Anomaly Detection

Payor Contracting and Credentialing

 

Create a data culture of trust and transparency with DataDx Anomaly Detection.

At the heart of any healthcare data initiative are the people that collaborate in service of patients. Through this everyday process, information is entered in many systems across the organization to track everything from expenses to how patients are scheduled, checked-in, billed for services, and communicated with. This creates a store of information the organization can draw upon to discover best practices and support business objectives.

No matter how diligent the work ethic, we are all human and busy schedules can lead to missed information, data entry errors, or work arounds. This can create inconsistencies in data sets which are often overlooked or go by undetected. The unfortunate result is a lack of trust in the data due to outliers or “bad data” points that do not reliably reflect the true operations of the practice. Also known as anomalies, these data points that deviate from the normal expected range skew an analysis and can lead to poor decision-making.

Just as relationships and teamwork are vital to making patients feel confident in the care they are receiving, getting the most out of your data initiatives requires a shared sense of purpose and trust in the underlying data. With administrative burdens putting additional pressure on practices’ time management, how do you reduce the need to rely upon a “gut feeling” for business decisions with an increased trust in the accuracy of data analysis?

 

Introducing DataDx Anomaly Detection

Cleaning data through anomaly detection methods is nothing new in the technology world. In fact, you are likely familiar with some common applications for anomaly detection techniques, such as uncovering insurance and healthcare fraud[1], or using machine learning for identifying cancerous tumors in MRIs[2]. Astronomers even use anomaly detection to discover new galaxies and unusual behavior hidden within billions of data measurements[3]. Using technology to uncover inconsistencies in data has many practical applications and now it can be used to run your practice much more effectively.

Even if you don’t know it, “dirty data” is likely hindering your organization’s data initiatives. Unfortunately, data quality issues can be incredibly difficult to catch with the human eye when lumped in with thousands if not millions of data points generated by the practice.

 

What Causes Anomalies

Even though there is no way to completely prevent “dirty data”, you can build in processes to minimize them once you understand how they often occur. The most common cause of anomalies results from records being entered which are not fully complete. One way to shore up this problem is to remove the entire record as if it never existed when performing your data analysis. However, sometimes there are values entered in the incomplete data that you don’t want to discard. In this circumstance, imputation is a process used to replace missing data with substituted values based on what would be likely to occur. Whichever method is chosen to deal with this issue, it’s unlikely you could fully prevent similar incidents in the future.

Another persistent cause of anomalies are extreme values that fall significantly outside of normal margins. These outliers can thoroughly compromise the integrity of your data and pose significant difficulty in correcting them. First, you must investigate if the values are legitimate or if it’s a data entry or processing error. In the event the outlier is just a mistake, you can simply treat it like the incomplete data scenario above. If it’s legitimate, then you must weigh the advantages and disadvantages of including it in your overall analysis. If an extreme event happened that warped the numbers for a short period of time, you would probably decide not to include this data because it’s unlikely to occur again and will only skew the long-term analytics.

No matter what kind of anomalies are present in your data or how you might opt to minimize their impact, you can’t fix them if you miss them. Since most healthcare organizations use simple visualization methods for finding dirty data, you can imagine how many outliers could be overlooked. Not only that, but how many laborious hours would it take for human eyes to review millions of rows of data? Anomaly detection software can ensure every past outlier is revealed in a short timeframe and future outliers are automatically reported in real time.

 

How DataDx Uses Anomaly Detection

DataDx utilizes advanced machine learning algorithms to detect inconsistencies within the data, then provides you with all the information uncovered in our Data Integrity Report (DIR), allowing you to investigate the source or root cause of the data anomaly. Once a direct connection has been built to your data source during the onboarding process, your team never has to worry about manual analytics again. New DIRs will be generated automatically so you can be alerted and stay ahead of mistakes and potential fraud. Our Anomaly Detection System will highlight incorrect patient information, unusually large transactions, journal entry or RVU errors, and much more. Our anomaly detection methods will save your office time and revenue from the first day it is implemented.

Often, the best way to understand the impact of a solution is to relay a few real-world examples that demonstrate its value:

Example 1:

Data entry errors can be difficult to identify and have far-reaching impacts. One of our oncology practices had an accountant erroneously enter an account transaction as much smaller than it actually was. This went unnoticed for several weeks until it ended up causing their physician compensation formula to overpay one of their physicians by more than 10x. The DataDx Anomaly Detection system combined with our Physician Compensation module flagged this transaction and kept the issue from repeating itself.

Example 2:

Having appropriate demographic data helps with everything from treatment and billing, to practice analytics. One of our pediatric clients had a patient that was labeled as being 119 years old. This patient had been seen over a dozen times and the issue persisted without being uncovered. The DataDx Anomaly Detection system flagged this record and allowed for the practice to fix the issue in all their systems.

Example 3:

Fraud is a large problem in practices of all sizes. The National Health Care Anti-Fraud Association estimates that billions of dollars are lost to medical fraud.[4] DataDx Anomaly Detection system is designed to identify transactions that are substantially different from what a medical organization has seen before. This provides an additional line of defense against fraud and theft in your practice.

DataDx has revolutionized healthcare software by facilitating a unique all-encompassing conversation between financial, operational, and clinical data sets, which is further enhanced by our anomaly detection methodology. Anomalies within Electronic Health Records complicate accurate billing and cost your practice future revenue. Considering the examples above, you may wonder how many anomalies have been missed at your practice? How much have they cost you?

 

Our Step-by-Step Approach

Once you’ve decided to take advantage of DataDx software, our team of experts will:

  • Survey your systems – Initially, we need to determine if your software platforms have established connection methodologies previously built by DataDx. If so, the onboarding process is quick and painless. We’ll utilize our pre-built logic and relationship with these vendors to get you on board lightning fast. If connections for your software systems have not been pre-built, we’ll take care of the leg work to understand how your vendors prefer to integrate your data with DataDx.
  • Integrate our software – After a connection is built, it’s time for your source systems to integrate with ours by loading your data to a common structure within our platform. This allows for benchmarking and validation among your peers and specific specialties.
  • Run tests – We will then conduct a thorough validation process using our software’s anomaly detection features combined with validations against your platforms’ existing reporting structure. This ensures the integration is functioning properly and trust is established between source systems.
  • Complete Onboarding – Our goal is to have your systems onboarded inside a 6-week window. Often, that timeframe is cut down significantly, especially if we are able to use our pre-built connectors. Most analytics vendors will take six or more months to onboard new clients and require significant investments. By combining cloud-based infrastructure with advanced machine learning and artificial intelligence algorithms, DataDx can bring valuable insights to your practice quicker and cheaper than anyone else.
  • Follow-up Regularly – We will want to check back with your team once per week for your feedback until we are certain we’ve worked out any potential errors.
  • Determine Other Needs – It’s not uncommon for anomaly detection to reveal areas within your healthcare organization that may need some additional support. We will go over our findings and make recommendations to ensure optimal value is achieved through the use of our platform and services. Our consultants are always available to help you achieve whatever goals you have for your practice.

From beginning to end, DataDx takes the time to thoroughly familiarize ourselves with your systems and specific business needs so you can benefit from reliable analytics as soon as our reports begin generating.

 

Final Thoughts

Operating a healthcare organization is complicated enough. The last thing your leadership needs is untrustworthy data to complicate the decision-making process further. Without anomaly detection software to reveal abnormal values, how can you ever have confidence you are guiding your practice in the right direction? DataDx has developed the only comprehensive way to automatically reveal medical and dental software outliers so you can make timely decisions that reflect an accurate analysis of your business. Finally, you can bring trust back into the equation in a way that frees your provider staff to focus on what matters most, patient care.

Contact our team to learn more about how your practice can benefit from the unique innovations only DataDx has to offer.

 

About DataDx

DataDx is a company born from a team of professionals with decades of experience in various roles within the healthcare industry. Over our many years of advising and managing medical organizations, we saw significant gaps in their ability to access real-time analytics for decision-making. Our Founder, Kate Othus, had a vision of better software combined with targeted consulting services filling in those missing pieces to help independent physicians, dentists, and their patients thrive. As a company, DataDx has created an invaluable resource for healthcare leaders to plan for tomorrow based on accurate data today. We offer the most cutting-edge cloud-based software in the medical and dental environment and our Anomaly Detection is a critical feature that ensures your data can be trusted.

DataDx exists to provide real-time analytics that empower you to manage your practice with information you can believe in.

Frequently Asked Questions

What is an anomaly?

An anomaly is an event that occurs outside of normal parameters or measurements.

What is anomaly detection?

Anomaly detection is the process of highlighting abnormal or peculiar events for further investigation.

What is anomaly detection software?

Anomaly detection software uses advanced machine learning algorithms to detect inconsistencies automatically.

Why is anomaly detection good for healthcare?

Because of the complexity of managing medical practices, anomalies can easily go undetected when relying on manual processes. That can lead to lost revenue and less efficient patient care.

[1]
https://www.plugandplaytechcenter.com/resources/detecting-insurance-fraud-machine-learning/

[2]
https://www.frontiersin.org/articles/10.3389/fgene.2019.00599/full

[3]
https://phys.org/news/2021-02-anomaly-pipeline-astronomical-discovery.html

[4]
https://www.nhcaa.org/tools-insights/about-health-care-fraud/the-challenge-of-health-care-fraud/