Synthetic Onboarding — Emergency Contact Data

What this document is

The emergency contact form collects a primary and a secondary contact for a new hire — name, relationship, mobile and alternate phone, address for each — plus the employee's physician and a free-text medical notes line for allergies and conditions responders should know about. It looks like the most trivial page in the onboarding packet and is in fact the one carrying health information inside an HR file.

Why generate synthetically

Two problems make this form worth generating. The first is that it is the packet's purest block-repetition test: two contact blocks with identical field structure, stacked, where the only thing distinguishing them is position. The second is that the medical notes line means a real corpus of these forms carries health data about employees and their families, which puts collection off the table for exactly the vendors who need to parse them. Synthetic forms solve both at once.

What makes synthetic data useful

Contacts are drawn from the employee's own simulated household, so the relationships are coherent — the primary contact is a spouse on 47.1% of forms and a sibling, parent or partner on the rest, while the secondary spreads near-evenly across parent, friend, relative and sibling. The medical notes line follows a realistic distribution too: 44% record nothing known, 27.5% no known allergies, and the remainder name a specific condition such as a penicillin allergy, asthma or a peanut allergy. Every field on the form is populated on every document, which makes this the corpus's cleanest fully-dense page.

Training challenges

Two structurally identical blocks with the same labels one above the other is the classic block-assignment failure: an extractor that returns the right values attached to the wrong contact produces output that validates and is wrong, and no value-level metric catches it. Every field is free text, including phone numbers and addresses, so nothing but content and position separates a mobile from an alternate number. And the medical notes line is a short unconstrained free-text field whose content is clinically meaningful — the hardest kind of field to extract reliably and the most consequential to get wrong.

Who uses this data

HR-tech onboarding suites and HRIS integrators capturing new-hire packets, employer-of-record and PEO platforms consolidating records across client employers, and workplace safety systems that need contact and medical-alert data extracted reliably from a scanned page.

Document complexity profile

13 fields on a single page, all typed as free text, with one label-to-value relation per field. The page is two structurally identical five-field contact blocks stacked vertically, plus an employee name, a physician and a medical notes line. Every field is populated on every document, so difficulty comes entirely from block assignment and free-text interpretation rather than from sparsity.

Key stats from our synthetic corpus

Quantitative characteristics of the Onboarding — Emergency Contact documents our generator produces.

How this document co-occurs with others

Rates at which identities in our corpus that produce a Onboarding — Emergency Contact also produce other documents.

Prevalence rates and category distributions above are corpus-derived: they were computed over the 641 employment-eligible identities inside a local synthetic corpus of 1,000 identities generated by SymageDocs' World Simulation Engine at seed 20260421 — the shipped definition gates onboarding forms on active employment. Field, type and relation counts come from the shipped emergency contact definition in the SymageDocs form library. No real employee, contact, or health data was used at any stage.

Frequently asked questions

What data format do synthetic emergency contact forms include?

Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes, field types, and ground-truth values for all 13 fields, with one label-to-value relation per field. Annotations record which contact block each value belongs to, so block assignment can be scored rather than assumed. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.

Are the contacts related to the employee?

Yes. Contacts come from the employee's simulated household and extended family, so a married employee's primary contact is usually their spouse and the relationships stay consistent with the family structure that appears on their beneficiary designation form in the same packet.

Why does an HR form contain medical information?

Because the real form does. A free-text notes line for allergies and conditions is standard on emergency contact sheets, which is exactly why these documents attract handling requirements that make a real training corpus impossible to assemble.

How dense is the page?

Fully dense — every field is populated on every generated document. That makes it a useful control against the sparser forms in the packet: if a model's recall drops here, the problem is not empty fields.

Can I use this data commercially?

Yes. Every contact, phone number, address and medical note is synthetic, contains no real personal or health data, and is licensed for commercial use including model training, benchmarking, and redistribution inside your own products.