EduAIttractive (ORBISOxJOTOZO)

We’ll create a bespoke patient video based on the specialist documentation: the doctor selects the diagnosis and patient group, the AI suggests scenes, and the doctor approves them.

The problem Doctors explain the same information to patients over and over again, but they have little time for each one. Under stress, patients cannot remember everything, and when they look things up online at home, they find information that frightens them even more.

Solution EduAIttractive transforms specialist material (a document, transcript or recording) into a short video tailored to a specific patient. The video follows on from the consultation with the doctor, explaining things simply, step by step, and clearly distinguishing between what applies generally and what depends on the patient’s specific condition.

How it works

1. Source material. The doctor uploads a PDF, Word document, text, transcript or recording. They may optionally select a diagnosis from the ICD-10-CZ code list.
2. Target audience. Age (from children to the over-75s), situation, language style, length, purpose and tone of the video, format (landscape, portrait, square) and image style (3D or flat illustrations).
3. Three perspectives on the same task. The AI proposes three scenarios: a step-by-step explanation, a practical perspective, and the patient’s questions and concerns. The doctor edits these in the editor (up to 20 scenes).
4. Approval. The doctor goes through the scenes one by one. They approve each scene individually, and approval is tied to the exact wording of each sentence.
5. Video. A video is rendered with voice-over, subtitles and illustrations.

What we believe and what we don’t The video content is based on recorded specialist materials, not on the internet as a whole. Each sentence includes a reference to its source and a measure of support (stated in the source / transcribed / beyond the scope of the source). A sentence that goes beyond the scope of the source cannot be approved without editing. The AI makes suggestions; the doctor makes the decision.

Technologies Azure OpenAI (scripting), Azure AI Speech (transcription and voice), an image model on Azure (illustrations), Remotion (rendering), Next.js.

What we’ve achieved The entire workflow, from uploading a document to the finished video, is fully operational. It includes an animated tutorial using sample data.

Next steps

• A library of approved reference materials linked to ICD-10 diagnoses, so that doctors do not need to upload anything.
• Measuring the benefits: time saved by doctors and patients’ understanding after watching the video.
• Regular clinical review and content updates.