EEG spectral abnormalities and psychosis as predictors of cognitive and functional decline in probable Alzheimer's disease

We examined whether either psychotic features (e.g., delusions and hallucinations) or EEG abnormalities are associated with more rapid progression of Alzheimer's disease (AD). AD patients with psychosis have exhibited more EEG abnormalities than those without psychosis, and both abnormal EEG and psychosis have been noted to be predictors of functional and cognitive decline in AD. Ninety-five probable AD patients participating in a longitudinal study of dementia had an EEG and a semistructured psychiatric interview at baseline. Using EEG spectral analysis, we classified records as normal/abnormal based on the parasagittal mean frequency. Patients with abnormal EEGs were more functionally (e.g., Blessed Rating Scale for activities of daily living) and cognitively (e.g., Mini-Mental State) impaired than patients with normal EEG. AD patients with psychosis were only more functionally impaired than patients without psychosis. A two-factor analysis showed no interaction between abnormal EEG and psychosis. In addition, using a Cox proportional hazard model adjusted for age and education, the presence of an abnormal EEG or psychotic symptom at study entry was associated with higher risk of reaching severe cognitive and functional impairment during follow-up. Neither abnormal EEG nor the presence of psychosis predicted death. These results indicate that both abnormal EEG and psychosis are independent predictors of disease progression but not of physical survival.     

Triple check for antigen specificity of B cells during germinal centre reactions

Somatic hypermutation of the immunoglobulin variable genes during germinal reactions might permit the expansion of B-cell clones with unwanted (e.g. autoreactive) specificity. Here, Ernst Lindhout and colleagues propose three antigen-specific checkpoints that ensure the appropriate antigen specificity of activated B cells is maintained by regulating the activation, selection and further differentiation of B cells.     

8-epi PGF2 alpha generation during coronary reperfusion. A potential quantitative marker of oxidant stress in vivo

BACKGROUND: Myocardial reperfusion is believed to be associated with free radical injury. However, indexes of oxidative stress in vivo have been limited by their poor specificity and sensitivity. Isoprostanes are stable products of arachidonic acid formed in a nonenzymatic, free radical-catalyzed manner. We have developed a sensitive and specific assay for one of these compounds, 8-epi prostaglandin (PG) F2 alpha. METHODS AND RESULTS: To address its utility as an index of oxidative stress during coronary reperfusion, we measured urinary levels by gas chromatography/mass spectrometry in a canine model of coronary thrombolysis, in patients with acute myocardial infarction treated with thrombolytic therapy, and in patients after elective coronary artery bypass surgery. Urinary 8-epi PGF2 alpha was unchanged after circumflex artery occlusion in a canine model of coronary thrombolysis (n = 13; 437.2 +/- 56.4 versus 432.7 +/- 55.2 pmol/mmol creatinine) but increased significantly (P < .05) immediately after reperfusion (553.8 +/- 64.7 pmol/mmol). Urinary levels were increased (P < .001) in patients (n = 12) with acute myocardial infarction given lytic therapy (265.8 +/- 40.8 pmol/mmol) compared with age-matched control subjects (n = 20; 91.5 +/- 11.8 pmol/mmol) and patients with stable coronary disease (n = 20; 95.7 +/- 6.3 pmol/mmol). Preoperative levels rose from 113.2 +/- 11.8 to 248.2 +/- 86.3 pmol/mmol at 30 minutes into revascularization to 332.2 +/- 82.6 pmol/mmol by 15 minutes after global myocardial reperfusion (P < .05) and dropped to 181.2 +/- 50.4 pmol/mmol at 30 minutes and 120.2 +/- 9.9 pmol/mmol at 24 hours after bypass surgery (n = 5). Corresponding changes in spin adduct formation, found with electron paramagnetic resonance, were noted in 2 patients. CONCLUSIONS: These data support the hypothesis that free radical generation occurs during myocardial reperfusion. Measurement of isoprostane production may serve as a noninvasive index of oxidative stress.     

Adherence to pharmacologic management of hypertension

Adherence to pharmacologic therapy of hypertension is low (in the range of 50-70%) and has important implications both for blood pressure control and cardiovascular complications. Based on a review of the literature using the levels of evidence grading technique, determinants of adherence to the pharmacologic therapy of hypertension have been assessed. Additionally, interventions to improve compliance were evaluated. Patient-centred, health care provider-centred and drug-specific factors have all been shown to affect adherence rates. We conclude that the extent of adherence to pharmacologic therapy is modifiable. Measurable improvements in adherence can be obtained from simplified medication regimens and a combination of behaviour strategies, including the tailoring of pill-taking to patients' daily habits and rituals, the advocacy of self-monitoring of pills and blood pressure, and the institution of reward systems.     

Moment arms and lengths of human upper limb muscles as functions of joint angles

Modeling of musculoskeletal structures requires accurate data on anatomical parameters such as muscle lengths (MLs), moment arms (MAs) and those describing the upper limb position. Using a geometrical model of planar arm movements with three degrees of freedom, we present, in an analytical form, the available information on the relationship between MAs and MLs and joint angles for thirteen human upper limb muscles. The degrees of freedom included are shoulder flexion/extension, elbow flexion/extension, and either wrist flexion/extension (the forearm in supination) or radial/ulnar deviation (the forearm in mid-pronation). Previously published MA/angle curves were approximated by polynomials. ML/angle curves were obtained by combining the constant values of MLs (defined by the distance between the origin and insertion points for a specific upper limb position) with a variable part obtained by multiplying the MA (joint radius) and the joint angle. The MAs of the prime wrist movers in radial/ulnar deviation were linear functions of the joint angle (R2 > or = 0.9954), while quadratic polynomials accurately described their MAs during wrist flexion/extensions. The relationship between MAs and the elbow angle was described by 2nd, 3rd or 5th-order polynomials (R2 > or = 0.9904), with a lesser quality of fit for the anconeus (R2 = 0.9349). In the full range of angular displacements, the length of wrist, elbow and shoulder muscles can change by 8.5, 55 and 200%, respectively.     

Learning Automata from Ordered Examples

Connectionist learning models have had considerable empirical success, but it is hard to characterize exactly what they learn. The learning of finite-state languages (FSL) from example strings is a domain which has been extensively studied and might provide an opportunity to help understand connectionist learning. A major problem is that traditional FSL learning assumes the storage of all examples and thus violates connectionist principles. This paper presents a provably correct algorithm for inferring any minimum-state deterministic finite-state automata (FSA) from a complete ordered sample using limited total storage and without storing example strings. The algorithm is an iterative strategy that uses at each stage a current encoding of the data considered so far, and one single sample string. One of the crucial advantages of our algorithm is that the total amount of space used in the course of learning for encoding any finite prefix of the sample is polynomial in the size of the inferred minimum state deterministic FSA. The algorithm is also relatively efficient in time and has been implemented. More importantly, there is a connectionist version of the algorithm that preserves these properties. The connectionist version requires much more structure than the usual models and has been implemented using the Rochester Connectionist Simulator. We also show that no machine with finite working storage can iteratively identify the FSL from arbitrary presentations.