Transforming Prescription Verification with Google Cloud AI
Key Challenges
Initial reliance on manual interpretation of handwritten prescriptions was labor-intensive and error-prone. Integrating and customizing AI models to accurately interpret diverse prescriptions was challenging. Improving OCR accuracy for reliable patient and medication detail extraction was essential. Complex data processing tasks, such as image cropping and medication matching using FuzzyWuzzy, added to the difficulties. Training the custom AI model on Google Vertex AI to meet Kauvery Hospital's specific needs required significant adaptation.
Key Results
Document AI accuracy improved from 40% to 60%, and Vision AI accuracy increased from 60% to over 80%, thanks to a custom model on Google Vertex AI. The automated system reduced manual effort in prescription verification, streamlined patient and medication information extraction, and minimized manual entry errors. Overall efficiency improved, allowing staff to focus on critical tasks and reducing time and labor involved in decoding handwritten prescriptions.
Overview
Kauvery Hospital is a multi-forte hospital chain in India, turned into facing a vast undertaking in verifying affected person prescriptions. This verification process concerned manually decoding and extracting records from handwritten prescriptions, which turned into exertions-in depth, time-eating, and liable to errors. The hospital wished a extra green and correct way to streamline this procedure and decrease the guide attempt concerned.
Ankercloud led the implementation of the AI-driven prescription verification solution at Kauvery Hospital. Ankercloud managed the project from start to finish, including custom model training and post-processing techniques.
Challenges
Kauvery Hospital aimed to automate the verification of prescriptions, which previously relied on manually interpreting handwritten documents. This effort sought to streamline the extraction of patient and medication information directly from prescription images, enhancing accuracy and efficiency in processing. By adopting an automated system, Kauvery aimed to minimize errors with manual entry.
Solution
Kauvery Hospital uses YOLO v7 to crop prescription images for patient and medication details. Google Document AI and Vision AI perform OCR with improved accuracy through matchmaking. A custom model on Google Vertex AI annotates prescriptions for precise extraction. Post-processing involves bounding boxes & FuzzyWuzzy for medication matching. This automated solution enhances efficiency & accuracy.
Business Outcome
At Kauvery Hospital, automation of prescription verification using YOLO v7 for image cropping and Google's AI technologies significantly improved accuracy and efficiency. Document AI enhanced from 40% to 60%, Vision AI from 60% to over 80% and Vertex AI's custom model boosted precision through detailed annotation. These advancements reduced manual effort, minimized errors and optimized efficiency.
.png)
.png)
