Gemini gives you control over your medical records archive
- Automatic sorting of medical record pages
- Validation of completeness and correctness: duplicate, missing and wrong page identification
- Conversion to EMR (Electronic Medical Record) standard record formats
- Data mining and visualization
Medical records are legal documents that are preserved in state and private archives for hundreds of years.
Most of these archives host large sections of paper documentation that remain detached from digital platforms for electronically-stored health information.
Gemini is an artificial intelligence-based system for medical records management, capable of automating medical record normalization processes, extracting and classifing relevant contents, converting them to electronic medical record formats and mining and visualizing data
Document page classification
Gemini first inspects all scheduled medical records and then, it classifies every single page into one of the thousands of classes. Gemini integrates most recent solutions in deep learning-based natural language processing and image classifcation to perform robust document page recognition tasks. Currently Gemini can achieve a recognition rate of 99.5% over large datasets of medical record images characterized by more than 2000 medical classes. Each data entry can contain only graphical elements, free form or structured and templated text, drawings or a mix among these cases.
Sorting Document Pages
Gemini sorts all pages to facilitate content accessing and extensive reading. You can choose or design new sorting criteria. For instance each medical record can be sorted by:
- medical class (e.g. anamnesis, diaries, exams and consultancies, consensus forms, surgeries, therapies, …)
- temporal coordinates
- others
Automatic metadata extraction
Gemini, automatically extracts all metadata (name, last name, sanitary code, nosografic index, dates, …) and uses them to validate each page. For instance it can
- identify document pages that belong to another patient medical record
- identify duplicate pages
- make hypotheses on missing pages
Data Extraction, Classification and Visualization
Gemini integrates latest Deep Learning architectures to perform semantic image segmentation oriented to medical images.
Gemini can robustly extracts and classify:
- visual contents of medical reports such as: electrocardiograms, medical ultrasounds, magnetic resonances, brain scans
- text sections: data tables, metadata sections
Each single page is then treated separately to acquire digital data, signals or structures. This step allows Gemini to fully convert digitized medical record images to EMR (Electronic Medical Record) standard record formats.
Gemini also integrates powerful inspection features to allow doctors, researchers and students quickly:
- visualize health data and trends
- mine models
- inspect 3D structures
Gemini is rapidly evolving driven by our customers’ needs. Do not hesitate to tell us how Gemini could be expanded to meet your speficic requirements.