Content listed below is not originally mine. I have added it from the PDF file so users won't be required to open random PDF attachments in order to view the content.
----------------------
SAS Methods for Fraud Discovery vs DOGE Open Discussion for the SAS Community By Melvin Ott, PhD Adjunct in Data Science, Northwestern University Since 2012 Introduction Invite to an open discussion: Who is better at Fraud Discovery? DOGE or SAS Analysts? In my career as an analyst, manager, and director, I have frequently examined large data transactions within the healthcare sector, working for organizations like Blue Shield and Ingenix Inc. Detecting fraud requires a systematic approach, leveraging advanced analytics and domain expertise. SAS has proven to be an invaluable tool in this endeavor, offering robust methods for identifying anomalies and irregularities in financial transactions. As a group, we have been doing this for a long time in the private sector. If we had been asked, we would have provided expertise to the government as well. Part 1 My Steps for Fraud Discovery Step 1: Sorting Data and Identifying Outliers The first critical step in fraud detection is sorting data to reveal outliers. Outliers often indicate unusual behavior, which could signify fraudulent activity. Step 2: Detecting Duplicate Payments Duplicate payments are a common form of fraud in healthcare transactions. SAS efficiently identifies duplicates by using PROC SORT followed by a check for duplicate claim numbers, dates, and identical charge amounts. Step 3: Validating Charges Invalid charges—such as those associated with unapproved services or inflated prices—are another red flag for fraud. SAS allows cross-referencing of claims data with approved billing codes and allowed pricing. Step 4: Predictive Modeling and Anomaly Detection Beyond rule-based methods, predictive analytics and anomaly detection models add a layer of sophistication to fraud detection. SAS supports the development of predictive models. Part 2. ChatGPT and perplexity.ai responses re DOGE Question: What analytics methods has DOGE used to find fraud? ChatGPT said: The Department of Government Efficiency (DOGE), led by Elon Musk, has implemented several analytical methods to detect and prevent fraud within federal operations. One primary approach involves the use of machine learning models, such as Random Forest and AdaBoost classifiers, to identify anomalies in financial transactions. These models analyze various features, including the standard deviation and average of transaction volumes, number of trades, and price fluctuations, to detect irregular patterns indicative of fraudulent activities. arxiv.org In addition to machine learning techniques, DOGE representatives have gained access to key payment and contracting systems within agencies like the Centers for Medicare and Medicaid Services (CMS). By examining the technology infrastructure and spending flows, they aim to pinpoint instances of fraud or waste. reuters.com However, it's important to note that recent reports have highlighted inaccuracies in DOGE's reported savings from fraud detection efforts. A review found errors such as unawarded contracts being listed as canceled and multiple listings of the same savings, raising concerns about the reliability of their reported data. politico.com Overall, DOGE employs a combination of advanced machine learning analytics and direct system audits to detect and address fraud within government operations. From perplexity.ai Question: What analytics methods has doge used to find fraud? Answer DOGE (Department of Government Efficiency) has employed several analytics methods to identify fraud in government systems, though specific details are limited: 1. Access to key systems: DOGE staff have been given access to payment and contracting systems at the Centers for Medicare and Medicaid Services (CMS) to look for fraud and ineffective use of resources3. 2. AI and predictive algorithms: DOGE is reportedly working on integrating more AI into government systems, building on existing predictive algorithms and models used by Medicare and Medicaid to detect fraud4. 3. Data analysis: DOGE has been analyzing federal contracts and expenditures, as evidenced by their release of data on canceled contracts and claimed savings5. 4. System probing: Young engineers recruited by DOGE have been tasked with accessing various government systems to look for fraud and inefficiencies3. However, it's important to note that the effectiveness and accuracy of DOGE's fraud detection methods have been questioned. An NPR analysis found that DOGE's claimed savings from canceled contracts were overestimated by billions of dollars5. Additionally, there are concerns about the privacy implications of DOGE's broad access to sensitive government systems, particularly in healthcare3. Part 3. Discussion and Request for Response The theory is that the SAS community, with its extensive domain expertise and proven methodologies, can outperform DOGE's machine learning-centric approach in detecting and preventing fraud. SAS tools not only provide transparency but also facilitate thorough auditing and validation of results. Question to the SAS Community: What methods have you used and were they successful? Theory is, we are better at this than DOGE. What do you think? Your insights and experiences are valuable. Join the discussion!
... View more
