The Problem
Imagine this scenario:
A patient named Emily has been suffering from severe migraines and finally gets a referral to a highly recommended neurologist. She calls the number listed online but it’s disconnected. Frustrated, she tries another number, but it routes her to a different department. Meanwhile, her condition worsens, causing her to miss more days at work. This is merely the tip of the iceberg when it comes to setbacks caused by inaccurate practitioner data. Let’s dive into some of the major issues.
The Patient’s Experience
For Emily, finding the right specialist has become a time-consuming and stressful task. Each incorrect phone number and outdated address is a barrier to her receiving timely care. She spends hours on the phone, navigating through automated systems and leaving voicemails that go unanswered. The delay in getting an appointment exacerbates her condition, leading to more frequent and intense migraines, affecting her quality of life and productivity at work.
The Provider’s Perspective
Dr. Smith, the neurologist Emily is trying to reach, is unaware of these issues. His practice moved to a new location six months ago, but the NPPES database and several online directories still have his old contact information. This oversight means he’s missing out on new patients who need his expertise. His office receives complaints about outdated contact information, which he must address, diverting resources and time that could be better spent on patient care.
The Insurance Company’s Dilemma
Emily’s insurance company, aiming to provide seamless care coordination, finds itself entangled in this web of inaccuracies. Each time Emily calls to verify Dr. Smith’s information, the insurer must cross-check and confirm details, consuming valuable time and resources. Claims are delayed due to mismatches in provider information, leading to increased administrative costs. These inefficiencies contribute to higher premiums for policyholders as the insurer tries to cover the additional administrative burden.
The Financial Impact
Inaccurate provider data not only causes inconvenience and delays in care but also has significant financial implications. According to a report by the Office of Inspector General (OIG), inaccuracies in provider data can lead to improper Medicare payments and administrative inefficiencies, costing millions annually (Office of Inspector General - HHS). The cumulative effect of these errors can inflate healthcare costs, as resources are wasted on verifying and correcting data rather than on patient care.
The Solution
In order for healthcare to scale affordably, medical practitioner information must be up to date and accurate. Addressing the inaccuracies requires both innovative technology and practical, real-world validation.
At Orderly, we tackle this issue by integrating advanced AI with traditional verification methods, ensuring that data corrections are not just fast, but also highly reliable. In the sections that follow, we’ll explore the specific challenges involved in maintaining accurate data, how our approach leverages large language models (LLMs) and direct provider outreach, and why this combination is crucial for solving the data accuracy problem at scale. Our commitment is to deliver precise, up-to-date information that benefits patients, providers, and insurers alike.
Challenges in Current Data Systems
Medical practitioner data, such as that found in the National Plan and Provider Enumeration System (NPPES), is increasingly plagued by inaccuracies. Studies have highlighted the low accuracy of NPPES data, which is critical for various healthcare operations, including insurance claims and provider verification. A report by the Office of Inspector General (OIG) found that provider data in NPPES was inaccurate in 48% of records, with inconsistencies between NPPES and the Provider Enrollment, Chain, and Ownership System (PECOS) in 97% of records. Addresses, which are essential for contacting providers, were particularly prone to inaccuracies and inconsistencies (Office of Inspector General - HHS). Another study highlighted the challenges of using self-reported data in NPPES, noting that inaccuracies often arise due to the lack of verification and frequent changes in provider information such as addresses and phone numbers (Aetna).
Evolving Nature of Practitioner Information
Certain types of practitioner information, like specialty, tend to remain accurate over time. However, contact information, including phone numbers, fax numbers, and addresses, frequently changes as practitioners move between locations or their offices update contact details. This variability complicates the process of verifying practitioner information for insurance purposes, leading to increased costs that are ultimately passed on to patients.
The Internet’s Role in Data Accuracy
With the growing reliance on the internet for up-to-date information, practitioners often prioritize updating their online profiles. This shift makes the internet a crucial source for current contact information. Research indicates that healthcare providers and patients increasingly rely on online directories and provider websites for the most current contact information. This trend is driven by the convenience and accessibility of online resources compared to traditional methods like phone directories (Aetna) (BluePeak Advisors). The Centers for Medicare & Medicaid Services (CMS) has recognized the need for accurate online provider directories and encourages healthcare providers to keep their NPPES information up-to-date as part of their compliance efforts (BluePeak Advisors).
Harnessing Advanced AI: Large Language Models
Orderly Health has harnessed the power of large language models (LLMs) to address these challenges. LLMs, such as those developed using recent advances in artificial intelligence, can efficiently parse and extract relevant information from myriad online sources. Unlike traditional rule-based systems (such as software with hardcoded rules or policies), LLMs(like models from OpenAI, Anthropic and Google) have demonstrated a high capability for understanding and processing human language, which significantly enhances their ability to extract accurate information from complex web pages (BioMed Central).
Yet, this technology is not without its pitfalls. Website information can be outdated or incorrect, and LLMs may sometimes misinterpret or even fabricate data. To address these risks, we’ve implemented multiple safeguards. At Orderly, we cross-reference AI outputs with reliable sources and expert publications, compare them against expected values using test datasets, and continuously rely on human judgment to evaluate random samples. These checks and balances ensure the highest level of accuracy while mitigating the potential for AI-related errors.
Combining Traditional Verification with Modern Technology
A key component of our strategy is the integration of phone attestations with LLM-driven data extraction. Phone attestations involve directly contacting practitioners or their offices to confirm the accuracy of the extracted data. This traditional method serves as a critical validation step, ensuring that the information provided by LLMs is reliable. The integration of traditional verification methods with LLMs has proven effective in enhancing data accuracy. For instance, phone attestations help verify the information extracted by LLMs, reducing the likelihood of errors and ensuring higher reliability of the data.
Additionally, we use specially trained LLMs to moderate and verify the outputs of our primary models. This layered approach enhances the accuracy of our data, enabling us to detect changes in practitioner information within days or weeks and maintain an 80-90% accuracy level. This is a significant improvement over the typical 40-60% accuracy seen in many existing data sets. This significant improvement underscores the importance of using multiple methods to validate and verify data..
The Impact of Accurate Data
Improving Patient Experiences
Just like Emily’s struggle with incorrect provider information, our solutions ensure patients can find the right specialists quickly. Accurate, up-to-date contact details remove barriers, reducing stress and delays, helping patients get care when they need it most.
Supporting Healthcare Providers
For providers like Dr. Smith, accurate data minimizes missed connections and complaints about outdated contact details. Our advanced data correction methods allow providers to focus more on delivering care, not administrative tasks, leading to better outcomes for both patients and practices.
Streamlining Insurer Processes
Insurers, like Emily’s, benefit by cutting the time spent verifying practitioner data. Fewer inaccuracies mean quicker claim processing, lower administrative costs, and a more efficient system that helps maintain affordable premiums for everyone.
Reducing Financial Inefficiencies
Inaccurate data inflates costs across the healthcare system. By maintaining high data accuracy, we help prevent improper payments and reduce administrative waste, saving millions while improving care delivery.
Conclusion
In conclusion, the challenges of maintaining accurate medical practitioner data are significant but not insurmountable. Through the innovative use of large language models, traditional validation methods, and a commitment to rigorous data accuracy standards, Orderly Health is at the forefront of this critical aspect of healthcare administration. Our solutions enhance operational efficiency for providers and insurers, ensuring that patients receive better, more reliable healthcare services. As we continue to refine and expand our methods, we remain dedicated to leading the industry in data accuracy and reliability.
Want to find out more about what you can achieve with Orderly?
Contact us today to request your free Impact Estimate.
About Our Guest Author:
Aaron Beach is currently leading AI/ML products at Orderly Health, where he enjoys tackling complex healthcare challenges with data-driven solutions. Over the past 18 years, he has applied these methods in renewable energy, mobile advertising, email marketing, and fraud detection, resulting in over 40 publications, 1000+ citations, and 2 patents 1,2. In his spare time, Aaron enjoys brewing beer, designing board games, and spending time with his wife and four children.
Enjoyed this content and hungry for more?
Subscribe now to receive the Orderly newsletter directly in your inbox, packed with insights and updates to keep you ahead of the curve in the world of healthcare.