About the Client
DentalXChange is a leader in building Electronic Data Interchange (EDI) solutions for the dental industry. Their specialized solutions deliver value to a wide network of over 50 thousand dental offices. Their products simplify the interactions between dentists, their patients and insurance providers. DentalXChange’s secure portals process a massive volume of nearly 45 million EDI transactions consisting of more than 30 million dental claims annually.
Current Document Processing Solution
Currently DentalXChange uses an OCR solution to extract data from ADA claims forms. Some challenges inherent to the existing solution are around flexibility, accuracy, efficiency, cost, and scalability.
The previously existing OCR solution relied on a templated approach where every claim or account statement form was processed using a corresponding template. This resulted in a two-fold challenge: First, as the number of dental offices is in the high thousands, the variety in the statement forms used is high. Managing these templates was a time-consuming and effort intensive process. Similarly there are six variants of the claims forms with other variations based on the dental office processing it. Second, if there is a change in the structure of the form, then the existing template doesn’t function optimally.
There were challenges in terms of accurately extracting information while minimizing the delays and inefficiencies caused because of the dependence on manual verification. The previous process was dependent on humans reviewing the documents and deciphering certain fields manually which were not extracted by the OCR solution accurately. As many 40% of the total forms had to be manually verified, resulting in hours worth of wait for downstream processing. In the ADA claim forms, important information about dental procedures performed for a particular patient is contained in tabular format. This flows through to the insurance providers for downstream processing. The OCR solution fell short in detecting the table contents accurately. The prior solution was also less scalable because it depended on templates; thereby requiring additional time for a new document to be onboarded. It was unable to optimally adjust capacity by scaling up or scaling down to maintain steady, consistent performance at optimal cost.
Quantiphi’s Solution
Quantiphi has developed and deployed a document processing platform to accurately extract information from the ADA claim and statement forms along with providing confidence scores for each field extracted. Subsequently, Quantiphi also continues to maintain and manage the platform to continuously improve the AI models used in the solution and to empower the client with responsive, accurate and scalable support for additional documents. Some of the key AWS services that power Quantiphi’s solutions are as follows.
 |
Amazon Simple Storage Service (Amazon S3) is storage that is designed to make web-scale computing easier.Quantiphi’s solution uses this service as a repository for the input and output files. |
 |
Amazon Elastic Compute Cloud (Amazon EC2) is a web service that provides secure, resizable compute capacity in the cloud. It is designed to make web-scale cloud computing easier for developers. Quantiphi’s solution uses this service for validating input images, building docker images for speedy new deployments. |
 |
Amazon Elastic Container Service (Amazon ECS) is a fully managed container orchestration service that helps you easily deploy, manage, and scale containerized applications. This service plays a critical role in having various containers poll SQS queues for messages to spin up containers to manage scale with great efficiency. ECS is used to scale up the solution to spin up extra instances having more containers through ASG (Auto Scaling Group) only if the set instance consists of a set number of containers that are being used at its full capacity. Eventually when the load decreases the extra containers in these instances start getting destroyed, it automatically starts scaling down. |
 |
AWS Lambda provides serverless compute service which performs the file format conversions and invocations of Textract processing. This empowers the solution with the ability to run code without provisioning or managing servers and managing scalability and runtimes efficiently. |
 |
Amazon Textract is a machine learning service that automatically extracts text, handwriting and data from scanned documents that goes beyond simple optical character recognition (OCR) to identify, understand, and extract data from forms and tables. Amazon Textract is a base on which the brain of Quantiphi’s solution resides for claims and statement forms processing. Custom ML models intelligently decipher the information extracted to ensure that various types of fields such as key-value pairs, checkboxes and tables are detected with maximum accuracy. The confidence scores for each field is also calculated. |
 |
Amazon Relational Database Service (Amazon RDS) instances have been used for storing the document and page level data processed through the solution. |
 |
Amazon Simple Queue Service (Amazon SQS) enables the solution to be truly decoupled by offering secure, durable and available hosted queueing service. SQS helps in the integration of various components of the solution such as pre-processing components, Textract processing and ECS. It also plays a critical role in making the solution robust through appropriate error handling mechanisms. |
Impact of Quantiphi’s Solution
The solution developed by Quantiphi is helping DentalXChange to improve the accuracy in extracting information from dental claims. The computed accuracy has been in excess of 98% consistently across multiple variants of claim forms. This can be observed across all the types of fields in the forms viz. key-value pairs, table contents, check-boxes and also unique fields such as the numeric values which are crossed-off to be selected.
The solution is dynamic to the type of document selected i.e. it is able to identify the type as well as the variant of the document being processed. The solution has been designed keeping in mind the massive volumes of documents that DentalXChange processes on a daily basis. The AWS Platform lends itself perfectly to the scale and flexibility requirements from the solution.
The solution has been designed keeping DentalXChange’s teams in mind; to assist them in processing claims quickly and efficiently. It provides the team a granular view of the confidence scores of each field which is extracted; thus helping them in the downstream processes of review and consumption of the data. The reliance on human review through the solution developed by Quantiphi has fallen drastically to 1% from the 40% volume earlier. This results in faster processing and drastically improved customer experience. DentalXChange’s solutions powered by this solution will be able to cater to their customer expectations in a matter of few seconds as opposed to hours needed previously. To sum it up, Quantiphi’s solution helps DentalXChange process a large volume of documents responsively, accurately and dynamically while minimizing the overhead involved in processing and scaling.
To learn how you can leverage Quantiphi's document processing solution for your business, get in touch with our experts.
