Form Digitalization

Gameday Men’s Health was created for one simple purpose: To help men recover from low testosterone by optimizing their hormone levels so they can live life on their own terms. Set in a man-cave environment – plush with flat screen TVs, leather sofas, refreshments, and a stock ticker – Gameday Men’s Health helps men like you get back to the top of your game. When you need to feel your best in life, be your best at home, and perform your best at work – it’s Gameday.

Carlsbad

Location

Gameday Men's Health

Healthcare

Industry

Challenge

The company was processing a vast number of medical forms, relying on over 1,200 offshore resources to manually read, interpret, and digitize patients’ medical conditions and prescriptions. This approach was not only time-consuming and costly but also prone to human error. The company needed a more efficient, accurate, and scalable solution to automate this process.

Solution

Our team developed a customized Optical Character Recognition (OCR) solution tailored specifically for handling complex medical documents. The solution consisted of multiple processing stages, including text detection, filtering and preprocessing, text recognition, and post-processing. We fine-tuned state-of-the-art models such as CRAFT (Character Region Awareness for Text detection) and MORAN (Multi-Orientation Recognition with Attention Network) to handle the diverse and often challenging nature of these documents, which included misaligned text and handwritten notes.

Given the variability in the documents, we also implemented custom fine-tuning to improve accuracy in recognizing difficult handwriting and resolving alignment issues. To streamline the process further, we developed a mobile application that allowed annotators to scan documents directly, automatically aligning them for digitization. This mobile app significantly reduced manual intervention and enabled real-time digitization of medical records, making the entire process faster and more reliable.

Discover Our Approach

Customized OCR Pipeline

We implemented a multi-step OCR pipeline with stages for text detection, filtering, preprocessing, recognition, and post-processing to maximize accuracy.

STAGE 1

STAGE 2

Model Fine-Tuning

We fine-tuned models like CRAFT (CNN-based) and MORAN to handle complex text structures, including handwritten notes and misaligned text common in medical forms.

Mobile Application Development

We built a user-friendly mobile app that allowed annotators to scan documents, align them automatically, and submit them for digitization in real time.

STAGE 3

STAGE 4

Scalability and Automation

The solution reduced the reliance on manual labor, enhancing scalability while significantly cutting operational costs and minimizing human errors.

Improved Accuracy and Speed

By integrating advanced OCR models and automating the digitization workflow, the company was able to reduce processing times and improve data accuracy, leading to better patient outcomes and operational efficiency.

STAGE 5

Interested in automation?