Pub Date : 2022-04-28DOI: 10.1109/sieds55548.2022.9799330
Jordan Hiatt, D. Howe, Lauren Neal
At present, visual neuroscientists must employ an inefficient, time-intensive process to study the ways in which various types of neurons react to characteristics of a visual stimulus; the standard procedure requires specifying and monitoring a single cell type per individual microscopy recording. This research paper proposes an alternative method: utilize a supervised classification algorithm to distinguish between several cell types – based on the cells’ behavior and response to stimuli – in the context of a single recording. This allows researchers to record multiple cell types at once and, subsequently, classify them by type for further analysis. For this classifier, the neuronal spatial footprints and neuronal temporal activity are extracted from raw microscopy recordings using constrained nonnegative matrix factorization. From these data, neuronal features are engineered for the classifier, which-along with features engineered from the visual stimulus corresponding to the neuronal activity-are used by various models to predict the cell type of the recorded neurons. Several algorithms are tested to compare their classification performance, including random forest classifiers, neural networks, and K-nearest neighbors classifiers. This research concludes that the relationship between stimulus and fluorescent response is a moderate predictor of cell type. We develop a cell type classification model that leverages one-hot encoding and engineering of visual stimulus and fluorescent response features, sliding time/frame windows, and dimensionality reduction to generate inputs in a model to classify multiple neuronal cell types in a single microscopy recording. We originally hypothesized that the K-nearest neighbors and/or neural network implementations would produce the strongest classification performance due to the algorithms’ ability to flexibly fit nonlinear feature spaces. Due to the imbalanced nature of the dataset, with five classes total and one class making up nearly 50% of the data, balanced accuracy is a better indicator of model performance than accuracy. Classifying cells via random chance would yield a balanced accuracy of 20%. Our best cell type classifier, a convolutional neural network optimized for time series classification, gives us an accuracy score of 70.6% and balanced accuracy of 53.7%.
{"title":"Improving Visual Neuroscience Cell Type Classification with Supervised Machine Learning*","authors":"Jordan Hiatt, D. Howe, Lauren Neal","doi":"10.1109/sieds55548.2022.9799330","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799330","url":null,"abstract":"At present, visual neuroscientists must employ an inefficient, time-intensive process to study the ways in which various types of neurons react to characteristics of a visual stimulus; the standard procedure requires specifying and monitoring a single cell type per individual microscopy recording. This research paper proposes an alternative method: utilize a supervised classification algorithm to distinguish between several cell types – based on the cells’ behavior and response to stimuli – in the context of a single recording. This allows researchers to record multiple cell types at once and, subsequently, classify them by type for further analysis. For this classifier, the neuronal spatial footprints and neuronal temporal activity are extracted from raw microscopy recordings using constrained nonnegative matrix factorization. From these data, neuronal features are engineered for the classifier, which-along with features engineered from the visual stimulus corresponding to the neuronal activity-are used by various models to predict the cell type of the recorded neurons. Several algorithms are tested to compare their classification performance, including random forest classifiers, neural networks, and K-nearest neighbors classifiers. This research concludes that the relationship between stimulus and fluorescent response is a moderate predictor of cell type. We develop a cell type classification model that leverages one-hot encoding and engineering of visual stimulus and fluorescent response features, sliding time/frame windows, and dimensionality reduction to generate inputs in a model to classify multiple neuronal cell types in a single microscopy recording. We originally hypothesized that the K-nearest neighbors and/or neural network implementations would produce the strongest classification performance due to the algorithms’ ability to flexibly fit nonlinear feature spaces. Due to the imbalanced nature of the dataset, with five classes total and one class making up nearly 50% of the data, balanced accuracy is a better indicator of model performance than accuracy. Classifying cells via random chance would yield a balanced accuracy of 20%. Our best cell type classifier, a convolutional neural network optimized for time series classification, gives us an accuracy score of 70.6% and balanced accuracy of 53.7%.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131017372","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9841477
Taylor Yeazitzis, Kristin Weger, J. Clerkin, Bryan L. Mesmer
For the design of complex technological systems, the use of system architecture, a type of conceptual model, is being considered as a promising solution for formalizing and communicating the structure, behavior, and views of subsystems and components. A system architecture is used to enable the design of a system that includes both software and hardware; however, during conceptualization of a system architecture, heuristics, which are also referred to as mental shortcuts, are used and sometimes even encouraged to save mental energy during a decision-making process. One of the challenges inherent in developing a system architecture is producing the architecture with minimal biases. Biases may introduce unwanted influences on the design or conceptual model of a system. Presently, there is a lack of research regarding the heuristics or subsequent biases that may be present within a system architecture process. The purpose of this research is to better understand influential factors on the architecting process by conducting a literature review on heuristics and biases that may be found within system architecture. Specifically, this research will focus on the three main heuristics of availability, anchoring and adjustment, and representativeness as well as biases associated with each. Finally, future avenues for research are suggested and elaborated to further practitioners’ understanding of heuristics and biases within system architecture.
{"title":"Heuristics and Biases in System Architecture","authors":"Taylor Yeazitzis, Kristin Weger, J. Clerkin, Bryan L. Mesmer","doi":"10.1109/sieds55548.2022.9841477","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9841477","url":null,"abstract":"For the design of complex technological systems, the use of system architecture, a type of conceptual model, is being considered as a promising solution for formalizing and communicating the structure, behavior, and views of subsystems and components. A system architecture is used to enable the design of a system that includes both software and hardware; however, during conceptualization of a system architecture, heuristics, which are also referred to as mental shortcuts, are used and sometimes even encouraged to save mental energy during a decision-making process. One of the challenges inherent in developing a system architecture is producing the architecture with minimal biases. Biases may introduce unwanted influences on the design or conceptual model of a system. Presently, there is a lack of research regarding the heuristics or subsequent biases that may be present within a system architecture process. The purpose of this research is to better understand influential factors on the architecting process by conducting a literature review on heuristics and biases that may be found within system architecture. Specifically, this research will focus on the three main heuristics of availability, anchoring and adjustment, and representativeness as well as biases associated with each. Finally, future avenues for research are suggested and elaborated to further practitioners’ understanding of heuristics and biases within system architecture.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130839532","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799374
N. Barrington, C. Gayle, E. Hensien, G. Ko, M. Lin, S. Palnati, G. J. Gerling
Advances in user interactivity in digital entertainment offer the potential to engage users beyond traditional passive and solitary experiences. Indeed, a high level of interactivity is inherent in tasks involving cooking, auto repair, and home improvement, all of which require users to complete multiple, detailed, and interdependent steps. Such tasks also require access to visual and audio instructions while a user's hands are engaged in a primary, physical task, and are often conducted in unique locations, e.g., kitchen or garage. This effort describes the design of an interactive, multimodal digital entertainment user experience for an ‘edutainment’ cooking show. The prototype wireframes incorporate three novel features, identified through requirements gathering and iterative design, of an interactive recipe map for hierarchical content navigation, voice command hands-free control, and avatars to further engage users. The overall design and three features were evaluated via usability testing in real kitchen settings in the conduct of actual cooking with real ingredients with a diverse range of seven participants. The results illustrate that the hierarchical navigational feature, alongside interactive voice communication, were effective at reducing users' cognitive load, and streamlined necessary information in order to support task completion.
{"title":"Developing a Multimodal Entertainment Tool with Intuitive Navigation, Hands-Free Control, and Avatar Features, to Increase User Interactivity","authors":"N. Barrington, C. Gayle, E. Hensien, G. Ko, M. Lin, S. Palnati, G. J. Gerling","doi":"10.1109/sieds55548.2022.9799374","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799374","url":null,"abstract":"Advances in user interactivity in digital entertainment offer the potential to engage users beyond traditional passive and solitary experiences. Indeed, a high level of interactivity is inherent in tasks involving cooking, auto repair, and home improvement, all of which require users to complete multiple, detailed, and interdependent steps. Such tasks also require access to visual and audio instructions while a user's hands are engaged in a primary, physical task, and are often conducted in unique locations, e.g., kitchen or garage. This effort describes the design of an interactive, multimodal digital entertainment user experience for an ‘edutainment’ cooking show. The prototype wireframes incorporate three novel features, identified through requirements gathering and iterative design, of an interactive recipe map for hierarchical content navigation, voice command hands-free control, and avatars to further engage users. The overall design and three features were evaluated via usability testing in real kitchen settings in the conduct of actual cooking with real ingredients with a diverse range of seven participants. The results illustrate that the hierarchical navigational feature, alongside interactive voice communication, were effective at reducing users' cognitive load, and streamlined necessary information in order to support task completion.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133688696","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799373
Nazanin Abolfazli, Masoud Eshghali, S. F. Ghomi
The purpose of this study is to address the issue of coordination and pricing in a single-period and three-stage green supply chain in which green products and non-green products exist together and can be substituted with each other in the market. We examine the equilibrium results for two production modes, green production mode and hybrid production mode, in the cooperative and non-cooperative game to demonstrate the importance of entering supply chain members in collaboration. Theoretical analysis shows that different production costs lead the manufacturer to decide on different production modes when customers have further evaluations about various types of products. Furthermore, the results indicate that the system's performance in a cooperative game is better than that in a non-cooperative game, implying that supply chain members will respond positively to collaboration as their profit is higher than that under the non-cooperative strategy. The cooperative pricing strategy implemented by the Rubinstein bargaining model can provide the Pareto optimal solution for the supply chain system's profit and members' profits considering different production modes. Finally, the proposed model is applied to a generated numerical example to validate the suggested coordinated pricing strategy's validity and results.
{"title":"Pricing and Coordination Strategy for Green Supply Chain Under Two Production Modes","authors":"Nazanin Abolfazli, Masoud Eshghali, S. F. Ghomi","doi":"10.1109/sieds55548.2022.9799373","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799373","url":null,"abstract":"The purpose of this study is to address the issue of coordination and pricing in a single-period and three-stage green supply chain in which green products and non-green products exist together and can be substituted with each other in the market. We examine the equilibrium results for two production modes, green production mode and hybrid production mode, in the cooperative and non-cooperative game to demonstrate the importance of entering supply chain members in collaboration. Theoretical analysis shows that different production costs lead the manufacturer to decide on different production modes when customers have further evaluations about various types of products. Furthermore, the results indicate that the system's performance in a cooperative game is better than that in a non-cooperative game, implying that supply chain members will respond positively to collaboration as their profit is higher than that under the non-cooperative strategy. The cooperative pricing strategy implemented by the Rubinstein bargaining model can provide the Pareto optimal solution for the supply chain system's profit and members' profits considering different production modes. Finally, the proposed model is applied to a generated numerical example to validate the suggested coordinated pricing strategy's validity and results.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125787578","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799360
George Corbin, Nora Dale, Aatmika Deshpande, Katherine Korngiebel, Paige Krablin, Emma Wilt, L. Alonzi, Neal Goodloe, Michael C. Smith, K. P. White
The United States is the world's leading country in incarceration. American citizens constitute five percent of the global population, but 20% of the world's inmates [5]. Those suffering from mental illnesses are disproportionately affected. According to a 2017 study by the Department of Justice, 64% of inmates in local jails have a history of mental health problems, and 60% are actively experiencing symptoms [2]. To lower the number of Americans behind bars, effective mental health treatment needs to be provided to those in need within the criminal justice system. This project, supported by the Jefferson Area Community Criminal Justice Board, is the continuation of a decade of research into the intersection between mental illness and incarceration in the Central Virginia. The primary goal was to evaluate the efficacy of the Brief Jail Mental Health Screener (BJMHS) used by the region's two jails to determine whether an inmate needs further mental health evaluation following their release. Data was obtained from both jails: the Albemarle-Charlottesville Regional Jail (ACRJ) and the Central Virginia Regional Jail (CVRJ), as well as two community programs that provide services to former inmates, Offender's Aid and Restoration (OAR) and Region Ten Community Services (R10). The BJMHS was found to predominantly identify people who had already received treatment. The screener's effectiveness was also found to vary by the location it was given and by the recipient's demographics: Females tended to make up a statistically significantly larger proportion of the screened-in population than expected, and black individuals a smaller proportion. When people took the screener multiple times at different locations (ACRJ, CVRJ, or OAR) and were changing their answers to therapeutic questions, they were more likely to acknowledge they were previously hospitalized for mental health treatment at OAR than they were at either jail. Additionally, of the cohort of inmates screening in multiple times at ACRJ, it was found that as their number of arrests increased, so did the proportion of the group that screened in and group that matched with R10. The findings of this paper will be used to improve the screener process and ideally increase its ability to correctly identify those who require mental health services.
{"title":"Evaluating Administered Differences of Brief Jail Mental Health Screener and Impacts of Diagnoses & Treatment of Linked Inmates with Severe Mental Illness","authors":"George Corbin, Nora Dale, Aatmika Deshpande, Katherine Korngiebel, Paige Krablin, Emma Wilt, L. Alonzi, Neal Goodloe, Michael C. Smith, K. P. White","doi":"10.1109/sieds55548.2022.9799360","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799360","url":null,"abstract":"The United States is the world's leading country in incarceration. American citizens constitute five percent of the global population, but 20% of the world's inmates [5]. Those suffering from mental illnesses are disproportionately affected. According to a 2017 study by the Department of Justice, 64% of inmates in local jails have a history of mental health problems, and 60% are actively experiencing symptoms [2]. To lower the number of Americans behind bars, effective mental health treatment needs to be provided to those in need within the criminal justice system. This project, supported by the Jefferson Area Community Criminal Justice Board, is the continuation of a decade of research into the intersection between mental illness and incarceration in the Central Virginia. The primary goal was to evaluate the efficacy of the Brief Jail Mental Health Screener (BJMHS) used by the region's two jails to determine whether an inmate needs further mental health evaluation following their release. Data was obtained from both jails: the Albemarle-Charlottesville Regional Jail (ACRJ) and the Central Virginia Regional Jail (CVRJ), as well as two community programs that provide services to former inmates, Offender's Aid and Restoration (OAR) and Region Ten Community Services (R10). The BJMHS was found to predominantly identify people who had already received treatment. The screener's effectiveness was also found to vary by the location it was given and by the recipient's demographics: Females tended to make up a statistically significantly larger proportion of the screened-in population than expected, and black individuals a smaller proportion. When people took the screener multiple times at different locations (ACRJ, CVRJ, or OAR) and were changing their answers to therapeutic questions, they were more likely to acknowledge they were previously hospitalized for mental health treatment at OAR than they were at either jail. Additionally, of the cohort of inmates screening in multiple times at ACRJ, it was found that as their number of arrests increased, so did the proportion of the group that screened in and group that matched with R10. The findings of this paper will be used to improve the screener process and ideally increase its ability to correctly identify those who require mental health services.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"105 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115760849","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799406
Stephen P Ford, Rehan Merchant, Avinaash Pavuloori, Ryan Williams, C. Dreisbach, A. Saunders, Christian Wernz, Jonathan Michel
Resource allocation, including decisions about clinical and administrative staffing, language interpreter requirements, and billing procedures, is challenging in a complex medical system. In the setting of limited resources and high patient need, identification of patients who require a high amount of medical, nursing, and clinical services need to be identified for optimal care. The purpose of this paper is to identify the factors that predict patient phenotypes, a set of observable characteristics of an individual, that reflect their primary care resource usage. The data used in this study are de-identified, patient level data (n=34,957) between January 2019 to December 2021. We used k-means clustering to identify patient phenotypes based on the frequency of primary care and emergency department visits. Using multinomial regression, we then identified insurance type, comorbidity score, age, race, language, gender, hypertension, chronic opioid, obesity, prediabetes, tobacco usage, congestive heart failure, and chronic obstructive pulmonary disease as significant predictors for the primary care usage phenotypes. Having a more complete, holistic understanding of patient resource phenotypes can help leaders to make important decisions regarding optimal hospital resource allocations. Future work using our methods could be used to prospectively identify patients in high-need resource phenotypes compared to individuals with average annual usage.
{"title":"Patient Phenotypes to Identify Resource Allocation and Usage in Primary Care","authors":"Stephen P Ford, Rehan Merchant, Avinaash Pavuloori, Ryan Williams, C. Dreisbach, A. Saunders, Christian Wernz, Jonathan Michel","doi":"10.1109/sieds55548.2022.9799406","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799406","url":null,"abstract":"Resource allocation, including decisions about clinical and administrative staffing, language interpreter requirements, and billing procedures, is challenging in a complex medical system. In the setting of limited resources and high patient need, identification of patients who require a high amount of medical, nursing, and clinical services need to be identified for optimal care. The purpose of this paper is to identify the factors that predict patient phenotypes, a set of observable characteristics of an individual, that reflect their primary care resource usage. The data used in this study are de-identified, patient level data (n=34,957) between January 2019 to December 2021. We used k-means clustering to identify patient phenotypes based on the frequency of primary care and emergency department visits. Using multinomial regression, we then identified insurance type, comorbidity score, age, race, language, gender, hypertension, chronic opioid, obesity, prediabetes, tobacco usage, congestive heart failure, and chronic obstructive pulmonary disease as significant predictors for the primary care usage phenotypes. Having a more complete, holistic understanding of patient resource phenotypes can help leaders to make important decisions regarding optimal hospital resource allocations. Future work using our methods could be used to prospectively identify patients in high-need resource phenotypes compared to individuals with average annual usage.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117019941","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799299
Anahita H. Sharma, Burke W. Lawlor, Jason Y. Wang, Y. Sharma, S. Sengupta, P. Fernandes, Fatima Zulqarnain, Eve May, S. Syed, Donald E. Brown
The current gold standard for Crohn's disease diagnosis involves the examination of biopsied tissue by a trained physician. However, endoscopic images and histological features are only evident when the appropriate biopsy site is chosen and the image is of high quality [1]. Thus, to prevent delayed diagnoses or reclassifications over time, additional tools to reinforce pathologists' diagnoses are necessary. Recent studies have showcased successful applications of deep learning for developing whole-slide classifications of digital histology images. In this study, we developed a patch-level image classification model for prediction of Crohn's disease using a convolutional neural network. This study obtained data from two different hospitals: INOVA and Cincinnati Children's Hospital Medical Center (CCHMC). When trained and validated on the same data set, our INOVA and CCHMC models achieved validation accuracies of 84.6 % and 93.9 %, respectively. However, the models performed poorly when trained on data from one site and tested on data from the other site. To investigate this issue, we built an additional patch-level model that was able to predict hospital source of the biopsy with 99 % accuracy. These results suggest the presence of site-specific artifacts which are detectable by machine learning models. We reduced the effects of these artifacts using color-normalization, image cropping, and other transformations, lowering site-predictive accuracy to 74%. Therefore, we suggest further works investigate reasons for inter-site biopsy differences such that site-generalizable, histopathological deep learning models can be developed.
{"title":"Deep Learning for Predicting Pediatric Crohn's Disease Using Histopathological Imaging","authors":"Anahita H. Sharma, Burke W. Lawlor, Jason Y. Wang, Y. Sharma, S. Sengupta, P. Fernandes, Fatima Zulqarnain, Eve May, S. Syed, Donald E. Brown","doi":"10.1109/sieds55548.2022.9799299","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799299","url":null,"abstract":"The current gold standard for Crohn's disease diagnosis involves the examination of biopsied tissue by a trained physician. However, endoscopic images and histological features are only evident when the appropriate biopsy site is chosen and the image is of high quality [1]. Thus, to prevent delayed diagnoses or reclassifications over time, additional tools to reinforce pathologists' diagnoses are necessary. Recent studies have showcased successful applications of deep learning for developing whole-slide classifications of digital histology images. In this study, we developed a patch-level image classification model for prediction of Crohn's disease using a convolutional neural network. This study obtained data from two different hospitals: INOVA and Cincinnati Children's Hospital Medical Center (CCHMC). When trained and validated on the same data set, our INOVA and CCHMC models achieved validation accuracies of 84.6 % and 93.9 %, respectively. However, the models performed poorly when trained on data from one site and tested on data from the other site. To investigate this issue, we built an additional patch-level model that was able to predict hospital source of the biopsy with 99 % accuracy. These results suggest the presence of site-specific artifacts which are detectable by machine learning models. We reduced the effects of these artifacts using color-normalization, image cropping, and other transformations, lowering site-predictive accuracy to 74%. Therefore, we suggest further works investigate reasons for inter-site biopsy differences such that site-generalizable, histopathological deep learning models can be developed.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124377336","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 : 2022-04-28DOI: 10.1109/SIEDS55548.2022.9799314
Charles F. Bass, Matthew Fitzsimmons, S. Keith, Thomas Lam, A. O’Neill, V. Lakshmi
Tensions between Egypt, Sudan, and Ethiopia have grown as a result of concerns regarding water security. These tensions have been magnified by the construction of the Grand Ethiopian Renaissance Dam along the Nile River. The dam has potential to increase power production of the region while also creating risk for downstream countries. Therefore, this research will focus on quantifying the economic impact of the Grand Ethiopian Renaissance Dam to understand its implications for the Nile River basin. This will be accomplished by utilizing historical data and case studies to identify factors which may significantly change as a result of the dam construction for the countries of Egypt, Sudan, and Ethiopia. Cases of interest include the High Aswan Dam in Egypt as well as the Merowe Dam located in Sudan. Ultimately, the results of this research take the form of analysis conducted on water security, land use, agriculture, hydropower and the broader economic considerations for the Nile River basin. Additionally, despite the uncertainty of future management strategies, revenue generation was projected using two filling timelines. By quantifying the economic impact of the dam, the results of this research will provide an understanding of how the Grand Ethiopian Renaissance Dam will influence the future of the Nile River region.
{"title":"Quantifying the Economic Impact of the Grand Ethiopian Renaissance Dam on the Nile River Basin","authors":"Charles F. Bass, Matthew Fitzsimmons, S. Keith, Thomas Lam, A. O’Neill, V. Lakshmi","doi":"10.1109/SIEDS55548.2022.9799314","DOIUrl":"https://doi.org/10.1109/SIEDS55548.2022.9799314","url":null,"abstract":"Tensions between Egypt, Sudan, and Ethiopia have grown as a result of concerns regarding water security. These tensions have been magnified by the construction of the Grand Ethiopian Renaissance Dam along the Nile River. The dam has potential to increase power production of the region while also creating risk for downstream countries. Therefore, this research will focus on quantifying the economic impact of the Grand Ethiopian Renaissance Dam to understand its implications for the Nile River basin. This will be accomplished by utilizing historical data and case studies to identify factors which may significantly change as a result of the dam construction for the countries of Egypt, Sudan, and Ethiopia. Cases of interest include the High Aswan Dam in Egypt as well as the Merowe Dam located in Sudan. Ultimately, the results of this research take the form of analysis conducted on water security, land use, agriculture, hydropower and the broader economic considerations for the Nile River basin. Additionally, despite the uncertainty of future management strategies, revenue generation was projected using two filling timelines. By quantifying the economic impact of the dam, the results of this research will provide an understanding of how the Grand Ethiopian Renaissance Dam will influence the future of the Nile River region.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"30 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125697671","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799352
Ce Johnson, Hannah E. Schmuckler
The US Census Bureau uses its decennial census codes for industry and occupation in the monthly Current Population Survey. The Census Bureau has regularly revised these three- and four-digit codes to more accurately reflect the reality of work in the United States. These changes make it difficult to study industries and occupations over time. While limited crosswalks exist, there is currently no way to translate an individual's coded occupation or industry to every other scheme for long-term comparison by social scientists. This project aims to impute the most likely code for an individual's occupation and industry into each year's coding scheme by using random forest models to translate industry and occupation across decades. To our knowledge, this is the first tool that can map industry and occupation at scale with a high degree of accuracy into any year's scheme.
{"title":"Longitudinal Classification and Predictive Modeling for Historical CPS Data Using Random Forests","authors":"Ce Johnson, Hannah E. Schmuckler","doi":"10.1109/sieds55548.2022.9799352","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799352","url":null,"abstract":"The US Census Bureau uses its decennial census codes for industry and occupation in the monthly Current Population Survey. The Census Bureau has regularly revised these three- and four-digit codes to more accurately reflect the reality of work in the United States. These changes make it difficult to study industries and occupations over time. While limited crosswalks exist, there is currently no way to translate an individual's coded occupation or industry to every other scheme for long-term comparison by social scientists. This project aims to impute the most likely code for an individual's occupation and industry into each year's coding scheme by using random forest models to translate industry and occupation across decades. To our knowledge, this is the first tool that can map industry and occupation at scale with a high degree of accuracy into any year's scheme.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131809268","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 : 2022-04-28DOI: 10.1109/sieds55548.2022.9799310
Liam Whitenack, R. Mahabir
Daily, companies compete for customers in order to increase their revenue. The food industry, and in particular, very large restaurant chains, are no different. Customers are drawn to the opportunity to conveniently retrieve their food with minimum wait times using drive-thru services. While such services are not new and are used by a large number of restaurants, the fundamental paradigm (i.e., the configuration of employee agents and their interaction with consumer agents) through which drive-thru services continue to be used is difficult to observe in a meaningful way. Recently, with the onset of the COVID-19 pandemic, drive-thru services were heavily relied upon to provide much of the limited person-to-person contact service necessary to help reduce the spread of disease. While this presented many opportunities for existing businesses to scale their operations, it also revealed many inefficiencies with drive-thru services and the way they conduct their business, leading to longer waiting times. This paper addresses this issue by developing a simulation-based tool for identifying inefficiencies in existing drive-thru services. The tool allows a range of both employee and customer agent scenarios to be tested, providing important situational awareness for restaurant owners. Questions that the tool can help businesses answer include: identifying the most optimized configuration for minimizing customer wait times due to resources constraints (e.g., employee availability), possible impacts to business with switching strategies, and service point bottlenecks. A set of best practices, in line with industry standards and based on a review of the literature, were used in the design phase of this work. The developed tool is open-sourced1 and presents an interactive and easy-to-use interface that businesses can use to improve their service wait times.
{"title":"A Tool for Optimizing the Efficiency of Drive-Thru Services","authors":"Liam Whitenack, R. Mahabir","doi":"10.1109/sieds55548.2022.9799310","DOIUrl":"https://doi.org/10.1109/sieds55548.2022.9799310","url":null,"abstract":"Daily, companies compete for customers in order to increase their revenue. The food industry, and in particular, very large restaurant chains, are no different. Customers are drawn to the opportunity to conveniently retrieve their food with minimum wait times using drive-thru services. While such services are not new and are used by a large number of restaurants, the fundamental paradigm (i.e., the configuration of employee agents and their interaction with consumer agents) through which drive-thru services continue to be used is difficult to observe in a meaningful way. Recently, with the onset of the COVID-19 pandemic, drive-thru services were heavily relied upon to provide much of the limited person-to-person contact service necessary to help reduce the spread of disease. While this presented many opportunities for existing businesses to scale their operations, it also revealed many inefficiencies with drive-thru services and the way they conduct their business, leading to longer waiting times. This paper addresses this issue by developing a simulation-based tool for identifying inefficiencies in existing drive-thru services. The tool allows a range of both employee and customer agent scenarios to be tested, providing important situational awareness for restaurant owners. Questions that the tool can help businesses answer include: identifying the most optimized configuration for minimizing customer wait times due to resources constraints (e.g., employee availability), possible impacts to business with switching strategies, and service point bottlenecks. A set of best practices, in line with industry standards and based on a review of the literature, were used in the design phase of this work. The developed tool is open-sourced1 and presents an interactive and easy-to-use interface that businesses can use to improve their service wait times.","PeriodicalId":286724,"journal":{"name":"2022 Systems and Information Engineering Design Symposium (SIEDS)","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116267424","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：