OCR software designed for the healthcare sector

OCR (Optical Character Recognition, or Optical Character Recognition) is a technology that makes it possible to convert scanned documents, PDF files or photos into digital text that can be edited and used, as in the case of prescriptions or analysis results.

Unlike traditional OCRs that are limited to transcribing text, Koncile uses artificial intelligence and language models (LLM) to understand the content of the document and automatically extract useful information such as patient name, date, date, dosage, or act codes.

The data structured in this way can be directly integrated into your existing systems such as the electronic patient record, billing tools or analysis.

Yes, integration is a key feature of Koncile's OCR solution.
It is designed to be easily integrated into your medical management software or your EHR thanks to:

• A flexible API : Our API (Application Programming Interface) allows a secure and personalized connection with your health software, whether it is a EHR, a LIMS, or another business tool.
• Pre-built connectors : We are actively developing connectors with the most used solutions in the medical sector, as well as with automation platforms such as Zapier or Make.

This integration allows a automatic transfer of extracted data (prescriptions, reports, care sheets...) to your tools, reducing manual tasks and the risk of errors.

Yes, OCR is capable of processing printed or handwritten prescriptions. If the document is handwritten, OCR uses advanced techniques to recognize handwritten characters and convert them into digital text, while ensuring the accuracy of the extracted data.

‍

Yes, you can extract specific information from a treatment form, such as medical procedures, reimbursed amounts, codes, or practitioner information. OCR can be set up to extract this data based on the sections or formats you are dealing with.

‍

Yes, Koncile’s OCR solution is specifically designed to deliver a high level of reliability when extracting data from medical prescriptions.
Unlike traditional OCR tools, our technology combines advanced methods to achieve accuracy rates exceeding 99%.

OCR enhanced by machine learning
Our OCR engine is powered by machine learning algorithms that continuously adapt to a wide range of prescription formats, layout variations, font styles, and even poor-quality documents (such as blurry scans or fine handwriting). It can also accurately detect complex structures like tables, columns, and handwritten sections.

Large Language Models (LLMs) for contextual understanding
In addition to OCR, we use AI language models capable of interpreting medical context. This allows Koncile to:

• Differentiate between dosage, frequency, and treatment duration—even when these elements are not clearly labeled
• Resolve ambiguities caused by print quality issues or overlapping data
• Identify and extract relevant information from unstructured text, such as treatment instructions or contraindications

The result
Reliable data extraction, fewer manual corrections, and substantial time savings in processing medical prescriptions.

Koncile supports the import of medical documents in PDF formats as well as the main image formats, such as PNG and JPEG.

Koncile intègre un système de score de confiance permettant d’évaluer la fiabilité des données extraites à partir de vos documents de santé. Ce score prend en compte plusieurs critères, notamment la lisibilité du texte, la complexité des informations médicales à identifier et le volume de données traité.

Par exemple, un dossier patient très dense ou une ordonnance manuscrite peu lisible peuvent impacter la qualité de l’extraction. L’algorithme analyse à la fois des aspects visuels (qualité de l’image, netteté du texte) et sémantiques (cohérence du contenu, reconnaissance contextuelle) pour produire un score de confiance combiné.

Ce score vous permet d’identifier rapidement les données les plus fiables, facilitant ainsi les contrôles et la prise de décision en environnement médical.

‍

Koncile’s OCR solution automates the processing of medical and administrative documents by transforming manual tasks into a smooth and secure digital workflow:

• No more manual data entry: Automatically extracts data from a wide variety of documents, such as prescriptions, medical reports, treatment forms, certificates, administrative forms, and supporting documents.
• Intelligent extraction: Koncile analyzes the document structure, identifies relevant information (e.g. patient name, consultation date, procedure code, prescribing physician, social security number), and extracts meaningful data—not just raw text.
• Structured data: Extracted information is delivered in standardized formats (JSON, XML, CSV, etc.) that can be seamlessly integrated into medical management software, electronic health records (EHRs), or administrative tools.
• Seamless integration: Data is automatically transferred to business software or hospital information systems via connectors or APIs.
• Advanced automation: Documents are automatically categorized (e.g. prescription, certificate, proof of identity), business workflows are triggered (e.g. validation of a sick leave or reimbursement), and alerts are generated when data is missing or inconsistent.

In short, Koncile automates the digitization, extraction, structuring, classification, and integration of medical and administrative data—reducing repetitive tasks and increasing reliability across healthcare workflows.

Yes, Koncile offers advanced data extraction customization that can be precisely tailored to your operational needs. With an intuitive interface, you can easily define the information to be extracted, no technical expertise required. You can create custom fields such as “Social Security Number,” “Practitioner Name,” or “Prescription Date,” and assign each a specific data type (text, number, date, medical code, etc.). This improves both extraction accuracy and data reliability.

You can also guide the algorithm using extraction rules, for example by indicating that the LPP code or patient ID usually appears near a specific keyword.