Healthcare patient education videos from clinical guidelines

A GP practice receives the same questions from forty patients each month about post-operative care, diabetes management, or asthma triggers. Each time, a nurse spends fifteen minutes explaining what's already documented in clinical guidelines. The patient leaves with a vague recollection of what was said. Three weeks later, they ring back with the same questions. Now imagine: each patient walks out with a personalised five-minute video, narrated by a professional voice, that they can watch at home as many times as they need. The gap between clinical knowledge and patient understanding costs the NHS time, money, and patient outcomes. Patient education videos address this directly, yet most clinics never produce them because the workflow feels overwhelming: extract information from guidelines, write a script, produce a video, publish it. When done manually, a single three-minute video takes six to eight hours. Automating this process reduces that to under an hour of human oversight. This workflow uses PDF documents as the source of truth, converts them into structured scripts, generates professional narration, and produces video with a virtual presenter. The entire chain runs with a single trigger and minimal human intervention. For more on this, see Healthcare clinic patient education video creation. For more on this, see Healthcare patient education video creation from clinical....

The Automated Workflow

This workflow uses Make (Integromat) as the orchestration engine because it offers reliable file handling, strong PDF integrations, and good webhook support. The flow moves through four stages: extract content from clinical guidelines, write a patient-friendly script, generate audio narration, and produce video.

Stage 1: PDF ingestion and content extraction

A clinic member uploads a PDF clinical guideline to a shared cloud folder (Google Drive or Dropbox). Make monitors this folder and triggers on new file detection. The file is passed to Chat With PDF by Copilot.us via its API.

CopyPOST https://api.copilot.us/chat-with-pdf/query
Content-Type: application/json { "file_url": "https://drive.google.com/uc?export=download&id=FILE_ID", "query": "Extract the key patient education points from this clinical guideline. List them as numbered bullet points. Focus on what patients need to do, not technical clinical information.", "model": "gpt-4o"
}

Chat With PDF returns a structured list of key points. Make stores this output in a variable for the next step.

Stage 2: Script generation

The extracted points feed into Copy.ai's API. Copy.ai generates a patient-friendly script that sounds conversational, not like a medical textbook. You'll request a specific tone: friendly and reassuring, typical duration (two to three minutes when read aloud), and target audience (adult with no medical background).

CopyPOST https://api.copy.ai/generate
Content-Type: application/json { "prompt": "Write a patient education script for a three-minute video. Tone: friendly, reassuring, non-technical. Content: ${extractedPoints}. Include a brief introduction, three main sections covering the extracted points, and a closing call to action. Format as a single continuous script for a presenter to read aloud.", "max_tokens": 600, "temperature": 0.7
}

Copy.ai returns a finished script. Make stores this in a text variable. At this point, a clinical reviewer might step in and approve or edit the script before it proceeds (a simple approval step in Make pauses execution until a human confirms).

Stage 3: Audio generation

The approved script is sent to ElevenLabs Turbo v2.5. This creates high-quality narration with natural inflection. You choose a voice (professional, calm, gender-neutral options available) and request British English pronunciation.

CopyPOST https://api.elevenlabs.io/v1/text-to-speech/21m00Tcm4TlvDq8ikWAM
Content-Type: application/json
xi-api-key: YOUR_API_KEY { "text": "${approvedScript}", "model_id": "eleven_turbo_v2_5", "voice_settings": { "stability": 0.5, "similarity_boost": 0.75 }, "language_code": "en-GB"
}

ElevenLabs returns an audio file URL. Make downloads this and stores it temporarily.

Stage 4: Video production

Hour One accepts a script, audio file, and presenter preference. It synchronises the narration with a virtual presenter, adding subtle gestures and expressions that feel natural. The API call looks like this:

CopyPOST https://api.hourone.com/v1/generate-video
Content-Type: application/json { "script": "${approvedScript}", "audio_url": "https://elevenlabs-output.com/audio-file.mp3", "presenter_id": "professional_presenter_001", "duration_seconds": 180, "background": "clinic_white", "output_format": "mp4"
}

Hour One processes this asynchronously. Make stores the job ID and polls the API every 30 seconds until the video is ready.

CopyGET https://api.hourone.com/v1/jobs/${jobId}
Content-Type: application/json

Once the status shows "completed", Make downloads the video file and uploads it to your clinic's content library (Google Drive, Vimeo, or a custom LMS).

Final step: Notification and logging

Make sends a Slack message to your clinical content team with a link to the finished video, the original guideline reference, and a reminder to publish and track views. A database record logs the guideline title, date created, script version, and video URL for future auditing.

The Manual Alternative

If you prefer to retain more control or your clinic's workflow requires clinical sign-off at each stage, you can run this semi-automated. Use Chat With PDF to extract points, then have your clinical educator write the script directly in a Google Doc (no Copy.ai involvement). Once approved, feed the script to ElevenLabs and Hour One manually via their web interfaces. This takes roughly three hours per video but gives you tighter control over messaging and allows your team to refine scripts based on patient feedback before committing to video.

Pro Tips

Rate limits and batch processing.

ElevenLabs and Hour One both have rate limits; if you're producing multiple videos in one day, stagger submissions by 30 seconds.

Make's throttle function prevents API overwhelm.

Cost optimisation for audio.

ElevenLabs Turbo v2.5 is cheaper than their premium tier but produces excellent quality for patient education. If your clinic produces fewer than five videos per month, the pay-as-you-go plan is more cost-effective than a monthly subscription.

Error handling.

Set up Make's error logging so failed jobs trigger a Slack alert to your team immediately. Hour One sometimes takes longer than expected on longer scripts; set your polling timeout to 5 minutes, then trigger a notification rather than failing the workflow.

Version control for scripts.

Store all generated scripts in a Google Drive folder with timestamps. This lets you track what changed between versions and quickly revert if a clinical reviewer flags an issue after video publication.

Presenter consistency.

Hour One allows you to lock a presenter ID. Use the same presenter across all your videos so patients develop familiarity and your clinic builds a recognisable content identity.

Cost Breakdown