Continuous electroencephalography in the medical intensive care unit.

OBJECTIVES: To examine predictors and the prognostic value of electrographic seizures (ESZs) and periodic epileptiform discharges (PEDs) in medical intensive care unit (MICU) patients without a primary acute neurologic condition. DESIGN: Retrospective study. SETTING: MICU in a university hospital. PATIENTS: A total of 201 consecutive patients admitted to the MICU between July 2004 and January 2007 without known acute neurologic injury and who underwent continuous electroencephalography monitoring (cEEG) for investigation of possible seizures or changes in mental status. INTERVENTION: None. MEASUREMENTS AND MAIN RESULTS: Median time from intensive care unit (ICU) admission to cEEG was 1 day (interquartile range 1-4). The majority of patients (60%) had sepsis as the primary admission diagnosis and 48% were comatose at the time of cEEG. Ten percent (n = 21) of patients had ESZs, 17% (n = 34) had PEDs, 5% (n = 10) had both, and 22% (n = 45) had either ESZs or PEDs. Seizures during cEEG were purely electrographic (no detectable clinical correlate) in the majority (67%) of patients. Patients with sepsis had a higher rate of ESZs or PEDs than those without sepsis (32% vs. 9%, p < 0.001). On multivariable analysis, sepsis at ICU admission was the only significant predictor of ESZs or PEDs (odds ratio 4.6, 95% confidence interval 1.9-12.7, p = 0.002). After controlling for age, coma, and organ dysfunction, the presence of ESZs or PEDs was associated with death or severe disability at hospital discharge (89% with ESZs or PEDs, vs. 39% if not; odds ratio 19.1, 95% confidence interval 6.3-74.6, p < 0.001). CONCLUSION: In this retrospective study of MICU patients monitored with cEEG, ESZs and PEDs were frequent, predominantly in patients with sepsis. Seizures were mainly nonconvulsive. Both seizures and periodic discharges were associated with poor outcome. Prospective studies are warranted to determine more precisely the frequency and clinical impact of nonconvulsive seizures and periodic discharges, particularly in septic patients.

Mapping of Environmental Data Using Kernel-Based Methods

Recently, kernel-based Machine Learning methods have gained great popularity in many data analysis and data mining fields: pattern recognition, biocomputing, speech and vision, engineering, remote sensing etc. The paper describes the use of kernel methods to approach the processing of large datasets from environmental monitoring networks. Several typical problems of the environmental sciences and their solutions provided by kernel-based methods are considered: classification of categorical data (soil type classification), mapping of environmental and pollution continuous information (pollution of soil by radionuclides), mapping with auxiliary information (climatic data from Aral Sea region). The promising developments, such as automatic emergency hot spot detection and monitoring network optimization are discussed as well.

Nocturnal hypoglycaemias in type 1 diabetic patients: what can we learn with continuous glucose monitoring?

AIM: In type 1 diabetic patients (T1DM), nocturnal hypoglycaemias (NH) are a serious complication of T1DM treatment; self-monitoring of blood glucose (SMBG) is recommended to detect them. However, the majority of NH remains undetected on an occasional SMBG done during the night. An alternative strategy is the Continuous glucose monitoring (CGMS), which retrospectively shows the glycaemic profile. The aims of this retrospective study were to evaluate the true incidence of NH in T1DM, the best SMBG time to predict NH, the relationship between morning hyperglycaemia and NH (Somogyi phenomenon) and the utility of CGMS to reduce NH. METHODS: Eighty-eight T1DM who underwent a CGMS exam were included. Indications for CGMS evaluation, hypoglycaemias and correlation with morning hyperglycaemias were recorded. The efficiency of CGMS to reduce the suspected NH was evaluated after 6-9 months. RESULTS: The prevalence of NH was 67% (32% of them unsuspected). A measured hypoglycaemia at bedtime (22-24 h) had a sensitivity of 37% to detect NH (OR=2.37, P=0.001), while a single measure < or =4 mmol/l at 3-hour had a sensitivity of 43% (OR=4.60, P<0.001). NH were not associated with morning hyperglycaemias but with morning hypoglycaemias (OR=3.95, P<0.001). After 6-9 months, suspicions of NH decreased from 60 to 14% (P<0.001). CONCLUSION: NH were highly prevalent and often undetected. SMBG at bedtime, which detected hypoglycaemia had sensitivity almost equal to that of 3-hour and should be preferred because it is easier to perform. Somogyi phenomenon was not observed. CGMS is useful to reduce the risk of NH in 75% of patients.