Monday, April 29, 2019 |
10:30 AM - 11:30 AM |
Tyler Smith |
Education and Career Development: Education |
Integrating Case Studies in a Health Analytics Curriculum |
Advanced informatics layered with customized analytics has driven major change in every discipline in the past decade. In the healthcare environment, individual level determinants of health are being leveraged to reveal personalized patient management that considers disease patterns, high-risk attributes, hospital acquired conditions, and performance measures for specialized treatment approaches. The growing need for health analytics expertise in light of these health informatics advancements during the last decade has created a critical void for higher education to fill. An abundance of new data-based opportunities that have made large public-use data sets accessible for easy download and use in the classroom have allowed for a much more applied classroom experience. More hands-on applications are positioning students for greater impact in the real world upon graduation and entry into the job market. In this presentation, we highlight the use of controlled and adaptive case studies leveraging SAS® and real-world data to provide a more realistic classroom experience.
|
Monday, April 29, 2019 |
11:30 AM - 12:00 PM |
Tom Sabo |
Business Analytics/ Data Visualization: Business Analytics |
An Artificial Intelligence Framework on SAS® Viya® to Counter International Human Trafficking |
Efforts to counter human trafficking internationally must assess data from a variety of sources to determine where best to devote limited resources. These varied sources include the US Department of State’s Trafficking in Persons (TIP) reports, verified armed conflict events, migration patterns, and social media. How can analysts effectively tap all the relevant data to best inform decisions to counter human trafficking? This paper builds on two previous SAS® Global Forum submissions that apply SAS® text analytics to the TIP reports, as well as to the Armed Conflict Location & Event Data (ACLED) project. We show a framework supporting artificial intelligence on SAS Viya for exploring all data related to counter human trafficking initiatives internationally, incorporating the TIP and ACLED sources as a starting point. The framework includes SAS rule-based and supervised machine learning text analytics results that were not available in the original data sets, providing a depth of computer-generated insight for analysts to explore. We ultimately show how this extensible framework provides decision-makers with capabilities for countering human trafficking internationally, and how it can be expanded as new techniques and sources of information become available.
|
Monday, April 29, 2019 |
12:00 PM - 12:30 PM |
Andrea Barbo |
Programming: Data Presentation |
Visualizing Hospital Readmission Rates: How Well Does Your Hospital Perform? |
Hospital readmission is an adverse but often preventable event that has been shown to be related to a hospital’s quality of care (Frankl et al., 1991). To reduce unplanned 30-day readmission rates among Medicare beneficiaries, the Centers for Medicare & Medicaid Services introduced measures of risk-standardized readmission rates (RSRRs) in both its pay for reporting and performance programs. RSRRs are calculated for five medical conditions, two procedures, and one hospital-wide measure for participating hospitals. These results are publicly reported in Hospital Compare, an online tool that patients can use to guide their decisions about where to seek care. Radar plots provide a tool that can illuminate differences and similarities of rates within and across hospitals. This can be useful for patients seeking the best possible care, or for hospital quality departments trying to understand their hospital’s performance compared to their peers. This technique can be implemented using the POLYGON statement in the SGPANEL procedure (Hebbbar, 2013). The code can be modified to add grids, tick marks, and labels, which provide more information on the estimates. Plotting the hospitals’ RSRRs against the observed national rate enables one to quickly see which hospitals perform worse or better than the national average and which perform similarly in certain measures. In summary, radar plots are an effective way to display multiple hospital RSRRs at once and make quick comparisons.
|
Monday, April 29, 2019 |
12:00 PM - 12:30 PM |
Anjali Bansal |
Analytics: Life Sciences |
Are You in Danger of Stroke? An Insight into the Leading Causes |
Stroke (also known as cerebrovascular accident) is one of the leading causes of death and a significant root of disability. According to the World Health Organization, 15 million people suffer stroke worldwide each year. Of these, five million die and another five million are permanently disabled. Currently stroke is the fifth leading cause of death in U.S. and the most important cause of disability. The question is: What leads to STROKE? A stroke is usually the result of lack of blood supply to the brain either from interruption of flow or reduction, thereby depriving the brain tissue of oxygen and necessary nutrients. Brain cells begin to die within minutes of the event, which can lead to permanent disability. Hypertension plays a significant role in the occurrence of stroke. Hypertension weakens arterial walls in the brain that can lead to a rupture resulting in hemorrhagic stroke. This paper gives detailed insight about the occurrence of stroke in the United States. This project also attempts to study the association between different cardiovascular diseases and stroke, and to understand the importance of hypertension in bringing about a stroke. The data was obtained from the Center for Health Systems Innovation at Oklahoma State University. Data regarding patient demographics, patient disease status, hospital encounters, and treatment were examined. SAS® Viya® was used to study the patients’ demographics with respect to stroke. SAS® Enterprise Miner™ was used for predictive model building.
|
Monday, April 29, 2019 |
12:00 PM - 12:30 PM |
Lauren Agrigento |
Analytics: Text Analytics |
Identification of the Factors Associated with Human Trafficking Recruitment |
Human trafficking networks commonly move from city to city in an effort to both avoid the police and to take advantage of certain events that might increase their business. In this study, unsupervised and unstructured textual data scraped from advertisements on Backpage.com is analyzed using SAS® Viya® textual machine learning analytics techniques to ultimately identify the types of appeals that human traffickers use in recruiting ads. We plan to use the results of this analysis to assist law enforcement in properly using its resources when identifying advertisements in which recruiting ads and their contents are properly categorized. The idea behind our analysis is to find out more about those recruiting ads by separating them from other advertisements, based on several unique features. These ads appear all over the country, but they originate in only a small number of places. Using these places and the unique features from each ad enables us to map each potential recruitment ring. Custom Concepts in SAS Viya enables us to classify indicators of the ad being the reason for the advertisement’s appearance, such as certain words or phrases that might indicate whether the trafficking network moved to the area in order to benefit from the recruitment’s increased demand. Assigning term densities to the Custom Concepts enables us to group similar posts based on their duration and attractor, and we can therefore identify direct links to traffickers.
|
Monday, April 29, 2019 |
1:30 PM - 2:00 PM |
Jim Brittain |
Programming: Data Presentation |
Using the ODS Report Writing Interface to Streamline Publication of Existing Reports |
This paper examines the Output Delivery System (ODS) Report Writing Interface (RWI) to streamline the publication process of "Health, United States", an annual report about the health status of the people living in the U.S. Trend tables within the report are complex, with multiple nested categorical values within rows, and are designed to be suitable for printing as PDF files. Section 508 of the Rehabilitation Act of 1973 requires all federal agency website content to be accessible to people with disabilities. The current process to publish these tables involves Dynamic Data Exchange (DDE), aging technology from SAS® that exports Microsoft Excel data to the trend tables. Manual verification, conversion to PDF, and tagging for accessibility is a labor-intensive process. Using RWI, the publishing process can be automated and the delivery of output can be enhanced. We use the ODS templates and Cascading Style Sheets (CSSs) to standardize and enhance the format of the output, varying formatting by using conditional processing. The current PDF style needs to be maintained while limiting the manual processing. We also examine additional output format options for the extended data that is currently provided as an Excel file designed for readability. Output includes PDF, Excel, and HTML5. Using the RWI and SAS accessibility features, we hope to alleviate the majority of the manual Section 508 compliance task of the output.
|
Monday, April 29, 2019 |
1:30 PM - 2:00 PM |
Kaito Kobayashi |
Analytics: Forecasting |
Forecasting CO2 Emission of Electrical Generation By Using SAS® Software |
Currently, global warming is one of the most severe problems in the world. From many researches, it is widely believed that CO2 is the main factor in global warming. Therefore, in many countries, introducing renewable energy is actively promoted so as to avoid high-emission power production methods such as thermal power generation. In this paper, the objective is to forecast future CO2 emission by using SAS® software. I analyzed the past generating trend of the amount of electricity demand or generation in order to predict future production quantity and CO2 emission. Also, taking worldwide tendencies into consideration, I made some probable eco-friendly scenarios, like shifting drastically to solar power, and made an analysis of each of them. Finally, I proposed some electrical generation plans for a sustainable world based on these analyses.
|
Monday, April 29, 2019 |
1:30 PM - 2:00 PM |
Margaret Kline |
Programming: General Programming |
Nothing is More Powerful than the Apache Helicopter, Except the APACHE II Macro |
The Acute Physiology and Chronic Health Evaluation (APACHE) II classification system is commonly used in intensive care units (ICU) to classify disease severity and predict hospital mortality. Disease severity scales in the ICU are a necessary component to assist in predicting patient outcomes, comparing quality of care, and stratifying patients for clinical trials. The APACHE II score is calculated from patient demographics and physiologic variables measured in the patient’s first 24 hours following ICU admission. The score comprises three components: 1) an acute physiology score (APS); 2) age; and 3) chronic health conditions. Although this information is accessible in electronic medical records (EMR), some systems do not have a way to automate the calculation of the APACHE II score, leaving physicians to calculate it by hand, which decreases patient care time and increases calculation error rates. This paper shows how to calculate the APACHE II score based on information from EMRs using our macro.
|
Monday, April 29, 2019 |
3:00 PM - 3:30 PM |
Mamadou Dakouo |
Analytics: Population and Public Health |
Recurrent-Event Survival Analysis: 33-Year Follow-Up of Lymphohaematopoietic Cancer Risk in Ontario |
Ontario and other Canadian jurisdictions have never had occupational disease surveillance systems to identify high-risk industries and target prevention efforts. The Occupational Disease Surveillance System (ODSS) was created to identify patterns and monitor trends in work-related disease in Ontario. In this study, we examined existing patterns and emerging trends in work-related lymphohaematopoietic (LH) cancer by industry groups. The ODSS established 2.2 million workers in Ontario by linking a cohort of workers with data from Workplace Safety and Insurance Board (WSIB) claims, to administrative health databases. Our definition of LH cancer includes lymphoma, Hodgkin's lymphoma, non-Hodgkin's lymphoma, and myeloma. A subject might have been diagnosed with LH cancer several times during the follow-up period. A new LH cancer diagnosis was considered to be a Recurrent Event. The Counting Process model was used to develop an age-adjusted, sex-stratified Cox proportional hazard regression. The risk of LH cancer within a particular industry versus the risk in all other industry groups in the cohort was calculated. All analyses were performed using SAS® 9.4. Of 2,188,302 workers included, 25,535 LH cancers were diagnosed. An increased risk was observed among workers in tobacco products industries for both sexes and among males in metal mines. This study identified several distinct industries at high risk for LH cancer and could be used for surveillance of occupations in Ontario.
|
Monday, April 29, 2019 |
3:30 PM - 4:00 PM |
Jennifer Sabourin |
Education and Career Development: Education |
How Access to STEM Can Save the World |
For every challenge we face on the planet, there is data that can help us find the solution. The next generation of problem solvers, “the digital natives”, are poised to see issues such as poverty, hunger, gender equality, and climate change not as insurmountable but as solvable puzzles. With unprecedented access to technology and data, and an unquenchable thirst for digital connection, this generation holds incredible promise in solving social challenges that affect every population, every race, every gender. An interdisciplinary approach to education that incorporates science, technology, engineering, and math (STEM) and project-based learning is the best way to prepare these students for the world that awaits their contributions. GatherIQ™, a mobile and web app produced by SAS in conjunction with their education software division, Curriculum Pathways, invites students to join the global quest to reach the United Nation's Sustainable Development Goals for 2030 through experiences in STEM and project-based learning. Using GatherIQ, students not only learn about the issues but combine their wits to address social challenges in their own back yards and around the world.
|
Monday, April 29, 2019 |
3:30 PM - 4:30 PM |
Richa Sehgal |
Analytics: Data Mining and Machine Learning |
Using Machine Learning and Clustering Methods to Improve the Effectiveness of Tumor Biopsies |
Currently, tumor biopsies do not provide doctors with all the possible information that could be used in determining a treatment plan for patients with cancer. The reason for this is that biopsies only remove a very small part of the tumor. However, there are certain “rare” cells (those that have a very large impact on determining key facts like how aggressive the cancer is and how fast it will spread and grow) that exist scattered around the tumor, but because of its size and lack of direction, the biopsy often does not pick these up. Once a dataset is formulated with different datapoints representing different cells within the tumor as a whole, clustering methods and related algorithms available in SAS can be used to find the groups of data points with the greatest variation. This shows the area of the tumor that contains the greatest variety in cell types, which is where a surgeon would want to aim during a biopsy in order to get the most accurate representation of the tumor and therefore create the most accurate prognosis and treatment plan. This model can then be applied to tumors of different cancers, sizes, and stages. This presentation will demonstrate the use of clustering procedures available in SAS to identify a biopsy sample with maximum variation in cell type.
This work is an extension of the research that the author did during her internship at Stanford in the summer of 2017.
|
Monday, April 29, 2019 |
4:00 PM - 4:30 PM |
Cleester Heath IV |
Programming: Data Presentation |
Create a Combined Graph of Tumor Data |
Traditionally, tumor response and duration of treatment information have been displayed in separate graphs in which the subjects can be sorted by different criteria. In such cases, the clinician has to work harder to associate the subject across the graphs. Recently there has been increased interest in combining this information in one visual. Displaying the data together, sorted by the tumor response with associated duration information, makes it easier for the clinician to understand the data. Three-dimensional waterfall graphs, which have both pros and cons, have been proposed for such cases. This paper shows you how to build a 3-D graph using SAS® that shows both tumor response and duration of treatment. This paper also presents alternative 2-D visuals that were created using the SGPLOT procedure. These 2-D visuals enable easier decoding of the data, which enables you to display more information in the graph.
|
Tuesday, April 30, 2019 |
10:30 AM - 11:00 AM |
Sareh Meshkinfam |
Analytics: Statistics |
SAS® Analytics for Characterizing the Relationship between Teacher Judgment and Student Performance |
A teacher's assessment or judgment of an individual student's performance can impact other educators' expectations of that student's ability as well as the student's future academic placement. Exploring the relationship between teacher judgment and student performance in primary education is critical, as early barriers can evolve into significant academic hurdles for individual students at the middle school and high school levels. Correlation analysis and regression models have been used to analyze the longitudinal and cross-sectional relationship between students’ achievements and teacher judgment in reading and mathematics across grades and years, considering students' demographics. SAS® procedures (such as PROC CORR and PROC MIXED) were used to explore a large data set from the North Carolina Education Research Data Center (NCERDC), which includes 6,511,741 students in 3rd to 8th grades from 2006 to 2013. The data set includes information such as students' End-Of-Grade (EOG) test scores, demographic characteristics, and evidence of their academic performance in each grade and year. SAS provides an effective tool to explore this data set with accessible and easy-to-use analysis approaches. Results demonstrate moderate to high correlations, which are significantly higher for male and significantly lower for minority ethnic groups. The regression models reveal that students’ gender, ethnicity, and previous grade performance significantly affect their EOG achievement score.
|
Tuesday, April 30, 2019 |
11:00 AM - 12:00 PM |
Yuri Quintana |
Industry Specific Topics: Technology Solutions |
The New Learning Health-Care System - How Data and Connected Communities Can Improve Health Care |
Rising chronic disease trends and aging demographics are creating unprecedented challenges in the delivery of health-care services. New technologies have enabled us to collect massive amounts of data, but making that information useful for clinical care and research remains a challenge. This talk presents an overview of learning health-care systems that aim to integrate clinical data, education, and care communication from multiple health-care providers, patients, and families involved in care support. This presentation describes the Alicanto learning health-care system that supports connected communities of health-care providers, patients, and families. We provide an overview of Alicanto's networks for cancer, elder care, and maternal health, and we discuss the challenges of integrating and analyzing data from clinical sources, patient-collected health outcomes, and home-care providers. Recommendations for future learning health-care systems are given.
|
Tuesday, April 30, 2019 |
11:00 AM - 11:30 AM |
Di Chang |
Analytics: Population and Public Health |
A Unified SAS® Approach to Matching Administrative Medical Claims with Survey Data |
The Arkansas State and Public School Employees Health Plan (ASEPSE) is the largest self-insured health plan in Arkansas, representing public school employees, state employees, retirees, and their dependents. In 2017, enrollment in the plan included approximately 153,000 members and is administered by the Arkansas Employee Benefits Division (EBD). As part of the plan’s wellness program, a premium incentive is available for members who complete an annual health risk assessment (HRA). The health risks captured in the survey represent modifiable lifestyle factors of the population. To present how modifiable lifestyle factors impact medical and pharmacy costs in this population, we used SAS® to import SQL-based administrative claims and the SUMMARY procedure to roll-up 2017 costs over unique and anonymously identified individuals. This anonymous identifier was also included on cleaned and formatted HRA questionnaire data. Merging both data sources by individual, average health care costs by categories of modifiable lifestyle behaviors were produced. Clear dose-response increases in costs were obtained, especially for low levels of physical activity and high levels of sedentary behaviors. SAS graphics produced clear visualizations of the inverse relationship between health care costs and poor health behaviors. The comprehensive SAS code used to incorporate multiple data sources, analyses, and graphics were converted to a macro for replication on future years of data.
|
Tuesday, April 30, 2019 |
12:30 PM - 1:00 PM |
Colin Nugteren |
Analytics: Data Mining and Machine Learning |
Data4Good: Helping IOM Forecast Logistics for Refugees in Africa |
International Organization for Migration challenge: “Towards enhanced needs mapping for improved capacity and resource planning”. The International Organization for Migration (IOM) of the United Nations challenged start-ups to come up with a solution for better supply chain management by utilizing data and analytics. Together, Notilyze and Elva came up with a suitable solution to make more use of already available data by using SAS® Viya®. Specifically, the project team helped to create an analytical tool that allows for a better translation of existing analytical outputs (for example, the displacements tracking site assessments) for site planning and concrete supply chain management. It strives to realize this goal by achieving the following objectives: 1) Using both existing displacement tracking site assessment data and input from humanitarian field workers within IOM and other relevant stakeholders to identify a set of common indicators on humanitarian needs, related development needs, and supply chain gaps for needs mapping and forecasting during this project; 2) Based on these standardized indicators, develop a prototype algorithm that provides automated forecasting of humanitarian and development needs and supply chain requirements; and 3) Visualize these on-the-ground needs and supply gaps in interactive dashboards, maps, and automated forecasts.
|
Tuesday, April 30, 2019 |
1:30 PM - 2:00 PM |
Abbas Tavakoli |
Analytics: Population and Public Health |
Using Parallel Analysis to Determine the Dimensionality of a Computer-Based Prostate Cancer Screen |
Prostate cancer (PrCA) incidence and mortality rates are higher among African-American men than any other racial group. Informed decision-making about prostate cancer screening could result in early detection and potentially reduce cancer health disparities. Currently, there are some, but few, computer-based decision aids to facilitate PrCA decisions of African-American men, but no scale has been validated to assess the extent to which African-American men accept and use a computer-based PrCA screening decision aid. Using parallel analysis, this study determined the dimensionality of the Computer-Based Prostate Cancer Screening Decision Aid and Acceptance Scale using data from a purposive sample of 352 African-American men aged 40 years and older who resided in South Carolina. Exploratory factor analysis was conducted using maximum likelihood, squared multiple correlations, and Promax rotation. Internal consistency reliability was assessed using Cronbach’s alpha. Parallel analysis was used to determine the dimensionality of the scale using SAS® macro language. Results showed the optimal factor structure of the Computer-Based Prostate Cancer Screening Decision Aid among African-American men was a 24-item, 3-factor model. Factor loadings ranged from 0.32 to 0.94 with 11 items loading on Factor 1, 8 items on Factor 2, and 5 items on Factor 3. Parallel analysis is a valuable method for determining the dimensionality of the Computer-Based Prostate Cancer Screening Decision Aid.
|
Tuesday, April 30, 2019 |
1:30 PM - 2:00 PM |
Yang Yue |
Analytics: Population and Public Health |
A Program for the Calculation of Age-Adjusted Cancer Incidence and Mortality Rates |
Reporting incidence and mortality rates of various types of diseases are the major two aims of population-based studies. This template program was originally developed to calculate age-adjusted incidence rates and mortality rates of new cancer cases (or death) occurring among North Carolina residents. Because the variance calculation assumes that the cancer counts have Poisson distributions, this program can be used if the incidence rates are less than 10% in the population. The macro produces age-adjusted rates for each of cancer sites, races, genders, geographic locations combinations. SAS® 9.4 was used to write the macro, and the TABULATE procedure was applied to generate cancer cases (or death) and rates.
|
Tuesday, April 30, 2019 |
1:30 PM - 2:00 PM |
Ethan Ziemba |
Analytics: Forecasting |
A Storm Brewing over Drug Overdose |
Drug overdose, specifically opioid overdose, has been a growing problem in the United States (US). Typically, people attribute sociocultural and economic factors to the drug overdose epidemic. This paper examines some of the common economic factors that have an impact on opioid overdose. First, unemployment rates and median household income are examined over the years 1999 to 2016 to see the impact they have on opioid overdose. The paper then shifts to examine variables that have seen less attention in the literature. The US has seen several detrimental natural disasters over recent years. That being said, there has not been much research on the effect natural disasters have had on opioid overdoses. Additionally, political composition of states along with the legality of marijuana are variables that have yet to see a great amount of research regarding their impact on opioid overdose. Using data from a variety of sources and merging them into a singular table, this paper investigates the impact unemployment, median household income, natural disasters, political composition, and marijuana legality have on opioid overdose. Methods, including visualizations, predictive analysis, and clustering are used to analyze the data and draw conclusions.
|
Tuesday, April 30, 2019 |
2:30 PM - 3:30 PM |
David Corliss |
Data Management: Data Integration |
Capture-Recapture Databases for Data for Good Projects |
Capture-recapture is a statistical methodology that uses repeated and independent identification from samples of the subjects of interest. The method is able to provide accurate counts of the entire population from these samples. Originally developed by ecologists to count animal populations, capture-recapture has become a critically important analytic tool for social justice, providing accurate counts of the number of people affected by a wide variety of problems, including crimes and natural disasters. In this method, careful design, development, and management of the underlying database are critical tasks. This paper demonstrates development and management of databases for capture-recapture analysis, including database organization, integrating additional data sources, addressing privacy issues, and database management and governance.
|
Tuesday, April 30, 2019 |
3:30 PM - 4:30 PM |
Terry Clarke |
Programming: Data Presentation |
Hey, I Know What, Let's Include Everybody: Accessibility |
The advent of personal computers and then smartphones improved the lives of many people in many ways, but for disabled people it was a game-changer, allowing access to untold amounts of information that previously had not been available to them. Sadly, there is a downside: as with all data, access to the information contained within that data depends entirely on how that data has been presented. For disabled people, this understanding is even more important because if insufficient care is taken, information can be made completely inaccessible to large groups of people. I explain the nature of disability, the legality of not providing access to everybody, the standards and guidelines that are available, and how not considering inclusivity can affect the bottom line. I demonstrate some of the tools that disabled people use to access information and explain what can be done to make your data as inclusive as possible.
|
Tuesday, April 30, 2019 |
4:00 PM - 5:00 PM |
Michael Matthews |
Programming: Data Presentation |
Spatial Analysis of Humanitarian OpenStreetMap Team Data Using SAS® ODS Graphics Procedures |
The Humanitarian OpenStreetMap Team (HOT) community consists of volunteers from around the globe. HOT tasks involve developing maps that identify communities and infrastructures based on satellite imagery. These maps are then used to assist aid organizations such as the Red Cross during humanitarian crises and for general community development in areas that are often not covered by the mapping products that most of us take for granted. This presentation examines OpenStreetMap data and introduces HOT and some of the associated mapping tasks, including assisting the aid efforts during the May 2018 Ebola outbreak in the Democratic Republic of the Congo. Analysis is performed on the data using the SAS® ODS Graphics procedures (including PROC SGMAP) to visualize the contributions to OpenStreetMap both spatially and over time.
|
Tuesday, April 30, 2019 |
4:30 PM - 5:00 PM |
Jesse Sookne |
Business Analytics/ Data Visualization: Data Visualization |
Follow My Lead: Designing Accessible Reports by Example Using SAS® Visual Analytics |
Your Legal, IT, or Communications department said that your reports must be accessible to people with disabilities. They might have used terms like "Section 508" or "WCAG". Now what? This paper leads you down the path to creating accessible reports by using SAS® Visual Analytics. It includes examples of what to do—and what not to do—to make your reports accessible. It provides information about which types of objects to use and how to use them in order to maximize the accessibility of your reports. You can use the information in this paper to create accessible reports, comply with your organization's accessibility requirements, and enable people with disabilities to benefit from the information that you publish.
|
Tuesday, April 30, 2019 |
4:30 PM - 5:30 PM |
Greg Kraus |
Programming: Data Presentation |
The Incredible, Accessible Report: Creating PDF Reports that Meet Compliance Standards in SAS® 9.4M6 |
SAS® 9.4M6 enables SAS® programmers to create PDF reports that fully meet the Web Content Accessibility Guidelines 2.0 (WCAG 2.0) conformance requirements. These are the guidelines that government and industry use to determine whether electronically produced output is usable by people with disabilities. With the accessibility features in SAS 9.4M6, it is possible to create reports that require zero post-processing remediation work, thus saving you time and money. By using the PRINT, TABULATE, and REPORT procedures, and the ODS Report Writing Interface (RWI), SAS 9.4M6 can prompt you to address certain accessibility problems detected in your code, create tables of data that are structured to be fully accessible to users, and add alternative text for images inserted into your reports. This paper demonstrates how to use these reporting procedures to create accessible PDF reports.
|
Wednesday, May 1, 2019 |
11:00 AM - 12:00 PM |
David Olaleye |
Industry Specific Topics: Technology Solutions |
Real World Evidence and Population Health Analytics: Intersection and Application |
Outside of clinical trials, real world evidence (RWE) studies provide pragmatic evidence to investigate safety and effectiveness of marketed pharmaceutical products in real-world clinical settings. Leveraging the power of RWE studies to predict patient outcomes and forecast health care utilization have been met with the challenges of big data from disparate data sources. SAS® Real World Evidence and SAS® Visual Analytics enables quick discovery and creation of patient cohorts for population health analytics. Machine learning algorithms informed by industry-standard clinical diagnoses and episode-of-care definitions are used to capture the multi-dimensional nature and health care utilization of patients followed over time. Using publicly available claims data from the 2008–2010 Medicare population from the Centers for Medicare & Medicaid Services (CMS) and from a commercially managed care population, we show how interpretability of RWE studies can be extended with population health analytics to predict at-risk patient subgroups with a propensity to benefit from intensive care coordination and management. Our analyses focus on 1) creating episode-of-care and resource consumption profiles for a type 2 diabetes mellitus (T2DM) population cohort; 2) using unsupervised machine learning algorithms to identify safety events associated with the treatment pathways for the cohort; and 3) using the PSMATCH and CAUSALTRT procedures to obtain a matched patient sample and estimate the potential outcomes effects for the treatment pathways.
|
Wednesday, May 1, 2019 |
1:00 PM - 1:30 PM |
James Van Scotter |
Analytics: Text Analytics |
Human Trafficking Research: The Persistence and Movement of Phone Numbers in Escort Ads |
This paper explains how we used SAS® Viya® to investigate phone numbers used in human trafficking, using a combined sample of approximately 3 million classified ads posted on backpage.com. The contact phone numbers displayed in online sex ads might be the most important single clue for finding the traffickers and pimps who profit from these activities. Some investigators believe criminals use inexpensive “burner” phones and throw them away after a few months to make it difficult for law enforcement to track them. Another theory suggests traffickers use full-featured phones and keep them for much longer periods of time. In a previous study that used a sample of ads collected over one year and about 120 locations, we found that at least 50% of the numbers were still used in ads after 8–12 months. While 40% of the numbers were found in places they had previously appeared, 10% were in new locations. This suggested that some of the missing 50% of the phone numbers might have also moved to other areas—areas that were outside of the locations included in the original sample. To address this issue, we extended the original sample by collecting over 2.5 million additional ads, encompassing more than 300 locations and 12 more months. Results from the second study are presented.
|