patients and physicians for about 18 months. The results obtained during the project demonstration, even if obtained on a pilot study of 12 subjects, show the feasibility of the T-IDDM telemedicine service, and seem to substantiate the hypothesis that the use of the system could present an advantage in the management of insulin dependent diabetic patients, by improving communications and, potentially, clinical outcomes.

Data mining with Temporal Abstractions: learning rules from time series

A large volume of research in temporal data mining is focusing on discovering temporal rules from time-stamped data. The majority of the methods proposed so far have been mainly devoted to the mining of temporal rules which describe relationships between data sequences or instantaneous events and do not consider the presence of complex temporal patterns into the dataset. Such complex patterns, such as trends or up and down behaviors, are often very interesting for the users. In this paper we propose a new kind of temporal association rule and the related extraction algorithm; the learned rules involve complex temporal patterns in both their antecedent and consequent. Within our proposed approach, the user defines a set of complex patterns of interest that constitute the basis for the construction of the temporal rule; such complex patterns are represented and retrieved in the data through the formalism of knowledge-based Temporal Abstractions. An Apriori-like algorithm looks then for meaningful temporal relationships (in particular, precedence temporal relationships) among the complex patterns of interest. The paper presents the results obtained by the rule extraction algorithm on a simulated dataset and on two different datasets related to biomedical applications: the first one concerns the analysis of time series coming from the monitoring of different clinical variables during hemodialysis sessions, while the other one deals with the biological problem of inferring relationships between genes from DNA microarray data.

M-HTP: A system for monitoring heart transplant patients

A computer-based assistant for monitoring a patients clinical course requires the use of tools able to handle temporal issues. Thus, methodologies coming from two historically distinct worlds need to be combined: the traditional world of Data Base Management Systems (DBMS) and the world of Knowledge-Based Systems (KBS). This paper describes an intelligent system designed to assist the clinical staff in the management of a monitoring protocol for infections in heart transplant recipients. The system consists of a DBMS designed for the management of patient clinical data and of a KBS which is capable of reasoning about the large amount of data and embodied in a temporal model based on time-points and intervals. Moreover, the system aims at providing a synthetic view of a patients clinical history and some diagnostic and therapeutic suggestions. The KBS retrieves findings stored in the data base and creates a complex taxonomy of objects representing a Temporal Network of important events and episodes noted in the patient history; then, from this temporal representation, it develops its reasoning based on medical knowledge represented using frames and production rules. The system is implemented on a Fourth Generation System tool (4GS) and a KBS shell, both running on an IBM PC AT compatible platform.

A unified approach for modeling longitudinal and failure time data, with application in medical monitoring

This paper considers biomedical problems in which a sample of subjects, for example clinical patients, is monitored through time for purposes of individual prediction. Emphasis is on situations in which the monitoring generates data both in the form of a time series and in the form of events (development of a disease, death, etc.) observed on each subject over specified intervals of time. A Bayesian approach to the combined modeling of both types of data for purposes of prediction is presented. The proposed method merges ideas of Bayesian hierarchical modeling, nonparametric smoothing of time series data, survival analysis, and forecasting into a unified framework. Emphasis is on flexible modeling of the time series data based on stochastic process theory. The use of Markov chain Monte Carlo simulation to calculate the predictions of interest is discussed. Conditional independence graphs are used throughout for a clear presentation of the models. An application in the monitoring of transplant patients is presented.

Protocol-based reasoning in diabetic patient management

We propose a system for teleconsultation in Insulin Dependent Diabetes Mellitus (IDDM) management, accessible through the use of the net. The system is able to collect monitoring data, to analyze them through a set of tools, and to suggest a therapy adjustment in order to tackle the identified metabolic problems and to fit the patients needs. The therapy revision has been implemented through the Episodic Skeletal Planning Methodi, it generates an advice and employs it to modify the current therapeutic protocol, presenting to the physician a set of feasible solutions, among which she can choose the new one.

A General Framework for Building Patient Monitoring Systems

This paper describes a general framework designed to assist the physician in the management of patients long-term monitoring. It exploits a temporal model based on the temporal primitives time-point and interval and provides powerful mechanisms performing temporal abstractions and temporal reasoning that can be used to assess the patient clinical evolution in various medical domains. The framework is integrated into a clinical workstation providing several tools designed to assist the clinical staff in the management of the patients records and in the definition of the domain-specific knowledge.

Going mobile with a multiaccess service for the management of diabetic patients.

Background: Diabetes mellitus is one of the chronic diseases exploiting the largest number of telemedicine systems. Our research group has been involved since 1996 in two projects funded by the European Union proposing innovative architectures and services according to the best current medical practices and advances in the information technology area. Method: We propose an enhanced architecture for telemedicine giving rise to a multitier application. The lower tier is represented by a mobile phone hosting the patient unit able to acquire data and provide first-level advice to the patient. The patient unit also facilitates interaction with the health care center, representing the higher tier, by automatically uploading data and receiving back any therapeutic plan supplied by the physician. On the patients side the mobile phone exploits Bluetooth technology and therefore acts as a hub for a wireless network, possibly including several devices in addition to the glucometer. Results: A new system architecture based on mobile technology is being used to implement several prototypes for assessing its functionality. A subsequent effort will be undertaken to exploit the new system within a pilot study for the follow-up of patients cared at a major hospital located in northern Italy. Conclusion: We expect that the new architecture will enhance the interaction between patient and caring physician, simplifying and improving metabolic control. In addition to sending glycemic data to the caring center, we also plan to automatically download the therapeutic protocols provided by the physician to the insulin pump and collect data from multiple sensors.

TA-clustering: Cluster analysis of gene expression profiles through Temporal Abstractions

This paper describes a new technique for clustering short time series of gene expression data. The technique is a generalization of the template-based clustering and is based on a qualitative representation of profiles which are labelled using trend Temporal Abstractions (TAs); clusters are then dynamically identified on the basis of this qualitative representation. Clustering is performed in an efficient way at three different levels of aggregation of qualitative labels, each level corresponding to a distinct degree of qualitative representation. The developed TA-clustering algorithm provides an innovative way to cluster gene profiles. We show the developed method to be robust, efficient and to perform better than the standard hierarchical agglomerative clustering approach when dealing with temporal dislocations of time series. Results of the TA-clustering algorithm can be visualized as a three-level hierarchical tree of qualitative representations and as such easy to interpret. We demonstrate the utility of the proposed algorithm on a set of two simulated data sets and on a study of gene expression data from S. cerevisiae.

Regulation of Iron Metabolism

Abstract A mathematical model of iron metabolism is presented. It comprises the following iron pools within the body: transferrin-bound iron in the plasma.iron in circulating red cells and their bone marrow precursors, iron in mucosal, parenchimal and reticuloendothelial cells. The control exerted by a hormone, called erythropoietin, on bone marrow utilisation of iron for hemoglobin synthesis is taken into account. The model so obtained consists of a system of differential equations of retarded type. Most model parameters can be estimated from radiotracer experiments, others can be measured and numerical values can be assigned to the remaining ones making few reasonable assumptions consistent with the available physiological knowledge. Iron metabolism behavior under different therapeutical treatments was simulated. Model predictions were compared to experimental data collected in clinical routine.