SEARCH · PubMed Health

Results for “Deep generative models”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 235 records · Page 13Linked to original sources

PubMed2005

A model of glucose transport and conversion to lactate in the renal medullary microcirculation.

In this study, we modeled mathematically the transport of glucose across renal medullary vasa recta and its conversion to lactate by anaerobic glycolysis. Uncertain parameter values were determined by seeking good agreement between predictions and experimental measurements of lactate generation rates, as well as glucose and lactate concentration ratios between the papilla and the corticomedullary junction; plausible kinetic rate constant and permeability values are summarized in tabular form. Our simulations indicate that countercurrent exchange of glucose from descending (DVR) to ascending vasa recta (AVR) in the outer medulla (OM) and upper inner medulla (IM) severely limits delivery to the deep inner medulla, thereby limiting medullary lactate generation. If the permeability to glucose of OMDVR and IMDVR is taken to be the same and equal to 4 x 10(-4) cm/s, the fraction of glucose that bypasses the IM is calculated as 54%; it is predicted as 37% if the presence of pericytes in OMDVR reduces the glucose permeability of these vessels by a factor of 2 relative to that of IMDVR. Our results also suggest that red blood cells (RBCs) act as a reservoir that reduces the bypass of glucose from DVR to AVR. The rate of lactate generation by anaerobic glycolysis of glucose supplied by blood from glomerular efferent arterioles is predicted to range from 2 to 8 nmol/s, in good agreement with lower estimates obtained from the literature (Bernanke D and Epstein FH. Am J Physiol 208: 541-545, 1965; Bartlett S, Espinal J, Janssens P, and Ross BD. Biochem J 219: 73-78, 1984).

Anaerobiosis↗

PubMed2002

Membrane orientation and position of the C2 domain from cPLA2 by site-directed spin labeling.

The C2 domain is a ubiquitous Ca(2+)-binding motif that triggers the membrane docking of many key signaling proteins during intracellular Ca(2+) signals. Site-directed spin labeling was carried out on the C2 domain of cytosolic phospholipase A(2) in order to determine the depth of penetration and orientation of the domain at the membrane interface. Membrane depth parameters, Phi, were obtained by EPR spectroscopy for a series of selectively spin-labeled C2 domain cysteine mutants, and for spin-labeled lipids and spin-labeled bacteriorhodopsin cysteine mutants. Values of Phi were combined with several other constraints, including the solution NMR structure, to generate a model for the position of the C2 domain at the membrane interface. This modeling yielded an empirical expression for Phi, which for the first time defines its behavior from the bulk aqueous phase to the center of the lipid bilayer. In this model, the backbones of both the first and third Ca(2+)-binding loops are inserted approximately 10 A into the bilayer, with residues inserted as deep as 15 A. The backbone of the second Ca(2+)-binding loop is positioned near the lipid phosphate, and the two beta-sheets of the C2 domain are oriented so that the individual strands make angles of 30-45 degrees with respect to the bilayer surface. Upon membrane docking, spin labels in the Ca(2+)-binding loops exhibit decreases in local motion, suggesting either changes in tertiary contacts due to protein conformational changes and/or interactions with lipid.

Calcium-Binding Proteins↗

PubMed2003

Cortical surface registration for image-guided neurosurgery using laser-range scanning.

In this paper, a method of acquiring intraoperative data using a laser range scanner (LRS) is presented within the context of model-updated image-guided surgery. Registering textured point clouds generated by the LRS to tomographic data is explored using established point-based and surface techniques as well as a novel method that incorporates geometry and intensity information via mutual information (SurfaceMI). Phantom registration studies were performed to examine accuracy and robustness for each framework. In addition, an in vivo registration is performed to demonstrate feasibility of the data acquisition system in the operating room. Results indicate that SurfaceMI performed better in many cases than point-based (PBR) and iterative closest point (ICP) methods for registration of textured point clouds. Mean target registration error (TRE) for simulated deep tissue targets in a phantom were 1.0 +/- 0.2, 2.0 +/- 0.3, and 1.2 +/- 0.3 mm for PBR, ICP, and SurfaceMI, respectively. With regard to in vivo registration, the mean TRE of vessel contour points for each framework was 1.9 +/- 1.0, 0.9 +/- 0.6, and 1.3 +/- 0.5 for PBR, ICP, and SurfaceMI, respectively. The methods discussed in this paper in conjunction with the quantitative data provide impetus for using LRS technology within the model-updated image-guided surgery framework.

Adult↗

PubMed2005

Influence of oil type on the amounts of acrylamide generated in a model system and in French fries.

Acrylamide formation was studied by use of a new heating methodology, based on a closed stainless steel tubular reactor. Different artificial potato powder mixtures were homogenized and subsequently heated in the reactor. This procedure was first tested for its repeatability. By use of this experimental setup, it was possible to study the acrylamide formation mechanism in the different mixtures, eliminating some variable physical and chemical factors during the frying process, such as heat flux and water evaporation from and oil ingress into the food. As a first application of this optimized heating concept, the influence on acrylamide formation of the type of deep-frying oil was investigated. The results obtained from the experiments with the tubular reactor were compared with standardized French fry preparation tests. In both cases, no significant difference in acrylamide formation could be found between the various heating oils applied. Consequently, the origin of the deep-frying vegetable oils did not seem to affect the acrylamide formation in potatoes during frying. Surprisingly however, when artificial mixtures did not contain vegetable oil, significantly lower concentrations of acrylamide were detected, compared to oil-containing mixtures.

Acrylamide↗

PubMed2006

Patient-specific analysis of the volume of tissue activated during deep brain stimulation.

Despite the clinical success of deep brain stimulation (DBS) for the treatment of movement disorders, many questions remain about its effects on the nervous system. This study presents a methodology to predict the volume of tissue activated (VTA) by DBS on a patient-specific basis. Our goals were to identify the intersection between the VTA and surrounding anatomical structures and to compare activation of these structures with clinical outcomes. The model system consisted of three fundamental components: (1) a 3D anatomical model of the subcortical nuclei and DBS electrode position in the brain, each derived from magnetic resonance imaging (MRI); (2) a finite element model of the DBS electrode and electric field transmitted to the brain, with tissue conductivity properties derived from diffusion tensor MRI; (3) VTA prediction derived from the response of myelinated axons to the applied electric field, which is a function of the stimulation parameters (contact, impedance, voltage, pulse width, frequency). We used this model system to analyze the effects of subthalamic nucleus (STN) DBS in a patient with Parkinson's disease. Quantitative measurements of bradykinesia, rigidity, and corticospinal tract (CST) motor thresholds were evaluated over a range of stimulation parameter settings. Our model predictions showed good agreement with CST thresholds. Additionally, stimulation through electrode contacts that improved bradykinesia and rigidity generated VTAs that overlapped the zona incerta/fields of Forel (ZI/H2). Application of DBS technology to various neurological disorders has preceded scientific characterization of the volume of tissue directly affected by the stimulation. Synergistic integration of clinical analysis, neuroimaging, neuroanatomy, and neurostimulation modeling provides an opportunity to address wide ranging questions on the factors linked with the therapeutic benefits and side effects of DBS.

Deep Brain Stimulation↗

PubMed2000

Expression of matrix metalloproteinase-2 (MMP-2) and vascular endothelial growth factor (VEGF) in inflammation-associated corneal neovascularization.

Matrix metalloproteinase-2 (MMP-2), matrix metalloproteinase-9 (MMP-9) and vascular endothelial growth factor (VEGF) are all implicated in the development of neovascularization. To investigate the possible role of these factors in corneal neovascularization we have analysed the expression of MMP-2, MMP-9 and VEGF in a rat model of inflammation-associated corneal neovascularization. In this model, corneal neovascularization was induced in Long-Evans rats by krypton laser photocoagulation whereafter eyes were enucleated at 1, 4, 7, 10 and 20 days. Slit-lamp biomicroscopy and histologic analysis revealed a gradual development of corneal neovascularization that peaked 7-10 days after treatment when newly formed vessels could be seen throughout the corneal surface reaching deep into the stroma. Antisense and sense riboprobes were generated using DNA complementary to MMP-2, MMP-9 and VEGF, and mRNA expression was analysed using in situ hybridization. The expression of MMP-2 and MMP-9 in untreated corneas was low or absent whereas VEGF was weakly expressed in the corneal epithelium. MMP-2 expression was increased during corneal neovascularization and was mainly localized to the cells infiltrating areas of new vessel formation. Many of these cells appeared to be inflammatory cells. VEGF expression had a similar overall distribution to MMP-2 during neovascularization with the exception that its expression in the corneal epithelium remained and even increased slightly. MMP-9 was prominently expressed at the border of regenerating corneal epithelium in areas with epithelial wounding but was not detected in the vascularized stroma. Together, the results of the present study support a role for MMP-2 and VEGF in inflammation-associated corneal neovascularization whereas MMP-9 instead appears to be involved in corneal epithelial wound-healing.

Animals↗

PubMed2000

Identification of variants and dual promoters of murine serine/threonine kinase KKIAMRE.

KKIAMRE is a serine/threonine protein kinase whose transcripts increase in the deep cerebellar nuclei of the rabbit after eyeblink conditioning, a model of associative learning and memory. We here characterized the expression, isoforms, and promoters of murine KKIAMRE gene. The expression of KKIAMRE was detected, by in situ hybridization and immunohistochemistry, in neurons in various brain regions including deep cerebellar nuclei. The gene spans approximately 40 kb and consists of 15 exons. Analysis of cDNA clones revealed multiple variants, having diversity in the putative carboxy-terminal regulatory domain, generated by alternative splicing and intraexonal termination. Furthermore, they had alternative 5' noncoding sequences. Primer extension, RNase protection, and transient expression assays revealed that two alternative promoters linked to distinct noncoding exons direct the expression of KKIAMRE. The gene was mapped on chromosomes 5 and 4 in mouse and human, respectively.

Amino Acid Sequence↗

PubMed2005

Microscopic theory for the quantum to classical crossover in chaotic transport.

We present a semiclassical theory for the scattering matrix S of a chaotic ballistic cavity at finite Ehrenfest time. Using a phase-space representation coupled with a multibounce expansion, we show how the Liouville conservation of phase-space volume decomposes S as S=S(cl) plus sign in circle S(qm). The short-time, classical contribution S(cl) generates deterministic transmission eigenvalues T=0 or 1, while quantum ergodicity is recovered within the subspace corresponding to the long-time, stochastic contribution S(qm). This provides a microscopic foundation for the two-phase fluid model, in which the cavity acts like a classical and a quantum cavity in parallel, and explains recent numerical data showing the breakdown of universality in quantum chaotic transport in the deep semiclassical limit. We show that the Fano factor of the shot-noise power vanishes in this limit, while weak localization remains universal.

Journal Article↗

PubMed1996

Infusion of phospholipid vesicles amplifies the local thrombotic response to TNF and anti-protein C into a consumptive response.

Inflammation often is considered a contributing factor to both thrombosis and disseminated intravascular coagulation. The molecular mechanisms that dictate which of these clinical manifestations will result from the inflammatory stimulus remain obscure. Bacterial infection and certain tumors are common initiators of the disseminated intravascular coagulant response. Complement activation resulting from bacterial infection shares with selected tumors the capacity to generate or release membrane particles that lack functional adhesion receptors and hence could circulate to amplify a disseminated intravascular coagulant response. We developed a model of venous thrombosis that resulted in localized thrombus formation without disseminated intravascular coagulation. The model involves infusion of tumor necrosis factor, blockade of protein C and a partial decrease in venous flow caused by ligation of the superficial femoral vein without obstruction of the deep formal vein. Infusion of phospholipid vesicles into this model resulted in amplification of a localized thrombotic response into a consumptive response. Seven different groups of animals were studied. The first three groups established the conditions necessary to produce deep vein thrombosis. The second four groups established the conditions necessary to produce disseminated intravascular coagulation. The infusion of phospholipid vesicles plus tumor necrosis factor and anti-protein C antibody resulted in consumption of fibrinogen, the production of thrombin/antithrombin complexes, a fall in platelet count, and venous thrombosis. Without ligation and catheterization phospholipid vesicles failed to produce the consumptive response. We conclude, therefore, that phospholipid vesicles can amplify a local thrombotic response into a consumptive response, and that vesiculation accompanying inflammation is one means by which localized coagulant activity may be amplified to produce disseminated intravascular coagulation.

Animals↗

PubMed2003

Defining a role for the subthalamic nucleus within operative theoretical models of subcortical participation in language.

OBJECTIVE: To investigate the effects of bilateral, surgically induced functional inhibition of the subthalamic nucleus (STN) on general language, high level linguistic abilities, and semantic processing skills in a group of patients with Parkinson's disease. METHODS: Comprehensive linguistic profiles were obtained up to one month before and three months after bilateral implantation of electrodes in the STN during active deep brain stimulation (DBS) in five subjects with Parkinson's disease (mean age, 63.2 years). Equivalent linguistic profiles were generated over a three month period for a non-surgical control cohort of 16 subjects with Parkinson's disease (NSPD) (mean age, 64.4 years). Education and disease duration were similar in the two groups. Initial assessment and three month follow up performance profiles were compared within subjects by paired t tests. Reliability change indices (RCI), representing clinically significant alterations in performance over time, were calculated for each of the assessment scores achieved by the five STN-DBS cases and the 16 NSPD controls, relative to performance variability within a group of 16 non-neurologically impaired adults (mean age, 61.9 years). Proportions of reliable change were then compared between the STN-DBS and NSPD groups. RESULTS: Paired comparisons within the STN-DBS group showed prolonged postoperative semantic processing reaction times for a range of word types coded for meanings and meaning relatedness. Case by case analyses of reliable change across language assessments and groups revealed differences in proportions of change over time within the STN-DBS and NSPD groups in the domains of high level linguistics and semantic processing. Specifically, when compared with the NSPD group, the STN-DBS group showed a proportionally significant (p<0.05) reliable improvement in postoperative scores achieved on the word test-revised (TWT-R), as well as a reliable decline (p<0.01) in the accuracy of lexical decisions about words with many meanings and a high degree of relatedness between meanings. CONCLUSIONS: Bilateral STN-DBS affects certain aspects of linguistic functioning, supporting a potential role for the STN in the mediation of language processes.

Aged↗

PubMed2005

Sources and effects of electrode impedance during deep brain stimulation.

OBJECTIVE: Clinical impedance measurements for deep brain stimulation (DBS) electrodes in human patients are normally in the range 500-1500 Omega. DBS devices utilize voltage-controlled stimulation; therefore, the current delivered to the tissue is inversely proportional to the impedance. The goals of this study were to evaluate the effects of various electrical properties of the tissue medium and electrode-tissue interface on the impedance and to determine the impact of clinically relevant impedance variability on the volume of tissue activated (VTA) during DBS. METHODS: Axisymmetric finite-element models (FEM) of the DBS system were constructed with explicit representation of encapsulation layers around the electrode and implanted pulse generator. Impedance was calculated by dividing the stimulation voltage by the integrated current density along the active electrode contact. The models utilized a Fourier FEM solver that accounted for the capacitive components of the electrode-tissue interface during voltage-controlled stimulation. The resulting time- and space-dependent voltage waveforms generated in the tissue medium were superimposed onto cable model axons to calculate the VTA. RESULTS: The primary determinants of electrode impedance were the thickness and conductivity of the encapsulation layer around the electrode contact and the conductivity of the bulk tissue medium. The difference in the VTA between our low (790 Omega) and high (1244 Omega) impedance models with typical DBS settings (-3 V, 90 mus, 130 Hz pulse train) was 121 mm3, representing a 52% volume reduction. CONCLUSIONS: Electrode impedance has a substantial effect on the VTA and accurate representation of electrode impedance should be an explicit component of computational models of voltage-controlled DBS. SIGNIFICANCE: Impedance is often used to identify broken leads (for values > 2000 Omega) or short circuits in the hardware (for values < 50 Omega); however, clinical impedance values also represent an important parameter in defining the spread of stimulation during DBS.

Brain↗

PubMed2004

Simultaneous integrated boost for breast cancer using IMRT: a radiobiological and treatment planning study.

PURPOSE: The purpose of this work is to explore the possibility of using intensity-modulated radiation therapy (IMRT) to deliver the boost dose to the tumor bed simultaneously with the whole-breast IMRT to reduce the radiation treatment time by 1-2 weeks. METHODS AND MATERIALS: The biologically effective dose (BED) for different treatments was calculated using the linear-quadratic (LQ) model with parameters previously derived for breast cancer from clinical data (alpha/beta = 10Gy, alpha = 0.3Gy(-1)). A potential doubling time of 15 days (from in vitro measurements) for breast cancer and a generic alpha/beta ratio of 3 Gy for normal tissues were used. A series of regimens that use IMRT as initial treatment and an IMRT simultaneous integrated boost (SIB) were derived using biologic equivalence to conventional schedules. Possible treatment plans with IMRT SIB to the tumor bed were generated for 2 selected breast patients, 1 with a shallow tumor and 1 with a deep-seated tumor. Plans with a simultaneous integrated electron boost were also generated for comparison. Dosimetric merits of these plans were evaluated based on dose volume histograms. RESULTS: A commonly used conventional treatment of 45 Gy (1.8 Gy x 25) to the whole breast and then a boost of 20 Gy (2 Gy x 10) is biologically equivalent to an alternative plan of 1.8 Gy x 25 to the whole breast with a 2.4 Gy x 25 SIB to the tumor bed. The new regime reduces treatment time from 7 to 5 weeks. For the patient with a deep-seated tumor, the IMRT plans reduce the volume of the breast that receives high doses (compared with the conventional photon boost plan) and provides good coverage of the target volumes. CONCLUSION: It is biologically and dosimetrically feasible to reduce the overall treatment time for breast radiotherapy by using an IMRT simultaneous integrated boost. For selected patient groups, IMRT plans with a new regimen can be equal to or better than conventional plans.

Breast Neoplasms↗

PubMed1995

SMART: a solvent-accessible triangulated surface generator for molecular graphics and boundary element applications.

An algorithm is presented for generating a representation of the solvent-accessible molecular surface as a smooth triangulated manifold. The algorithm, called SMART (SMooth moleculAR surface Triangulator), divides the contact and reentrant portions of the solvent-accessible molecular surface into curvilinear three-sided elements. In contrast to the author's earlier implementation of this general approach [Zauhar, R.J. and Morgan, R.S., J. Comput. Chem., 11 (1990) 603], the SMART algorithm defines elements directly on the appropriate geometric surface types (rather than using interpolation over cubic elements), and has special features to handle highly distorted regions which often appear in deep crevices and internal cavities. While the method is designed for use with boundary element techniques in continuum electrostatics, it can also be applied to the accurate computation of molecular surface areas and volumes, and the generation of shaded surfaces for display with interactive computer graphics.

Algorithms↗

PubMed2025

Unveiling novel antimicrobial peptides from the ruminant gastrointestinal microbiomes: A deep learning-driven approach yields an anti-MRSA candidate.

INTRODUCTION: Antimicrobial peptides (AMPs) present a promising avenue to combat the growing threat of antibiotic resistance. The ruminant gastrointestinal microbiome serves as a unique ecosystem that offers untapped potential for AMP discovery. OBJECTIVES: The aims of this study are to develop an effective methodology for the identification of novel AMPs from ruminant gastrointestinal microbiomes, followed by evaluating their antimicrobial efficacy and elucidating the mechanisms underlying their activity. METHODS: We developed a deep learning-based model to identify AMP candidates from a dataset comprising 120 metagenomes and 10,373 metagenome-assembled genomes derived from the ruminant gastrointestinal tract. Both in vivo and in vitro experiments were performed to examine and validate the antimicrobial activities of the AMP candidates that were selected through bioinformatic analysis and subsequently synthesized chemically. Additionally, molecular dynamics simulations were conducted to explore the action mechanism of the most potent AMP candidate. RESULTS: The deep learning model identified 27,192 potential secretory AMP candidates. Following bioinformatic analysis, 39 candidates were synthesized and tested. Remarkably, all synthesized peptides demonstrated antimicrobial activity against Staphylococcus aureus, with 79.5% showing effectiveness against multiple pathogens. Notably, Peptide 4, which exhibited the highest antimicrobial activity against methicillin-resistant Staphylococcus aureus (MRSA), confirmed this effect in a mouse model with wound infection, exhibiting a low propensity for resistance development and minimal cytotoxicity and hemolysis towards mammalian cells. Molecular dynamics simulations provided insights into the mechanism of Peptide 4, primarily its ability to disrupt bacterial cell membranes, leading to cell death. CONCLUSION: This study highlights the power of combining deep learning with microbiome research to uncover novel therapeutic candidates, paving the way for the development of next-generation antimicrobials like Peptide 4 to combat the growing threat of MRSA would infections. It also underscores the value of utilizing ruminant microbial resources.

Animals↗

PubMed2002

Utility of simultaneous brain, CSF and hyperintensity quantification in dementia.

Improved methods of quantifying MRI are needed to study brain-behavior relationships in dementia. Rating scales are variable; lesion-tracing approaches can be subjective and ignore atrophy; segmentation of MRI hyperintensities is complicated by partial volume effects; and hyperintense lesions in different anatomical areas may have different effects. The goal of this study was to extend existing segmentation approaches to include hyperintensities and to demonstrate the utility of simultaneously assessing atrophy and lesion compartments in dementia. A semi-automated method was applied to quantify brain and cerebrospinal fluid (CSF) compartments and to subclassify hyperintensities into periventricular, deep white matter, thalamic and basal ganglia compartments. Twenty MR scans from participants in an ongoing dementia study were used to generate intra- and inter-rater reliability estimates. High intra- and inter-class correlation coefficients (0.83-0.99) were obtained for all measures and the semi-automated measurements were highly correlated with traced volumes. Brain, CSF and specific lesion volumes were significantly correlated with neuropsychological functions. In models using only total hyperintensity volumes, the effects of lesion compartments (such as thalamic) were masked. Simultaneous quantification of atrophy and anatomically distinct hyperintensities is important for understanding cognitive impairments in dementia.

Aged↗

PubMed2002

Building coalitions: lessons from the Alliance for a Healthy New England.

Official pronouncements that the U.S. economy is in recession have confirmed the bad news that health care advocates began experiencing months ago. State budgets are deep in deficit. Medicaid programs are in jeopardy. And making the case for expanded health care access now seems harder than ever. There is good news, though: in the Alliance for a Healthy New England, health advocates may have found a new funding mechanism. Built on a broad-based coalition model, the Alliance is seeking to raise tobacco taxes in all six New England states, with the revenue generated to be used for tobacco control and expanded health care access. This States of Health looks at how the diverse organizations that make up the Alliance are learning to work together--and why working together is so valuable.

Health Care Coalitions↗

PubMedDate not supplied

[Rational structures in health education models: basics and systematization].

The different Health Education (HE) models appeared in the scientific literature are analyzed, trying to eliminate the confusion produced by its great diversity, applying a general and systematic point of view. Due to the relevance of that topic in the activities of Health Promotion in Primary Health Care it is urgent a deep reappraisal due the heterogeneity of scientific papers dealing with that topic. The curriculum, as the confluence of thought and action in Health Education, is the basic concept thanks to which it is possible to integrate both scientific logic, the biological one and that pertaining to the social sciences. Of particular importance have been the different paradigms that have emerged in the field of HE from the beginning of the present century: a first generation with a "normative" point of view, a second one orientated from positivistic bases, and a third generation adopting an hermeneutic and critic nature. This third generation of paradigms in HE has taken distances from the behaviouristic and cognitive perspectives being more critical and participative. The principal scientific contributors in the field of HE, internationals as well as spaniards are studied and classified. The main conclusions obtained from this Health Education paradigm controversy are referred to both aspects: 1) planning, programming and evaluating activities, and 2) models, qualitative and quantitative methodologies. Emphasis is given to the need of including Community Participation in all phases of the process in critic methodologies of HE. It is postulated the critic paradigm as the only one able to integrate the rest of the scientific approaches in Health Education.

Community Health Centers↗

PubMed2001

Temporal clustering of hemorrhages from untreated cavernous malformations of the central nervous system.

OBJECTIVE: Hemorrhages from cerebral cavernous malformations (CMs) sometimes seem to occur in closely spaced "clusters" interspersed with long hemorrhage-free intervals. Clustering of hemorrhages could affect retrospective assessments of radiosurgery efficacy in prevention of CM rehemorrhage. However, this empirical observation had not been tested quantitatively. To test whether CM hemorrhages tend to cluster, we reviewed pretreatment rebleeding rates after a first symptomatic hemorrhage in CM patients who later underwent surgery or radiosurgery. METHODS: We performed a retrospective review of 141 patients with CMs who presented with clinically overt hemorrhage, and who subsequently underwent surgery or proton beam radiosurgery during an 18-year period. Statistical models were used to analyze all events per person and identify potential variation in rebleeding risk with time after a previous hemorrhage. RESULTS: Sixty-three of 141 patients experienced a second hemorrhage before treatment; 16 had additional hemorrhages. Five hundred thirty-eight patient years elapsed between first hemorrhages and treatment. The cumulative incidence of a second hemorrhage after the first CM hemorrhage was 14% after 1 year and 56% after 5 years. During the first 2.5 years after a hemorrhage, the monthly rehemorrhage hazard was 2%. The risk then decreased spontaneously to less than 1% per month, which represents a 2.4-fold decline (P < 0.001). Rehemorrhage rates were higher in younger patients (P < 0.01), but not in females or in patients with deep lesions. Shorter intervals between successive hemorrhages did not predict higher subsequent rehemorrhage risk. CONCLUSION: The rehemorrhage rate from untreated CMs is high initially, and it decreases 2 to 3 years after a previous hemorrhage. This hazard pattern generates the observed temporal clustering of hemorrhages from untreated CMs.

Adolescent↗