Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735623
Navin Kasa, Andrew Dahbura, Charishma Ravoori, Stephen Adams
Credit card fraud is a problem that can cost banks billions of dollars annually, leading to increased incentives among financial institutions for the development of fast, effective and dynamic fraud detection systems. This research paper addresses credit card fraud detection through a semi-supervised approach, in which clusters of account profiles are created and used for modeling classifiers. Accounts are profiled based on their behavioral trends and clustered into similar groups. Groups are further identified as distinct customer segments based on purchase characteristics such as amount, frequency or distance. Random forest and XGBoost classifiers are trained on an entire sample and compared against classifiers trained at the transaction level across each cluster. This research concludes that the overall weighted performance of classifiers trained at the cluster level does not significantly outperform classifiers trained on the full sample. However, this research finds that clustering can be used to find meaningful groups of account holders that also have varying fraud rates across each cluster. Additionally, some classifiers trained on specific clusters yield significant improvements in performance over the baseline, whereas classifiers for other clusters do not perform as well as the baseline. This research also concludes that the optimal classifier for a given cluster varies by cluster, highlighting the potential for further development of new classifiers which may perform well on clusters that currently exhibit underperforming models.
{"title":"Improving Credit Card Fraud Detection by Profiling and Clustering Accounts","authors":"Navin Kasa, Andrew Dahbura, Charishma Ravoori, Stephen Adams","doi":"10.1109/SIEDS.2019.8735623","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735623","url":null,"abstract":"Credit card fraud is a problem that can cost banks billions of dollars annually, leading to increased incentives among financial institutions for the development of fast, effective and dynamic fraud detection systems. This research paper addresses credit card fraud detection through a semi-supervised approach, in which clusters of account profiles are created and used for modeling classifiers. Accounts are profiled based on their behavioral trends and clustered into similar groups. Groups are further identified as distinct customer segments based on purchase characteristics such as amount, frequency or distance. Random forest and XGBoost classifiers are trained on an entire sample and compared against classifiers trained at the transaction level across each cluster. This research concludes that the overall weighted performance of classifiers trained at the cluster level does not significantly outperform classifiers trained on the full sample. However, this research finds that clustering can be used to find meaningful groups of account holders that also have varying fraud rates across each cluster. Additionally, some classifiers trained on specific clusters yield significant improvements in performance over the baseline, whereas classifiers for other clusters do not perform as well as the baseline. This research also concludes that the optimal classifier for a given cluster varies by cluster, highlighting the potential for further development of new classifiers which may perform well on clusters that currently exhibit underperforming models.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116878307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735648
Alberto Abrisqueta, C. Bishop, Spencer P. Perryman, Luke M. Shoebotham, Jimmy Wang, M. Porter
In order to minimize incident duration, reduce secondary crashes, and improve travel time reliability along interstates and primary roadways, the Virginia Department of Transportation (VDOT) employs a fleet of vehicles known as Safety Service Patrols (SSPs) to detect traffic incidents and disruptions, assist stranded motorists, and perform short-term traffic control and scene management. At the time of their origin in the 1960s, the SSP routes, loose schedules detailing which road segments to patrol during a shift, were developed independently in each of VDOT's five regions, using largely anecdotal evidence. Fifty years later, and these routes have largely remained the same, having never been formally analyzed to check for system efficiency. This has left the agency concerned that the current routes may not be maximizing the patrollers' potential effect on public safety and traffic management. This paper thus develops a route optimization model via a genetic algorithm to optimally position the SSPs to minimize their average incident response time. This model is informed from five years' worth of traffic incident data in Virginia, collected and analyzed to generate a probability distribution estimating the concentration of incidents along varying route segments across each day of the week and time of day. Once layered onto the deliverable, an interactive dashboard depicting the locations of all the incidents, VDOT personnel will be able to visualize where the SSPs are in relation to the incidents. The SSP route optimization and visualization tools have the potential to improve system performance by respectively reducing average response time and allowing VDOT to gain a better understanding of traffic incident hotspots.
{"title":"Evaluation of VDOT's Safety Service Patrols to Improve Response to Incidents","authors":"Alberto Abrisqueta, C. Bishop, Spencer P. Perryman, Luke M. Shoebotham, Jimmy Wang, M. Porter","doi":"10.1109/SIEDS.2019.8735648","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735648","url":null,"abstract":"In order to minimize incident duration, reduce secondary crashes, and improve travel time reliability along interstates and primary roadways, the Virginia Department of Transportation (VDOT) employs a fleet of vehicles known as Safety Service Patrols (SSPs) to detect traffic incidents and disruptions, assist stranded motorists, and perform short-term traffic control and scene management. At the time of their origin in the 1960s, the SSP routes, loose schedules detailing which road segments to patrol during a shift, were developed independently in each of VDOT's five regions, using largely anecdotal evidence. Fifty years later, and these routes have largely remained the same, having never been formally analyzed to check for system efficiency. This has left the agency concerned that the current routes may not be maximizing the patrollers' potential effect on public safety and traffic management. This paper thus develops a route optimization model via a genetic algorithm to optimally position the SSPs to minimize their average incident response time. This model is informed from five years' worth of traffic incident data in Virginia, collected and analyzed to generate a probability distribution estimating the concentration of incidents along varying route segments across each day of the week and time of day. Once layered onto the deliverable, an interactive dashboard depicting the locations of all the incidents, VDOT personnel will be able to visualize where the SSPs are in relation to the incidents. The SSP route optimization and visualization tools have the potential to improve system performance by respectively reducing average response time and allowing VDOT to gain a better understanding of traffic incident hotspots.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133674823","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735642
K. Dawar, Ashwanth J. Samuel, Raf Alvarado
The task of manually annotating text is often tedious and error-prone. There is a strong need to digitize landscape history because a scalable, relational database with refined texts simply does not exist, ultimately limiting the pedagogical extent of this rich field. The data for the study conducted is a comprehensive textbook (544 pages) titled, “Landscape Design: A History of Landscape Architecture,” by Elizabeth Rogers. The Landscape Studies Initiative and Data Science Institute at the University of Virginia have partnered together to construct a SQL aided application (Flask) that will assist in deep annotation of scholarly texts. Our goal was to utilize machine learning techniques, specifically named entity recognition models (NER) and topic models (TM), not only to optimize the annotation process, but also to provide a fresh perspective on the text through a new index. In this paper, we will look at the training system, design, and architecture of several different NER models, including Python's spaCy, Stanford's Named Entity Recognizer, and IBM Bluemix's Natural Language Understanding tool, and compare their accuracies. Additionally, this paper aims to explore topic modeling from different tools and techniques, such as the Python libraries Gensim and Mallet in order to compare and contrast the relevance of those models to our dataset. The impact that these techniques have on the humanities fields can be astoundingly influential, but severely limited by the availability, size, and domain of the training dataset. Entity Recognition and Topic Modeling, as a result, are far from solved tasks: we will address some of the fundamental challenges that can prevent these systems from being robust and accurate.
{"title":"Comparing Topic Modeling and Named Entity Recognition Techniques for the Semantic Indexing of a Landscape Architecture Textbook","authors":"K. Dawar, Ashwanth J. Samuel, Raf Alvarado","doi":"10.1109/SIEDS.2019.8735642","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735642","url":null,"abstract":"The task of manually annotating text is often tedious and error-prone. There is a strong need to digitize landscape history because a scalable, relational database with refined texts simply does not exist, ultimately limiting the pedagogical extent of this rich field. The data for the study conducted is a comprehensive textbook (544 pages) titled, “Landscape Design: A History of Landscape Architecture,” by Elizabeth Rogers. The Landscape Studies Initiative and Data Science Institute at the University of Virginia have partnered together to construct a SQL aided application (Flask) that will assist in deep annotation of scholarly texts. Our goal was to utilize machine learning techniques, specifically named entity recognition models (NER) and topic models (TM), not only to optimize the annotation process, but also to provide a fresh perspective on the text through a new index. In this paper, we will look at the training system, design, and architecture of several different NER models, including Python's spaCy, Stanford's Named Entity Recognizer, and IBM Bluemix's Natural Language Understanding tool, and compare their accuracies. Additionally, this paper aims to explore topic modeling from different tools and techniques, such as the Python libraries Gensim and Mallet in order to compare and contrast the relevance of those models to our dataset. The impact that these techniques have on the humanities fields can be astoundingly influential, but severely limited by the availability, size, and domain of the training dataset. Entity Recognition and Topic Modeling, as a result, are far from solved tasks: we will address some of the fundamental challenges that can prevent these systems from being robust and accurate.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124103254","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735641
Andrew Gerin, Ben Clougherty, Jack Chang, Kal Fernlund, K. Macdonald, Rakshith Raghu, Thomas L. Sample, A. Rashap, S. Guerlain
The Key National Indicators Act of 2008 was passed by the United States Congress with the intent of creating a system of citizen-based statistics, which would function as a tool for informing both policymakers and the American public on the current outlook of our nation with respect to the physical, financial, and social well-being of individuals. This act allocated $77.5 million over a 10-year period to establish a commission that would collect and analyze data. However, over the past decade, the commission has failed to produce tangible results in the form of standardized metrics. Our research intends to carry out the initial goal of the Key National Indicators Act by following the basic principles proposed within it. To do so, we have used the five pillars of the US Constitution (Justice, Tranquility, Defense, Welfare, and Liberty) as broad indicators of how to measure progress. Within each pillar, specific metrics such as Poverty Rate (Welfare) and Recidivism Rate (Justice) were deemed significant through survey results. Data were assembled for each metric and weighted according to survey findings. After normalizing the results, sub-scores were established for each pillar and combined into an overall wellness score, the Promise America Wellness Score (PAWS). An analysis was carried out for data acquired from 1995 up until 2016 in order to understand the growth and decline of the United States over time. To effectively visualize and report these findings, a final product was designed in the form of a website. Through this website, the metrics and trends are made visible and interpretable, and are more easily marketed as a potentially valuable source of data for decision-makers. The website also includes areas for future steps in which well-being in the US would be able to be compared to well-being in other countries.
{"title":"Building a Foundation to Measure National Well-Being","authors":"Andrew Gerin, Ben Clougherty, Jack Chang, Kal Fernlund, K. Macdonald, Rakshith Raghu, Thomas L. Sample, A. Rashap, S. Guerlain","doi":"10.1109/SIEDS.2019.8735641","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735641","url":null,"abstract":"The Key National Indicators Act of 2008 was passed by the United States Congress with the intent of creating a system of citizen-based statistics, which would function as a tool for informing both policymakers and the American public on the current outlook of our nation with respect to the physical, financial, and social well-being of individuals. This act allocated $77.5 million over a 10-year period to establish a commission that would collect and analyze data. However, over the past decade, the commission has failed to produce tangible results in the form of standardized metrics. Our research intends to carry out the initial goal of the Key National Indicators Act by following the basic principles proposed within it. To do so, we have used the five pillars of the US Constitution (Justice, Tranquility, Defense, Welfare, and Liberty) as broad indicators of how to measure progress. Within each pillar, specific metrics such as Poverty Rate (Welfare) and Recidivism Rate (Justice) were deemed significant through survey results. Data were assembled for each metric and weighted according to survey findings. After normalizing the results, sub-scores were established for each pillar and combined into an overall wellness score, the Promise America Wellness Score (PAWS). An analysis was carried out for data acquired from 1995 up until 2016 in order to understand the growth and decline of the United States over time. To effectively visualize and report these findings, a final product was designed in the form of a website. Through this website, the metrics and trends are made visible and interpretable, and are more easily marketed as a potentially valuable source of data for decision-makers. The website also includes areas for future steps in which well-being in the US would be able to be compared to well-being in other countries.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124191972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735588
M. Hito, Logan Kuo, A. Newman, Marysia Serafin, Sarah Yang, G. Lewin
Many home automation systems use the wireless protocol Z-Wave for communication between devices in the network. Installers of such systems often face issues related to Z-Wave connectivity during installations, as devices must be placed so that Z-Wave signal strength between them is strong enough to add the device to the mesh network. The goal of this project is to design and prototype a device that measures Z-Wave signal strength to aid installers in the process of placing devices and troubleshooting device connectivity. While devices that measure Z-Wave signal strength currently exist, they are not optimized for installer use. Some devices require a proprietary license to operate, and some such as spectrum analyzers are expensive and have a large learning curve. The design's hardware consists of an antenna, software-defined radio (SDR), Raspberry Pi, LED array, activation button, battery, and housing. The antenna captures a modulated signal in the Z-Wave band and converts it to a voltage signal to be processed by the SDR. The voltage signal goes through an analog to digital converter in the SDR and is streamed into the Raspberry Pi as discrete voltage samples over time. The Raspberry Pi utilizes firmware called LibAirspy to interface to the SDR and forwards the voltage samples over time to a Python program, which then processes the data via downsampling, multiplication by the conjugate, and conversion to a decibel scale. The program finds the maximum power level measured over the interval and compared to a strength threshold. Whether the threshold has been met is displayed via LEDs on the tool. This tool will be used to decrease the resource intensivity of installing Z-Wave devices, leading to increased installer efficiency and lower call-center workload.
{"title":"Development of a Portable Z-Wave Signal Detector for Home Security Installations","authors":"M. Hito, Logan Kuo, A. Newman, Marysia Serafin, Sarah Yang, G. Lewin","doi":"10.1109/SIEDS.2019.8735588","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735588","url":null,"abstract":"Many home automation systems use the wireless protocol Z-Wave for communication between devices in the network. Installers of such systems often face issues related to Z-Wave connectivity during installations, as devices must be placed so that Z-Wave signal strength between them is strong enough to add the device to the mesh network. The goal of this project is to design and prototype a device that measures Z-Wave signal strength to aid installers in the process of placing devices and troubleshooting device connectivity. While devices that measure Z-Wave signal strength currently exist, they are not optimized for installer use. Some devices require a proprietary license to operate, and some such as spectrum analyzers are expensive and have a large learning curve. The design's hardware consists of an antenna, software-defined radio (SDR), Raspberry Pi, LED array, activation button, battery, and housing. The antenna captures a modulated signal in the Z-Wave band and converts it to a voltage signal to be processed by the SDR. The voltage signal goes through an analog to digital converter in the SDR and is streamed into the Raspberry Pi as discrete voltage samples over time. The Raspberry Pi utilizes firmware called LibAirspy to interface to the SDR and forwards the voltage samples over time to a Python program, which then processes the data via downsampling, multiplication by the conjugate, and conversion to a decibel scale. The program finds the maximum power level measured over the interval and compared to a strength threshold. Whether the threshold has been met is displayed via LEDs on the tool. This tool will be used to decrease the resource intensivity of installing Z-Wave devices, leading to increased installer efficiency and lower call-center workload.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133274413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735618
S. Clonts, Lia Cooley, P. Freitag, Bryan R. Soltis
Over four thousand bridges in the Virginia Department of Transportation's (VDOT) inventory are structurally deficient or obsolete. This project aimed to determine relevant bride deterioration factors in Virginia while providing a historical overview of Virginia's bridge ratings and deterioration rates. The team specifically analyzed the differences in the impact of deterioration factors by Virginia districts and bridge structure types. Using VDOT's inspection data from their Bridge Resource Management (BrM) system, we analyzed Virginia responsible bridge inspections from 2000–2015. Then we worked with a team from VDOT's Structure and Bridge Division to establish the most important factors that constitute an inspection record. We used a random forest algorithm to determine variable importance and relationships between important variables. We created 27 models in total which determined the relative influence of bridge-specific and environmental factors on bridges' overall condition ratings, as well as the bridge component condition ratings. Our models gave us an understanding of the relative importance of all factors analyzed across all bridge types. With 28 variables, the full model was 84.6% accurate on the test set. Our team further analyzed how ratings differ by district and bridge structure type. District trends were especially important to understand overall state consistency. Results confirmed factors such as bridge age, daily traffic, and relative location were influential in determining condition ratings between different districts and structure types. Limitations in analysis include inaccurate data for inspection ratings and bridge characteristics. Analysis is also ongoing, limiting the current definitive conclusions we can propose.
{"title":"Virginia Bridge Deterioration Factors","authors":"S. Clonts, Lia Cooley, P. Freitag, Bryan R. Soltis","doi":"10.1109/SIEDS.2019.8735618","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735618","url":null,"abstract":"Over four thousand bridges in the Virginia Department of Transportation's (VDOT) inventory are structurally deficient or obsolete. This project aimed to determine relevant bride deterioration factors in Virginia while providing a historical overview of Virginia's bridge ratings and deterioration rates. The team specifically analyzed the differences in the impact of deterioration factors by Virginia districts and bridge structure types. Using VDOT's inspection data from their Bridge Resource Management (BrM) system, we analyzed Virginia responsible bridge inspections from 2000–2015. Then we worked with a team from VDOT's Structure and Bridge Division to establish the most important factors that constitute an inspection record. We used a random forest algorithm to determine variable importance and relationships between important variables. We created 27 models in total which determined the relative influence of bridge-specific and environmental factors on bridges' overall condition ratings, as well as the bridge component condition ratings. Our models gave us an understanding of the relative importance of all factors analyzed across all bridge types. With 28 variables, the full model was 84.6% accurate on the test set. Our team further analyzed how ratings differ by district and bridge structure type. District trends were especially important to understand overall state consistency. Results confirmed factors such as bridge age, daily traffic, and relative location were influential in determining condition ratings between different districts and structure types. Limitations in analysis include inaccurate data for inspection ratings and bridge characteristics. Analysis is also ongoing, limiting the current definitive conclusions we can propose.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115681131","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/sieds.2019.8735590
{"title":"[Copyright notice]","authors":"","doi":"10.1109/sieds.2019.8735590","DOIUrl":"https://doi.org/10.1109/sieds.2019.8735590","url":null,"abstract":"","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129653718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735633
Jordan E. Blanchfield, Meredith T. Hargroves, Peter J. Keith, Maryanna C. Lansing, Lars Hälsing Nordin, Rachel C. Palmer, Shelby E. St. Louis, Allyson J. Will, W. Scherer, Nicholas J. Napoli
Individualized biometric data are being incorporated into training and competitions by many coaches and trainers to provide insights into athletic performance and physical fitness of their athletes. Currently, fitness tracking software provides coaches with minimal descriptive statistics on the collected biometric data, resulting in limited actionable outcomes. The collection of biometric data provides an opportunity to understand the variables that are indicative of athletic performance, and to create predictive models to determine appropriate training and in-game strategies. In order to develop these informative decision support tools, predictive frameworks have to address the correct performance metrics, control of subject-to-subject variability, handle data limitations, and maintain model interpretability. We demonstrate that the strenuousness of training sessions leading up to a competitive match has significant impact on the outcome of the game (win or loss) in continuous-play team sports. Specifically, a high cardiovascular training load two days prior to competition was predictive of a win. Additionally, we show that statistically significant differences exist in the physiological behaviors of different player positions. Analysis of several performance metrics also demonstrates that singular metrics or combinations of simple statistics do not directly relate to the outcome of a game, particularly in low-scoring sports such as field hockey or soccer.
{"title":"Developing Predictive Athletic Performance Models for Informative Training Regimens","authors":"Jordan E. Blanchfield, Meredith T. Hargroves, Peter J. Keith, Maryanna C. Lansing, Lars Hälsing Nordin, Rachel C. Palmer, Shelby E. St. Louis, Allyson J. Will, W. Scherer, Nicholas J. Napoli","doi":"10.1109/SIEDS.2019.8735633","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735633","url":null,"abstract":"Individualized biometric data are being incorporated into training and competitions by many coaches and trainers to provide insights into athletic performance and physical fitness of their athletes. Currently, fitness tracking software provides coaches with minimal descriptive statistics on the collected biometric data, resulting in limited actionable outcomes. The collection of biometric data provides an opportunity to understand the variables that are indicative of athletic performance, and to create predictive models to determine appropriate training and in-game strategies. In order to develop these informative decision support tools, predictive frameworks have to address the correct performance metrics, control of subject-to-subject variability, handle data limitations, and maintain model interpretability. We demonstrate that the strenuousness of training sessions leading up to a competitive match has significant impact on the outcome of the game (win or loss) in continuous-play team sports. Specifically, a high cardiovascular training load two days prior to competition was predictive of a win. Additionally, we show that statistically significant differences exist in the physiological behaviors of different player positions. Analysis of several performance metrics also demonstrates that singular metrics or combinations of simple statistics do not directly relate to the outcome of a game, particularly in low-scoring sports such as field hockey or soccer.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124410898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735614
S. Pagsuyoin, Jiayue Luo, Jana Latayan
Drug addiction exerts tremendous burdens on healthcare systems and societies. In the United States alone, opioids addiction cost over $500 billion in 2015 and over 50,000 people die each year from opioid-related overdoses. Health policymakers rely on drug surveillance data to combat this problem through intervention programs. In this preliminary study, we examine the temporal trends and relationship between opioids-related deaths and opioids prescription in Massachusetts. Opioids-related overdoses and fatalities have significantly increased in the state in recent years; currently, Massachusetts is identified as a hotspot in the opioids crisis in the country. By county yearly data (2013–2017) on opioids prescriptions, related deaths, and census were obtained and collated from different state agencies. Yearly deaths and prescription by county were divided by the corresponding yearly population to yield death rates and prescription rates, respectively. All yearly data were found to be normally distributed, thus Pearson correlation analysis was carried out to determine relationships between yearly death rates and prescription rates. We found significant ($mathrm{p} < 0.05$) yearly increases in death rates from 2013 to 2016; average prescription rates were highest in 2015 but trends were decreasing in the last two years. We did not find strong correlations ($vert mathrm{r}_{max}vert = 0.24, mathrm{p}_{min}=0.4$) between same-year death and prescription rates. We also did not find strong correlations between current year death rates and immediate previous year prescription rates ($vert mathrm{r}_{max}vert =0.17, mathrm{p}_{min}=0.6$). It should be emphasized that this study is limited in scope in that only two variables were considered (prescription and death); there are other contributing factors (e.g., drug access, affordability, illness, etc.) that determine opioids addiction that were not included in our analysis. Furthermore, prescription data alone do not provide sufficient information regarding actual per capita consumption rate (i.e., dosages, drug type, and number of refills are also needed). Nonetheless, research findings can provide some insights with respect to access to opioids, particularly for opioids sourced through illicit routes or misuse. For example, where death rates continue to increase despite declining prescription rates, intervention measures may include controlling potential other drugs sources.
{"title":"Temporal Trends in Opioids-Related Overdose Deaths and Prescription Rates in Massachusetts","authors":"S. Pagsuyoin, Jiayue Luo, Jana Latayan","doi":"10.1109/SIEDS.2019.8735614","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735614","url":null,"abstract":"Drug addiction exerts tremendous burdens on healthcare systems and societies. In the United States alone, opioids addiction cost over $500 billion in 2015 and over 50,000 people die each year from opioid-related overdoses. Health policymakers rely on drug surveillance data to combat this problem through intervention programs. In this preliminary study, we examine the temporal trends and relationship between opioids-related deaths and opioids prescription in Massachusetts. Opioids-related overdoses and fatalities have significantly increased in the state in recent years; currently, Massachusetts is identified as a hotspot in the opioids crisis in the country. By county yearly data (2013–2017) on opioids prescriptions, related deaths, and census were obtained and collated from different state agencies. Yearly deaths and prescription by county were divided by the corresponding yearly population to yield death rates and prescription rates, respectively. All yearly data were found to be normally distributed, thus Pearson correlation analysis was carried out to determine relationships between yearly death rates and prescription rates. We found significant ($mathrm{p} < 0.05$) yearly increases in death rates from 2013 to 2016; average prescription rates were highest in 2015 but trends were decreasing in the last two years. We did not find strong correlations ($vert mathrm{r}_{max}vert = 0.24, mathrm{p}_{min}=0.4$) between same-year death and prescription rates. We also did not find strong correlations between current year death rates and immediate previous year prescription rates ($vert mathrm{r}_{max}vert =0.17, mathrm{p}_{min}=0.6$). It should be emphasized that this study is limited in scope in that only two variables were considered (prescription and death); there are other contributing factors (e.g., drug access, affordability, illness, etc.) that determine opioids addiction that were not included in our analysis. Furthermore, prescription data alone do not provide sufficient information regarding actual per capita consumption rate (i.e., dosages, drug type, and number of refills are also needed). Nonetheless, research findings can provide some insights with respect to access to opioids, particularly for opioids sourced through illicit routes or misuse. For example, where death rates continue to increase despite declining prescription rates, intervention measures may include controlling potential other drugs sources.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126402870","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.1109/SIEDS.2019.8735649
J. Askey, Jaclyn Bellefeuille, S. Éskin, Anya Welch
Many Nigerian healthcare facilities do not have consistent access to electricity, which is required for many modern-day healthcare services and operations. The energy needs for such healthcare facilities can be supported by supplemental energy technologies to increase the reliability of electric power. These additions can include green or traditional technologies and would be established within any facility's preexisting electricity infrastructure. In this project, we developed a decision support tool that provides a list of candidate healthcare facilities to fund for supplemental energy generation in Nigeria, based on the effectiveness of implementing reliable energy technologies. The selection of energy technologies is unique to the region and type of facility. The combination of location-specific factors provides a recommendation for the most reliable energy technology. The decision support tool is designed to be used by health policy decision makers and Non-Governmental Organizations (NGOs) to prioritize funding efforts that are in the best interest of the population, as well as the environment, in developing countries where infrastructure development projects would provide the greatest impact. The tool recommends the optimal energy generation technology for each region, prioritizes the facilities based on access and coverage parameters, and then takes into account budget and planning constraints. The project aims to encourage developing countries, starting with Nigeria, to reevaluate their energy infrastructure needs in relation to health facilities.
{"title":"Decision Support Tool for Selecting Supplemental Energy Technologies for Healthcare Facilities in a Developing Country","authors":"J. Askey, Jaclyn Bellefeuille, S. Éskin, Anya Welch","doi":"10.1109/SIEDS.2019.8735649","DOIUrl":"https://doi.org/10.1109/SIEDS.2019.8735649","url":null,"abstract":"Many Nigerian healthcare facilities do not have consistent access to electricity, which is required for many modern-day healthcare services and operations. The energy needs for such healthcare facilities can be supported by supplemental energy technologies to increase the reliability of electric power. These additions can include green or traditional technologies and would be established within any facility's preexisting electricity infrastructure. In this project, we developed a decision support tool that provides a list of candidate healthcare facilities to fund for supplemental energy generation in Nigeria, based on the effectiveness of implementing reliable energy technologies. The selection of energy technologies is unique to the region and type of facility. The combination of location-specific factors provides a recommendation for the most reliable energy technology. The decision support tool is designed to be used by health policy decision makers and Non-Governmental Organizations (NGOs) to prioritize funding efforts that are in the best interest of the population, as well as the environment, in developing countries where infrastructure development projects would provide the greatest impact. The tool recommends the optimal energy generation technology for each region, prioritizes the facilities based on access and coverage parameters, and then takes into account budget and planning constraints. The project aims to encourage developing countries, starting with Nigeria, to reevaluate their energy infrastructure needs in relation to health facilities.","PeriodicalId":265421,"journal":{"name":"2019 Systems and Information Engineering Design Symposium (SIEDS)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132873797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：