AI Scanning Doctor's Handwriting

Clients' Challenge

• Christian Mutuality (LCM) receives about 40,000 mostly handwritten doctor’s certificates daily.
• These certificates contain essential information such as patient details, reasons for the visit, and payment amounts.
• Manual labor is required to read and input data from handwritten certificates into applications and databases, leading to high labor costs and slow processing times.
• Automating this process would reduce labor costs and speed up certificate processing.

Our Solution

• We implemented an AI-based solution augmented with advanced image processing to automate reading and transcribing handwritten certificates.
• The AI system automatically processes close to 20% of all A-type doctor’s certificates, significantly reducing manual labor and shortening processing times without compromising quality.
• Ensured the AI maintains accuracy comparable to human performance, with continuous monitoring and periodic retraining to handle new data and maintain high standards.
• We custom-trained the AI model based on neural networks optimized for handwriting, significantly outperforming general third-party models. The production system operated entirely in a local data center, with the AI model training on Google Cloud Platform’s Vertex AI.

Impact Created

• Significant reduction in manual labor and faster processing of certificates.
• Speedier refunds, resulting in happier customers.
• Ensured high accuracy and quality of data transcription, comparable to human input, with continuous improvement through regular AI retraining.