Synthea: Open-source synthetic patient generation

I never expected to see MITRE on the front page of HN! We're actively adding more synthetic data sources to Synthea all the time.

Played around with this in my soon-to-be previous health-tech job and its great.<p>Actually the entire hl7-fhir ( <a href="https:&#x2F;&#x2F;www.hl7.org&#x2F;fhir&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.hl7.org&#x2F;fhir&#x2F;</a> ) standard seems to me quite solid. It would be wonderful if a new cohort of start-ups would leverage it to drastically improve the digital UX of healthcare generally.

Does anyone know if there is an equivalent for generating &quot;random&quot; viable products[1] in a PDM&#x2F;ERP system?<p>I&#x27;m demoing some systems in this field for outside interests, but I can&#x27;t use any &quot;real&quot; data due to ITAR and data restrictions like TC, NC, etc. Wait, what about the ERP? The ERP I&#x27;m developing against has &quot;sample&quot; data that&#x27;s basically useless. Not much better than <i>lorem ipsum</i> pasted across ten thousand cells. Actually, it&#x27;s worse than that, because . . ah hell, this is HN, I won&#x27;t waste your time. People here know what the ERP ecosystem is like. I also don&#x27;t want to build out from a bespoke, brittle ERP - that&#x27;s how we got into this mess in the first place.<p>[1] Like a multi-level BOM that makes sense, or a Service BOM &#x2F; Logistics Database that&#x27;s meaningful. Anything for making pseudo-random PLs that follow MIL-STD-100, which is still considered frickin&#x27; Holy Ground by these people.

Synthea is great! We use it a ton at Medplum - and the sample data that conforms to USCDI is especially useful we recommend for those who are getting started. <a href="https:&#x2F;&#x2F;www.medplum.com&#x2F;docs&#x2F;tutorials&#x2F;importing-sample-data">https:&#x2F;&#x2F;www.medplum.com&#x2F;docs&#x2F;tutorials&#x2F;importing-sample-data</a>

I actually had this idea when I worked for a local HIE. I just lacked the technical competency to make it real. I think this would be incredibly useful for the adoption of FHIR and also learning more about HL7. For security-minded folks this information could be a good tool for tuning DLP and other tools without using real patient data.

I recommend the OMOP schema as a goto standard for EHR data. There&#x27;s an ETL pipeline for converting Synthea output into OMOP.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;OHDSI&#x2F;ETL-Synthea">https:&#x2F;&#x2F;github.com&#x2F;OHDSI&#x2F;ETL-Synthea</a>

Neat! We made a synthetic patient generation prototype a few years ago: <a href="https:&#x2F;&#x2F;pau.treenotation.org&#x2F;synth&#x2F;" rel="nofollow">https:&#x2F;&#x2F;pau.treenotation.org&#x2F;synth&#x2F;</a><p>The challenge at the time was generating realistic correlations between the columns. How do you approach this?<p>I noticed LLMs are a huge breakthrough here with the downside that they currently rely on massive online models. I wonder if someone could train a tiny model that could fit on a local machine specifically to solve the synthetic data problem.

I’ve used Synthea for a whole assortment of small and large projects and it’s been boring in the best possible way: reliable and easy to use.<p>I’ve also had the pleasure of working directly with the team at MITRE that owns it on a consulting engagement (we needed some improvements to it) and they are a delight to work with.

I’ve worked at the intersection of AI &amp; healthcare for years and this has been an excellent tool I’ve leveraged in the past; synthetic data is particularly helpful in the context of healthcare!