Services Rendered
Product Design
Product Engineering
.svg)
Continuing Design
A healthcare revenue management company partnered with Zemoso Labs to revolutionize the claims reconciliation process using artificial intelligence (AI) and machine learning (ML). The partnership aimed to automate the labor-intensive manual process of converting paper EOPs (Explanation of Payments) to electronic 835s. The goal was to create an AI/ML-led system designed for flawless reconciliation of payments.
Mountains of paperwork. Endless forms. Delays in payments.
Healthcare providers struggle with complex and time-consuming claims processing systems, primarily due to diverse document formats, the requirement for manual data reviews, and a high error rate, leading to delayed reimbursements and inefficiencies. This not only affects cash flow, but also diverts valuable resources away from patient care.
The client wanted to automate the reconciliation process end-to-end using NLP (Natural Language Processing) and OCR (Optical Character Recognition). The aim was to build a platform that leverages AI to transform paper-based Explanation of Payments (EOPs) into structured digital data, significantly streamlining the process.
Leveraging our expertise in building nuanced ML models and NLP-led solutions, we co-created a platform that converts unstructured text into structured text, verifies it programmatically, and reduces manual effort, minimizing delays in payment processing.
Zemoso built ML models that could learn and helped our client navigate product innovation — turning a good idea into a great product. Zemoso's ability to collaboratively and iteratively work with the client's in-house team ensured success and a timely launch.
.svg)
.svg)
Our clients love what we do:
We helped the healthtech company succeed by leveraging our tried-and-tested innovation and execution frameworks, proactively managing risks throughout the process.
Collaborative Design: We conducted a GV-inspired design sprint to map the ideal user journey for streamlining EOP conversion and data extraction, and created a tangible prototype for the claims automation platform.
Architecture Sprint: Designed the platform's underlying technical structure, leveraging Apache Spark and NLP libraries for scalable data processing, efficient text extraction, and robust machine learning, laying the foundation for a high-performing system.
Text Conversion using OCR: Used Apache Spark and Python API (PYSPARK) to handle large-scale data processing, perform text extraction from scanned PDFs, store the extracted text data in Apache Hadoop HDFS for efficient storage and accessibility.
Text Extraction using NLP: Leveraged NLP libraries like SpaCy to process and extract relevant information from the text data and transform extracted data into structured formats using Pandas, enabling easy comparison and analysis.
ML Model to Automate Verification: Developed machine learning models to automate verification procedures with predefined rules and logic to identify anomalies or deviations from expected patterns in EOPs.
Accuracy Assessment and Improvement: Used Spark's machine learning capabilities to measure system accuracy and performance metrics while continuously updating and refining the model using PySpark based on analysis from discrepancies found.
Feedback Loop and Maintenance: Established a continuous integration/continuous deployment (CI/CD) pipeline using Docker containers to track and improve performance.
This partnership helped accelerate claims processing for healthcare providers, leading to reduced costs, improved accuracy, and faster reimbursements. The success of this partnership showcases the power of automation in revolutionizing healthcare revenue management, and Zemoso’s expertise in helping healthcare companies adopt complex technologies like AI and ML.
P.S. We have strict Non-disclosure agreements (NDAs) with many of our clients. The data, insights, and capabilities discussed in this case study have been anonymized to protect our client’s identity and don’t include any proprietary information.
.svg)
.svg){kind=icon}
