Patient Context: PDFs, Photos & Documents

Patient Context lets you attach richer reference material to a patient than just typed text. Upload an evaluation PDF, snap a photo of a medication list, capture a wound photo, or pull in a lab printout. The AI extracts or describes what is in each file and stores the resulting text on the patient. Every AI Note, Auto Fill, and chat response then sees that context alongside your typed eval summary.

Plan requirements. PDF and Word-document uploads are available on every plan (Unlimited and Unlimited+). Photo and camera analysis — medication-list photos, wound photos, lab snapshots, and other images — requires Unlimited+. Trial users have full access to everything during their 7-day trial.

What You Can Attach

• PDF — evaluation documents, discharge summaries, referrals, intake paperwork. Text-based and scanned PDFs both work (see below).
• Document (.docx) — Word documents exported from an EHR, referral letters, dictated notes.
• Photos (library) — pick one or many at once (up to ten per batch). Pill bottles, printed medication lists, wound photos, rash or swelling photos, x-ray screenshots, lab printouts, handwritten clinician notes.
• Photo (camera) — snap a single photo right from the patient editor.

How to Add Context

1. Open a patient (or add a new one).
2. Tap Add Context in the toolbar under the Patient Summary.
3. Choose Upload PDF, Upload Document, Choose Photos, or Take Photo.
4. The AI analyzes each upload and saves a named, categorized text entry. When you pick multiple photos at once, each one becomes its own context item so the AI can categorize a wound photo, a med list, and a lab printout differently in the same batch.

Each entry shows up as a row with the AI-suggested name (e.g. “Medication List”, “Right Lower Leg Erythema”, “Hospital Discharge Summary”). Tap a row to expand the extracted text. You can remove an entry any time from the same expanded view.

How the AI Treats Different Uploads

• Medication lists & pill photos — the AI transcribes drug name, strength, route, and frequency into a clean bulleted list. It does not infer doses that are not written.
• Wound / skin / clinical photos — the AI describes only what is visible (location, approximate size, color, texture, drainage). It deliberately does not diagnose or suggest treatments. The clinician interprets.
• Lab results, imaging printouts, clinical documents — the AI faithfully transcribes the text, preserving numbers, units, and dates.

What Happens to the Image

The image bytes are sent to Google Vertex AI under our BAA, processed, and immediately discarded. Only the extracted text is stored on the patient record. The image itself is never written to the device, the database, or the cloud. This matches how voice transcription works in the app.

Why This Matters

The AI Note, Auto Fill, and patient chat features all read the patient’s context when they generate output. The more accurate and complete that context is, the better the generated notes and form fills will be. Attaching a med list once means every future note and fill automatically knows the patient’s medications. Capturing a wound photo once means future notes can reference the observed findings without retyping them.

PHI

The AI service runs under a HIPAA BAA with Google Cloud. The Remove PHI button on the typed eval summary is unchanged. For uploads, the extracted text is what gets stored on the patient — review each entry after upload and edit out anything you do not want retained.

Supported PDF Types

PDF extraction first reads the text layer of the document, the same way most desktop PDF readers do. Most PDFs exported from an EMR, Word, or a “Save as PDF” print dialog include a text layer and extract instantly.

Scanned PDFs work too. When a page has no text layer (a faxed or paper-scanned document), SOAP Note Buddy renders that page to an image and runs the AI’s OCR over it. Mixed PDFs (some text pages, some scanned pages) are stitched back together so the final entry contains everything in document order. Scanned PDFs are capped at 20 pages per upload — for longer documents, split the file or re-export the relevant section.

Legacy .doc files (Word 97-2003) are not supported directly. Open the file in Word and save as .docx first, then upload.

Tips

• Review every extracted entry before saving. AI extraction is excellent but not perfect, especially for handwriting.
• Use Take Photo for paper docs that were never digitized — it usually beats scanning to PDF.
• Photos of wounds and skin findings should be well-lit and in focus. Include something nearby for scale if size matters.
• Med-bottle photos work best one bottle at a time when the labels are small, or grouped on a flat surface when labels are readable.