Madhuri S. Mulekar

Does increased emergency medical services prehospital time affect patient mortality in rural motor vehicle crashes? A statewide analysis.

BACKGROUND Fatality rates from rural vehicular trauma are almost double those found in urban settings. It has been suggested that increased prehospital time is a factor that adversely affects fatality rates in rural vehicular trauma. By linking and analyzing Alabamas statewide prehospital data, emergency medical services (EMS) prehospital time was assessed for rural and urban vehicular crashes. METHODS An imputational methodology permitted linkage of data from police motor vehicle crash (MVC) and EMS records. MVCs were defined as rural or urban by crash location using the United States Census Bureau criteria. Areas within Alabama that fell outside the Census Bureau definition of urban were defined as rural. Prehospital data were analyzed to determine EMS response time, scene time, and transport time in rural and urban settings. RESULTS Over a 2-year period from January 2001 through December 2002, data were collected from EMS Patient Care Reports and police crash reports for the entire state of Alabama. By using an imputational methodology and join specifications, 45,763 police crash reports were linked to EMS Patient Care Reports. Of these, 34,341 (75%) were injured in rural settings and 11,422 (25%) were injured in urban settings. A total of 714 mortalities were identified, of which 611 (1.78%) occurred in rural settings and 103 (.90%) occurred in urban settings (P < .0001). When mortalities occurred, the mean EMS response time in rural settings was 10.67 minutes and 6.50 minutes in urban settings (P < .0001). When mortalities occurred, the mean EMS scene time in rural settings was 18.87 minutes and 10.83 minutes in urban settings (patients who were dead on scene and extrication patients were excluded from both settings) (P < .0001). When mortalities occurred, the mean EMS transport time in rural settings was 12.45 minutes and 7.43 minutes in urban settings (P < .0001). When mortalities occurred, the overall mean prehospital time in rural settings was 42.0 minutes and 24.8 minutes in urban settings (P < .0001). The mean EMS response time for rural MVCs with survivors was 8.54 minutes versus a mean of 10.67 minutes with mortalities (P < .0001). The mean EMS scene time for rural MVCs with survivors was 14.81 minutes versus 18.87 minutes with mortalities (patients who were dead on scene and extrication patients were excluded) (P = .0014). CONCLUSIONS Based on this statewide analysis of MVCs, increased EMS prehospital time appears to be associated with higher mortality rates in rural settings.

Prevalence of intraatrial thrombus in patients with atrial flutter.

In summary, left atrial thrombus occurs with disproportionately high frequency in patients hospitalized with atrial flutter. Male gender and a left ventricular ejection fraction < 40% are predictors of left atrial thrombus formation in such patients.

Loss of tumor suppressor Merlin in advanced breast cancer is due to post-translational regulation.

Background: The role of Merlin in breast cancer is unknown. Results: Merlin protein is degraded in advanced breast cancer due to osteopontin-initiated signaling. Conclusion: Merlin is regulated at the post-translational level in breast tumors. Significance: We have defined a functional role for Merlin in limiting breast tumor growth and elucidated the utility of Merlin as an important biomarker in breast cancer. Unlike malignancies of the nervous system, there have been no mutations identified in Merlin in breast cancer. As such, the role of the tumor suppressor, Merlin, has not been investigated in breast cancer. We assessed Merlin expression in breast cancer tissues by immunohistochemistry and by real-time PCR. The expression of Merlin protein (assessed immunohistochemically) was significantly decreased in breast cancer tissues (although the transcript levels were comparable) simultaneous with increased expression of the tumor-promoting protein, osteopontin (OPN). We further demonstrate that the loss of Merlin in breast cancer is brought about, in part, due to OPN-initiated Akt-mediated phosphorylation of Merlin leading to its proteasomal degradation. Restoring expression of Merlin resulted in reduced malignant attributes of breast cancer, characterized by reduced invasion, migration, motility, and impeded tumor (xenograft) growth in immunocompromised mice. The possibility of developing a model using the relationship between OPN and Merlin was tested with a logistic regression model applied to immunohistochemistry data. This identified consistent loss of immunohistochemical expression of Merlin in breast tumor tissues. Thus, we demonstrate for the first time a role for Merlin in impeding breast malignancy, identify a novel mechanism for the loss of Merlin protein in breast cancer, and have developed a discriminatory model using Merlin and OPN expression in breast tumor tissues.

Adaptive spam filtering using dynamic feature space

Unsolicited bulk e-mail, also known as spam, has been an increasing problem for the e-mail society. This paper presents a new spam filtering strategy that 1) uses a practical entropy coding technique, Huffman coding, to dynamically encode the feature space of e-mail collections over time and, 2) applies an online algorithm to adaptively enhance the learned spam concept as new e-mail data becomes available. The contributions of this work include a highly efficient spam filtering algorithm in which the input space is radically reduced to a single-dimension input vector, and an adaptive learning technique that is robust to vocabulary change, concept drifting and skewed data distribution. We compare our technique to several existing off-line learning techniques including support vector machine, naive Bayes, k-nearest neighbor, C4.5 decision tree, RBFNetwork, boosted decision tree and stacking, and demonstrate the effectiveness of our technique by presenting the experimental results on the e-mail data that is publicly available

Confidence interval estimation of overlap: equal means case

The estimators of the commonly used measures of overlap are known to be biased by an amount which depends on the unknown overlap. In general it is dicult to calculate the precision or bias of most ecological measures because there is no exact formula for the variance of the statistic and the sampling distribution is unknown (Dixon, 1993. In: Scheiner, S.M., Gurevitch, J. (Eds.), Design and Analysis of Ecological Experiments. Chapman & Hall, New York, pp. 290{318.) Two resampling techniques, namely, Jackknife and Bootstrap along with the Taylor series approximation and transformation method are considered for the construction of condence intervals. Three measures of overlap frequently used in quantitative ecology and considered in this study are Matusita’s measure , Morisita’s measure, , and Weitzman’s measure, . c 2000 Elsevier Science B.V. All rights reserved.

Increased Rural Vehicular Mortality Rates: Roadways with Higher Speed Limits or Excessive Vehicular Speed?

OBJECTIVE The purpose of this study was to assess whether higher roadway speed limits and excessive vehicular speed were contributing factors to increased rural vehicular mortality rates in the State of Alabama. METHODS During a 2-year period from January 2001 through December 2002, data were collected from Alabama police crash reports and EMS patient care reports. Police crash reports and EMS patient care reports were linked utilizing an imputational methodology. Vehicular speeds were estimated speeds extracted from police crash reports. Vehicular speeding was defined as estimated speeds greater than posted speed limits. RESULTS A total of 38,117 reports were linked. Of those, 30,260 (79%) and 7,857 (21%) were injured in rural and urban settings, respectively. The frequency of vehicular speeding was significantly higher in rural (18.8%) than in urban settings (9.4%) (p < 0.0001). At vehicular speeds less than 26 mph, mortality rates for occupants of speeding and nonspeeding vehicles were not significantly different in rural (1.68%, 0.82%) and urban (1.44%, 0.59%) settings (p = 0.78,1.0), respectively. On roads with posted speeds of 26 to 50 mph, mortality rates for occupants in speeding vehicles were not significantly different in rural (3.75%) and urban (2.23%) settings (p = 0.1360). For occupants of nonspeeding vehicles on roads with posted speeds of 26 to 50 mph, mortality rates were significantly greater in rural (0.72%) than in urban (0.35%) settings (p < 0.0032). On roads with posted speeds of 51 to 70 mph, mortality rates for occupants in speeding vehicles were not significantly different in rural (5.80%) and urban (4.95%) settings (p = 1.0). For occupants of nonspeeding vehicles on roads with posted speeds of 51 to 70 mph, mortality rates were significantly greater in rural (1.92%) than in urban (0.94%) settings (p = 0.01). CONCLUSIONS Vehicular speeding occurs with significantly higher frequency in rural settings. This imparts a greater overall vehicular mortality rate. At higher rates of speed, mortality rates for travel above the posted speed limit are similar in rural and urban settings; however, mortality rates for travel within the posted speed limit are greater in rural settings. This suggests factors beyond higher and excessive vehicular speed impart higher rates in rural settings.

How useful are approximations to mean and variance of the index of dissimilarity

Sociologists, demographers, and economists often use the index of dissimilarity, D, to describe the extent of racial, ethnic, spatial, or areal dissimilarity (or segregation) of different socio-economic groups. Derivation of the mean and variance of D to develop inference techniques has been a subject of studies for over three decades. Some have attempted to develop approximations for the mean and variance in order to simplify calculations. However, due to difficulties in deriving the measure of variation and complex nature of approximations, D is almost always employed in practice for descriptive rather than inferential purposes. In this study, simulation technique is used to compare different approaches for approximating the mean and variance of D and standardizing of D values. The usefulness of approximation is discussed using residential segregation data classified by race and personal income data classified by gender and age group. The results show that the amount of bias in approximations negates the usefulness of these approximations in standardization process and possibly inference.

Improving rural emergency medical service response time with global positioning system navigation

OBJECTIVE Rural emergency medical services (EMS) often serves expansive areas that many EMS personnel are unfamiliar with. EMS response time is increased in rural areas, which has been suggested as a contributing factor to increased mortality rates from motor vehicle crashes (MVCs) and nontraumatic emergencies. The purpose of this study was to assess the effect of a global positioning system (GPS) on rural EMS response time. METHODS GPS units were placed in ambulances of a rural EMS provider. The GPS units were set for fastest route (not shortest distance) to the scene that depends on traffic lights and posted road speed. During a 1-year period from September 2006 to August 2007, EMS response time and mileage to the scene were recorded for MVCs and other emergencies. Response times and mileage to the scene were then compared with data from the same EMS provider during a similar 1-year period when GPS technology was not used. EMS calls less than 1-mile were removed from both data sets because GPS was infrequently used for short travel distances. RESULTS During the 1-year period before utilization of GPS, 893 EMS calls greater than 1 mile were recorded and 791 calls recorded with GPS. The mean EMS response time for MVCs was 8.5 minutes without GPS and 7.6 minutes with GPS (p < 0.0001). When MVCs were matched for miles traveled, mean EMS response time without GPS was 13.7 minutes versus 9.9 minutes with GPS (p < 0.001). CONCLUSION GPS technology can significantly improve EMS response time to the scene of MVCs and nontraumatic emergencies.

Loss of N-Myc interactor promotes epithelial-mesenchymal transition by activation of TGF-β/SMAD signaling.

Epithelial–mesenchymal transition is one of the critical cellular programs that facilitate the progression of breast cancer to an invasive disease. We have observed that the expression of N-myc interactor (NMI) decreases significantly during progression of breast cancer, specifically in invasive and metastatic stages. Recapitulation of this loss in breast cell lines with epithelial morphology (MCF10A (non-tumorigenic) and T47D (tumorigenic)) by silencing NMI expression causes mesenchymal-like morphological changes in 3D growth, accompanied by upregulation of SLUG and ZEB2 and increased invasive properties. Conversely, we found that restoring NMI expression attenuated the mesenchymal attributes of metastatic breast cancer cells, accompanied by distinctly circumscribed 3D growth with basement membrane deposition and decreased invasion. Further investigations into the downstream signaling modulated by NMI revealed that NMI expression negatively regulates SMAD signaling, which is a key regulator of cellular plasticity. We demonstrate that NMI blocks TGF-β/SMAD signaling via upregulation of SMAD7, a negative feedback regulator of the pathway. We also provide evidence that NMI activates STAT signaling, which negatively modulates TGF-β/SMAD signaling. Taken together, our findings suggest that loss of NMI during breast cancer progression could be one of the driving factors that enhance the invasive ability of breast cancer by aberrant activation of TGF-β/SMAD signaling.

ADAPTIVE SPAM FILTERING USING DYNAMIC FEATURE SPACES

Unsolicited bulk e-mail, also known as spam, has been an increasing problem for the e-mail society. This paper presents a new spam filtering strategy that 1) uses a practical entropy coding technique, Huffman coding, to dynamically encode the feature space of the e-mail collected over time and, 2) applies an online algorithm to adaptively enhance the learned spam concept as new e-mail data becomes available. The contributions of this work include a highly efficient spam filtering algorithm in which the input space is radically reduced to a single-dimension input vector, and an adaptive learning technique that is robust to vocabulary change, concept drifting and skewed class distributions. We compare our technique with several existing off-line learning techniques including support vector machine, logistic regression, naive Bayes, k-nearest neighbor, C4.5 decision tree, RBFNetwork, boosted decision tree and stacking. We demonstrate the effectiveness of our technique by presenting the experimental results on the e-mail data that is publicly available. A more in-depth statistical analysis on the experimental results is also presented and discussed.