Sleep and sleep disorders during pregnancy and postpartum: The Life-ON study.

OBJECTIVE: to prospectively assess sleep and sleep disorders during pregnancy and postpartum in a large cohort of women. METHODS: multicenter prospective Life-ON study, recruiting consecutive pregnant women at a gestational age between 10 and 15 weeks, from the local gynecological departments. The study included home polysomnography performed between the 23rd and 25th week of pregnancy and sleep-related questionnaires at 9 points in time during pregnancy and 6 months postpartum. RESULTS: 439 pregnant women (mean age 33.7 ± 4.2 yrs) were enrolled. Poor quality of sleep was reported by 34% of women in the first trimester of pregnancy, by 46% of women in the third trimester, and by as many as 71% of women in the first month after delivery. A similar trend was seen for insomnia. Excessive daytime sleepiness peaked in the first trimester (30% of women), and decreased in the third trimester, to 22% of women. Prevalence of restless legs syndrome was 25%, with a peak in the third trimester of pregnancy. Polysomnographic data, available for 353 women, revealed that 24% of women slept less than 6 h, and 30.6% of women had a sleep efficiency below 80%. Sleep-disordered breathing (RDI≥5) had a prevalence of 4.2% and correlated positively with BMI. CONCLUSIONS: The Life-ON study provides the largest polysomnographic dataset coupled with longitudinal subjective assessments of sleep quality in pregnant women to date. Sleep disorders are highly frequent and distributed differently during pregnancy and postpartum. Routine assessment of sleep disturbances in the perinatal period is necessary to improve early detection and clinical management.

Computing short-interval transition matrices of a discrete-time Markov chain from partially observed data.

Markov chains constitute a common way of modelling the progression of a chronic disease through various severity states. For these models, a transition matrix with the probabilities of moving from one state to another for a specific time interval is usually estimated from cohort data. Quite often, however, the cohort is observed at specific times with intervals that may be greater than the interval of interest. The transition matrix computed then needs to be decomposed in order to estimate the desired interval transition matrix suited to the model. Although simple to implement, this method of matrix decomposition can yet result in an invalid short-interval transition matrix with negative or complex entries. In this paper, we present a method for computing short-interval transition matrices that is based on regularization techniques. Our method operates separately on each row of the invalid short-interval transition matrix aiming to minimize an appropriate distance measure. We test our method on various matrix structures and sizes, and evaluate its performance on a real-life transition model for HIV-infected individuals.

Automated test selection in decision-support systems: a case study in oncology.

Decision-support systems in medicine should be equipped with a facility that provides patient-tailored information about which test had best be performed in which phase of the patient's management. A decision-support system with a good test-selection facility may result in ordering fewer tests, decreasing financial costs, improving a patient's quality of life, and in an improvement of medical care in general. In close cooperation with two experts in oncology, we designed such a facility for a decision-support system for the staging of cancer of the oesophagus. The facility selects tests based upon a patient's health status and closely matches current routines. We feel that by extending our decision-support system with the facility, it provides further support for a patient's management and will be more interesting for use in daily medical practice. In this paper, we describe the test-selection facility that we designed for our decision-support system in oncology and present some initial results.

Generalised reliability characteristics for probabilistic networks.

BACKGROUND: In the medical domain, establishing a diagnosis typically amounts to reasoning about the unobservable truth, based upon a set of indirect observations from diagnostic tests. A diagnostic test may not be perfectly reliable, however. To avoid misdiagnosis, therefore, the reliability characteristics of the test should be taken into account upon reasoning. OBJECTIVE: In this paper, we address the issue of modelling the reliability characteristics of diagnostic tests in a probabilistic network. METHOD: To this end, we study the mathematical foundation of a test's characteristics and collate them with the probabilities required for a probabilistic network. RESULTS: We show that the standard reliability characteristics that are generally available from the literature have to be further detailed and stratified, for example by experts, before they can be included in a network. We demonstrate these modelling issues by means of a real-life probabilistic network in oncology.

On the use of vignettes for eliciting test-selection strategies.

Decision-support systems often include a strategy for selecting tests in their field of application. This strategy in essence captures procedural knowledge and serves to provide support for the reasoning processes involved. Generally, a test-selection strategy is offered in which tests are selected sequentially. For our field of application, we noticed that such a strategy would be an oversimplification, and decided to acquire knowledge about the actual strategy used by the experts. To this end, we composed a method that comprised an unstructured interview to gain general insight in the test-selection strategy used, and a subsequent structured interview, simulating daily practice through vignettes, to acquire full details. We used the method with two experts in our field of application and found that it closely fitted in with their daily practice and resulted in a large amount of detailed knowledge.