Development of a system for recognizing medical issues from images
We used computer vision and machine learning to automate the processing of medical documents
32desk contacted us to develop a system for recognizing medical tests from photographs and then conducting analytics based on the data obtained. This solution was supposed to automate the processing of medical documents, reduce data entry time and reduce the likelihood of human errors.
We have collected a dataset of 10,000 images of medical tests. To expand the amount of data and increase the model's resistance to various conditions, data augmentation methods were used: rotation, distortion, noise addition, and other techniques.
The pilot project has been successfully implemented in a network of dental clinics across the country. The system made it possible to automate the processing of medical tests, significantly reduce the time required to enter data and reduce the likelihood of errors. The clinics received a tool for rapid analytics and decision-making based on relevant medical data, which improved the quality of patient care and the efficiency of medical staff.
