Healthcare Interoperability: End-to-End Streaming FHIR Pipelines With Databricks & Redox
Summary
- FHIR streaming pipelines: Redox and Databricks enable fully-streaming ETL of FHIR bundles from EHRs directly into Unity Catalog volumes, making healthcare data immediately queryable and AI-ready.
- Actionable near real-time insights: Teams can implement use cases like preauthorization automation and sepsis risk alerts by leveraging Serverless SQL, Streaming Tables, and the VARIANT data type.
- End-to-end interoperability: AI/BI Dashboards and Agentic Frameworks with SQL UDFs enable writing FHIR messages back to Redox for direct EMR integration without data silos.
Healthcare Interoperability: End-to-End Streaming FHIR Pipelines With Databricks & Redox

Talk by Matthew Giglia, Solutions Architect, Databricks ; Tim Kessler, Field CTO, Redox, Inc
Redox & Databricks direct integration can streamline your interoperability workflows from responding in record time to preauthorization requests to letting attending physicians know about a change in risk for sepsis and readmission in near real time from ADTs. Data engineers will learn how to create fully-streaming ETL pipelines for ingesting, parsing and acting on insights from Redox FHIR bundles delivered directly to Unity Catalog volumes. Once available in the Lakehouse, AI/BI Dashboards and Agentic Frameworks help write FHIR messages back to Redox for direct push down to EMR systems. Parsing FHIR bundle resources has never been easier with SQL combined with the new VARIANT data type in Delta and streaming table creation against Serverless DBSQL Warehouses. We'll also use Databricks accelerators dbignite and redoxwrite for writing and posting FHIR bundles back to Redox integrated EMRs and we'll extend AI/BI with Unity Catalog SQL UDFs and the Redox API for use in Genie.
Here’s more to explore:
Production ready data pipelines for analytics and AI: https://www.databricks.com/solutions/data-engineering
The Big Book of Data Engineering: https://www.databricks.com/resources/ebook/big-book-data-engineering-2nd-edition
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements
Chapters
0:00Introduction and speakers1:53The role of Redox3:45Data interoperability cycle6:01Overcoming ingestion barriers7:16Real-time data architecture8:39Healthcare use cases10:13Real-time clinical insights12:35Data ingestion demonstration17:24Exploratory analysis with AI20:35Prior authorization write-back24:10AI agent framework27:39Closing and call to action
FAQs
How does the Redox and Databricks integration work?
Redox delivers FHIR bundles directly into Unity Catalog volumes where data engineers create fully-streaming ETL pipelines using SQL, the VARIANT data type, and Serverless Warehouses for near real-time processing.
What clinical use cases are enabled?
The platform enables preauthorization automation, sepsis and readmission risk alerts from ADT data, and clinician notifications—all with sub-second latency for operational decision-making.
How does data flow back to EMR systems?
AI/BI Dashboards and Agentic Frameworks use SQL UDFs and the Redox API to compose FHIR messages and write them back to Redox for direct push-down to integrated EMR systems.
What data type simplifies FHIR parsing?
The VARIANT data type in Delta Lake makes parsing FHIR bundle resources with SQL significantly simpler compared to traditional approaches, accelerating data engineering velocity.
Full transcript
0:00 Introduction and speakers
So, uh, really awesome that everyone showed up. Last session of the day. It's just between us and happy hour, I think, at the data after hours party. Um, hopefully everyone's had a great summit. Uh. I'm one of the solution architects here at Data Bricks. with me. I've got Hey everyone, field CTO at Redox. All right, so uh this is this has been a partnership that we've had data bricks and redux for quite a long time of and I have actually gotten to know each other the last couple of years and uh being able to do uh things with fire interoperability.
Um it's something that I've always wanted to do when I was a customer and now it's really a privilege to get to be able to do this as a solution architect at data bicks and working with a partner like redox. So great really quick introduction who I am. I mentioned I'm the field CTO at Redux. Been with Redox for the last nine years now. Um and actually before joining Redux I used to do implementations at Epic where I had the pleasure of helping people navigate how difficult it is to integrate data both outbound from these ecosystems as well as writing it back inbound.
Click a little sticky. So we'll a little quick introduction to myself. Of course, uh I am a a solution architect here at Data Bricks, but my background is I spent almost 15 years at Excelis Bluec Cross Blue Shield in upstate New York where I live with my family. Uh did everything you can imagine uh at that BlueC Cross Blue Shield. Started as an analyst and then I was eventually the first data scientist, principal data scientist, then principal analytics and data architect.
Uh and what you're going to see here with this healthcare data with interoperability, it's really what I've always wanted to build and uh now we get to do it. So Tim, great. Awesome. So just a quick introduction on Redux, who we are before we dig into really the meat of this presentation. So Redux, we exist to make healthcare data useful. And where we really look to enable that is, you know, we think about what the high-tech act was 2009, quite a while ago.
1:53 The role of Redox
So we've seen rapid adoption of healthcare data systems, EHR systems. We finally digitized healthcare. actually getting it out of these ecosystems into platforms like data bricks where you can really drive a lot of value is still quite difficult. So we started about 10 years ago in, Wisconsin. Uh if you're familiar with the healthcare industry, you might know another large company in there um in that town. Uh I worked there previously doing patient access revenue cycle implementations.
I think I was employee number 15 at Redux or so. I think like 19 of the first 20 were all former Epics. So we really got our roots kind of learning how to deploy these ecosystems, get them live up and running and then we wanted to find a way to get the data out of those ecosystems in a friendly developer friendly manner to actually make this data useful and flexible. So today we're over 150 employees. We're transacting over two billion transactions per month through our platform.
We're connected to over 10,000 healthcare organizations streaming birectional data. So not just pulling data out into data bricks but also writing it back into the EHR, putting it at the point of care. So when I say like pushing this back into the EHR, I actually mean like writing notes into a system, pushing flowheet values back in uh hitting clinical decision support web hooks to be able to pop up data for a provider at the point of order entry.
So really talking about getting this data back into the EHR at the hands of the the provider or the end user's fingertips to really drive as much value as possible ultimately shift some of the decision- making made by these folks within that ecosystem. When we think about the interoperability cycle especially within the data engineering ecosystem we really view Redux's role as handling that connectivity that translation of the data and the normalization of the initial normalization of that data.
3:45 Data interoperability cycle
So connectivity what I mean this is healthcare we still see a lot of VPN based connections a lot of MLLP overVVPN to trans or to transact event driven HL7 messages if you've ever worked with an HL7 message you're going to see it here today it's a kind of old pipe delimited format that can be a bit annoying to work with. So we translate that to uh JSON value. Specifically, we primarily work with fire data. We're really looking to push fire within the broader ecosystem, but of course we can support translating this to pretty much any format you're looking to work with.
And then we handle normalization of that data as well. So you've worked with EHR data, if you work with many, many different systems, you might have Epic in the inpatient setting, you could have Athena Health in the outpatient setting. You could have some other system in kind of a, you know, physical therapy ecosystem. Epic might send, you know, for patient sex, they might send male or female. Athena Health might send one, two, or three.
That other system might send an alpha code. We'll normalize those all into a consistent value so that when it lands into data bricks, there's a lot less work that you need to do to get that data ready to work with. But then when we look at actually driving a lot of value from the data, Redux doesn't log into your system. We don't deploy anything directly within data bricks. That's where we really rely on partners to help enrich that data, drive a lot of value and you know actually build with that data.
So when I think about leveraging AI in the healthcare ecosystem, when I think about a lot of what exists today, um that connectivity can be incredibly difficult. I have another slide that I'm not going to bore you with that highlights all of the archaic legacy ecosystems that exist today that you really kind of need to support in order to tap into these systems. Redux handles all of that out of the box. It's all within the Redux toolkit.
So, you know, when you when you deploy Redux in your ecosystem, you don't have to think about supporting IG profiles or XDS. B exchanges or how to translate XMLbased CCD into, you know, 45 FE resources to start making sense of that. I've also mentioned that you know I've really highlighted the EHR aspect of it but healthcare data can live in any system and connecting to all of those systems can be a bit different. So when you work with Redux we really want to deploy all of the tools within that ecosystem to handle that connectivity so you're never building custom.
6:01 Overcoming ingestion barriers
Uh I also want to mention that we're really really quick about doing this because we've built this all already because we have over 10,000 integrations with healthcare organizations throughout the US. A lot of the pipes already exist which means we don't necessarily have to build this from scratch. So in scenarios where we're already integrated with a healthcare organization, we already have data flowing. Routing that into a data bricks environment takes about 20 minutes. however long it takes to deploy that data bricks environment, however long it takes to actually kind of stand up that new destination.
So that can be extremely fast. But even if it's a net new integration from scratch, it's about three weeks uh to get that deployed. And that's mainly just going around finding the network administrator, finding the bridges integration analyst, finding analyst who's going to determine what data flows. And again, our role is how do we make that as easy as possible? How do we put the tools in your hands? you have that control to ultimately build that connection yourself too.
7:16 Real-time data architecture
This kind of puts everything together. So when we think about where Redux sits within that ecosystem, our goal is connecting to all these source systems, providing the tooling to make that incredibly easy simple to go from zero to production level go live in as quickly as possible so that you can start landing data in data bricks immediately and start le actually driving value from that from that data that's been kind of locked in many of these different silos for so long.
Before you move on, I just want to kind of point out something on this slide. So, uh, I was actually in the HLS industry booth, uh, earlier today and yesterday. I got several questions about each one of these types of clinical data. Several of you are in the room right now. Do you handle X12? The answer with data bricks, of course, is yes. Can you handle XML? Of course, the answer is yes. But if you don't use Redox, then you have a data engineering problem for each one of these types of file formats.
Potentially now a data mart or a, you know, set of silver tables for each of these particular types of files. by having it go to Redox and having Redox then change that into fire. Now you have one standard way of ingesting the superset of your clinical data which then means you only have one data engineering pipeline to have to worry about to create a clinical data marks. You get value out of this data really quickly.
So you can be using it for machine learning insights and you know affecting patient care at the site of care. So of course after all that right uh we're here at the data bricks conference right data and AI summit and as you can see you know all of the different types of data that's going to be flowing into your lakehouse the idea here is that this clinical data is going to be able to unlock all of those use cases that um that you've always wanted to do or if you're a payer for example and you're thinking about plan to plan um you're going to be a great payer that's going to be receiving lots of new members to your to your uh to your uh you commercial Medicare or Medicaid lines of business.
8:39 Healthcare use cases
And what that plan to plan really means is that you're going to be able to know everything about that member before they even start with your plan. On day zero, you're going to be able to know from their clinical information they'll be you're receiving and fire. Uh things like which he denominators they're in, which HCC's for risk adjustment they might already have, which care plans they may uh currently be on from the prior uh prior plans uh care management team. all of the analytics and insights that you currently run now if you're a payer that requires claims analysis uh and claims run out to actually come in before you can actually do that plan to plan is going to allow you with fire to be able to know all those things and run all those analytics before the member even joins your plan.
So being able to take that data in, ingest it easily, parse it, make sense of it, so you can use it for all of these use cases that you want to do with it, including your existing use cases you do today with your members. That's what data bricks is going to make easy, and we're going to show you how that works with the direct integration redux data bricks in a few minutes. So this is what the demo is going to look like in a second here.
10:13 Real-time clinical insights
So of course, uh, there's going to be EHRs. If you are a provider, you very likely already have an EHR that you have to enter all your information in. Uh but it can be very difficult uh even if you own that system to pull information out of it. And often you're going to want to do things like real-time operations, real-time hospital operations. Redux is able to, you know, look at Epic or look at Cerner and listen for certain transactions, um, such as potentially ADTS you may already be generating today to send to other partner organizations, turn those into fire, which we can then land in data bricks of course with our lakehouse architecture and then use that for machine learning.
Right? So we actually have a customer it was a provider system that uh wanted to use their own ADTS which they were generating for payers of course uh to predict um and forecast their bed usage right so that they would actually be able to do nursing uh staff staffing uh much easier and then uh once they had that information they needed to post that back into the EMR itself so that they could then uh you know make sure that they had that information there those insights for uh for the folks that were actually working inside Epic itself right so That is a use case that we've actually delivered together.
Redox and data bricks. If you are a payer, uh you're making observations about your members all the time, right? So, uh we are maybe looking at somebody who is has an inpatient stay at a behavioral health facility, for example. And if you have a behavioral health impatient stay, you know that there's a heat measure that you have to have that person see a specialist within seven or 14 days. That's incredibly difficult to affect change on if you're not looking at real-time information, right?
So, but those ADTs, right, if you've actually used those ADTVs, HL7V2s, they can be actually quite difficult to really use and turn into enterprise having it go from something like those older HL7 V2s into fire in one clinical uh real-time ingestion fully streaming. Now, you'll be able to make that enterprise. to be able to use that for uh for those predictive models and then send that potentially even send those observations back using redox into epic concerners so that at site of care you know potentially a social worker or BH specialist or even schedule the BH specialist on behalf of the member uh that's all possible with this joint uh partnership that we have.
So this is what going to be what the demo is. Um and we're actually going to be using the prior authorization mandate as the example. Uh so for folks that are aware prior off mandate you will be required to use fire uh to send and receive uh these prior offs right so the mandate says that within 72 hours we have to be able to uh to respond to these and if you are a payer um you know that most of the time we actually approve uh claims we don't deny them most of the time and uh you know if we could have done it faster we would have a long time ago right because all of our stakeholders really hate the fact that it takes so long for something like a prior authorization to happen uh including the doctors and our members and our patients itself right so with the power of something like redox and data bricks we're going to be able to help you meet this mandate a little bit faster so that's what's the demo is going to be so what I'm going to do right now uh unfortunately they don't let me do a live demo here at summit so this is pre-recorded but uh this is my screen here I'm going to be logging into my uh version of redoxia pit play there we go perfect okay so I've got my username password there.
12:35 Data ingestion demonstration
I'll log in and of course uh this is the data bricks redux environment. I'm going to be using a dev environment today but of course redux has staging in production as well. Previously I've made a destination. This destination actually represents a volume in unity catalog which uh means that uh every transaction that is going to come through redox could be sent uh again from the EMR directly to data bricks and unity catalog with that volume where you'd be able to stream ingest it.
So what I've done is I've actually done uh I've actually sent some ADTS directly to uh our our uh Unity catalog volume here. As you could see from those logs, they were already successful. So, it's very quick. It turns it from that whatever format it might be into fire. Here you're seeing a workflow with a file arrival trigger that immediately gets kicked off as soon as those files hit your volume. And what we're going to do is using only SQL, no specialized software, right?
No Scala, no Python, no RDDs, literally just SQL with the new variant data type. We're actually able to completely stream that in using autoloader and some dynamic SQL running against a serless SQL warehouse and that's going to create first our bronze tables uh using variant which will completely parse that. You'll see that here in a second and uh hopefully the glare in the screen is not as bad as is for me right now.
I'm showing you this uh variant bronze table here and we'll look at the sample data and this is the way I like to do my bronze table sort of key value pair. You saw the file metadata there a second ago. This is that variant uh fire. This fire column is a variant data type. Without knowing anything about the JSON schema of this data, we're able to completely parse all the way down every single element of this data. Now, if you understand uh fire and you you're in the room and you probably do, the entry array is where all the really great resources are.
And you can have uh a CCDA that's been turned into fire and that might be hundreds of you know resource types in that entry array and the patient resource need not be the first one it might be the 15th one right or you might have one that's just for scheduling uh scheduling a clinician or PCP for the patient that might be only five or six uh resources in the entry array and they may be completely different what that means is that the normal big data schema on read you know from JSON where you know what the schema is does not work for fire which is why you've always needed specialized software before.
What you're seeing here is really in the last seven months with the advancements at data bricks for data bricks SQL is that now you can just use regular old SQL uh or you can use Python if you want as well but no more specialized software and this will work to be candid for any JSON if it works for fire it means it works literally for anything which is really cool so this workflow that you're seeing here uh that got kicked off um that allowed us to actually create both bronze and silver tables one silver table for each each resource if you come to the industry booth you would see this workflow and you may have already seen it preview earlier today.
All of that now is powering this AIBI dashboard. Many people don't even know that latitude and longitude is even in uh the patient resource for example, but you know right off of there we're able to create a map, pull that out. We can see some information about the encounters and we can even see some information about claims. Now for prior up the uh the service request uh bundle that would be created has a number of different resources but one of the resources that will uh come in a prior authorization uh style request is going to be actually a claim resource itself which sounds kind of silly because it's not really a claim yet.
It's what the claim will be once it is approved for example or what we're requesting the claim to look like. The way that you will know if it is a prior prior authorization claim is if the use actually says uh prior authorization. So a few minutes ago in the little AIBI dashboard we identified which ones actually had by filtering on the use. We saw prior off. Now we have a prior off bundle that is in my system. And here what we're doing I'm actually showing how you can make a map really quickly.
17:24 Exploratory analysis with AI
So uh which is great. So that's AIBI there right notebook. Um, you might have noticed there a second ago that was the bronze table that I was writing that off of, which and that's just to prove that you can actually write SQL directly against any element of those of that fire bundle immediately if you wanted to. I wouldn't I don't recommend it in reality, but it's just to prove that it is possible. You'll notice at the top there, there's that bundle uh UU ID.
That's the one for the uh prior authorization uh widget here. And you could just see now we're just again looking at the um looking at this specific bundle now as the fire variant column and this uh entry array is very specific for this type this particular type of bundle which is that service request bundle and uh there's the patient resource and so forth. So now what we can do is just using SQL again we can lateral explode this and actually um you know using that SQL start to parse that out pull out the resource type pull out the full URL which is really the primary key of the of the resources itself uh and then be able to write SQL against that even easier be able to filter it here what you can see is uh I've actually exploded it a little bit so that now the claim resource the person resource those are now columns if I want to do that because maybe I want to take this information and I want to shoot that over to my utilization management team uh such that uh my my my nurses and my medical directors maybe have some information that they want to get from that.
Of course, the bundles themselves, they don't tell me what the descriptions are, but I can use the power of uh the data intelligence platform here and the AI query function to ask it things like hey uh take these diagnosis codes and take the procedure code that I'm being requested and give me a summary of that just using an LLF. So here I'm I was using metal lambda 3 uh 70B and really quickly in SQL I'm able to actually now parse everything about this particular uh uh service request here.
You can see the diagnosis codes. It's a professional claim. Uh there's the diagnosis code summary. There's some knee injuries for example. And then the actual description, the long description, super long description, not the long description we're looking for used to, right? which just is usually 10 words if you're familiar with uh uh the ICD9 ICD10 sort of descriptions, but actually like a full description like something that maybe a medical director or um nurse would really actually be able to use as far as uh you know what's actually happening including things like how long that uh uh that particular service even takes.
So next point this right here. Okay, there we go. Oh, okay. So, now that we Where's the mouse to play this? Sorry, technical difficulties. There we go. Okay, so now we need to actually respond to this, right? So let's say that we've sent this information all the way off to our utilization management team and they now need to you know they've approved this particular service request and now we need to send it back as fire uh in order to actually let the uh let the uh clinician know that they are good to go.
20:35 Prior authorization write-back
This could actually happen really in sight uh at site of care. Um what you're seeing here is a little bit about the how you can actually get the schema of those things from variant but I want to talk a little bit more about prior off. If you think about it, because the fire bundles are able to actually take any clinical information, right? any encounter you know that previously had happened such that the criteria for this particular service would be uh would be uh approved any diagnostic reports any clinical notes that maybe back all this kind of stuff up all of that could be packaged in these bundles such that when the payer receives it if they're able to parse this really easily in the way that we can now with data bricks they instead of using things like interqual criteria they could actually just you know run analytics right on there and maybe approve it immediately and then shoot it back to that maybe even at sighted pair uh maybe that you know maybe the patient is actually able to even have that service be done you know while they're in that same visit potentially right so the ability for things to be truly real time even for something like prior authorization is possible with fire because of the ability for us to bring in so much great information so how do you take your lakehouse data and turn it into uh fire we use db Ignite which is an accelerator that we've developed here uh at data bricks and you can use a low code mapping like you see on the screen, this claim response for example, there's some outcome that I need to create and that's going to come from some other column potentially like event outcome or something like that.
Uh and then once you have that mapping, you're able to then throw it into our mapping manager and then that will turn it into the actual fire uh fire resources itself. Now the way to think about this low code mapping, you're very familiar probably with sourcetoarget mappings from your data management practice. You have a source column and table. you have a target resource, you know, information uh piece of information in uh fire.
All you need to do is do that, put it into that format that you saw. The transformation has happened by dbgnite automatically for you a few minutes ago. You may have saw that all those curly brackets and arrays were all presented there for that claim response. If you've tried to type that manually like I have and I know has, that is incredibly complicated. I would much rather have something like DBignite do that. And now we have another accelerator that um we actually the blog for this uh was just released just a couple of days ago.
My colleague uh is the one that wrote this package. This is the Redox write package. It's in Python where you're able to interact directly with their API. So as long as you have those fire bundles ready to go generated from the lakehouse, you can really easily post these back. So now this is an observation as I mentioned a minute ago. You may be making observations either again as the provider itself or as a data science team at a payer and now I maybe I want to post that back.
Maybe the probability of sepsis has increased because someone has transferred from one unit of a hospital to another unit of a hospital. I want to make sure that folks know that has increased so we can maybe take care of that immediately. There you can see the status 200 redox would then have taken that and then shot that right back over to the EMR so that we can do whatever we need to do. Here I am logging back into Redux now where I can see that the uh with the logs that the observation crate has come through and they've processed that and then they redux would understand how to push that right back into Epic uh in that way.
24:10 AI agent framework
Now this um what's really cool is that this particular uh package, this particular API, it's not just for posting things back. It also is for retrieving things. So you could have an agent framework. Of course, we're at the data AI summit. So, it would be remiss if you didn't talk about AI agents a little bit. Um, but suppose you were a care manager and you were working with a patient or a member and you were just curious like I wonder if there's any new diagnostic reports or any new lab values, right, that might be out there.
Uh, it's totally possible to actually take this uh and you're going to see it here in a second. You can take this API and actually serve it using Mosaic model serving with a MLflow Pyunk flavor. talk to me after if you want to learn more about that in great detail and then serve that as a SQL function which then you could use uh to then um for any for any AI agent to be able to use. So this right here is just a little bit of code that's showing how you would do that.
And then uh so this is a recording so I can't just jump to the to the fun part here but you can see that uh this is the way the steps that it's done. And you know what's great about this then is that then you can ask it any type of question that you want. So for example the diagnostic report or you could say um you know uh tell me the patient uh the most recent patient encounter. It would go out to Redox receive that information from something like Epic and then send that back to you where then you'd be able to display that know for example in a care management app or utilization management app.
U let's see this is a little too much detail probably for folks. So let me see if I can fast forward this. Okay, here we go. There's the make request. Um, so now we're able to use again AI query right there using parse JSON again that's turning it into into variant, right? So this is an API that's being called in SQL in the data bricks SQL editor and it's able to um you know again post that observation for example and because of that parse JSON turning into variant now anything that used to just be JSON now you can just write SQL against that turn that into a table function for example and let's see when I get to the good part really want to fast forward this but okay.
So there you can see it's uh that's the variant there. The mouse doesn't show up on here otherwise it'll totally fast forward through this. Apologize. Okay, here's the SQL function now. Now here's here's the example I was talking about. Now I've got this patient. I want to search for their latest diagnostic reports. So, I've just got a little CTE right here that just says I want to post for a uh diagnostic report search for that patient.
And then I can pass that into my Redux request SQL function, which again is just a SQL version of that API. I wrapped it in Mosaic model serving in order to do that. And then I can uh I can then write SQL against this. And then what you're going to see here is that it's going to return all of the diagnostic reports in order for this particular patient. Sometimes it's going to be riched uh riched text as you'll see there's some information about the where it came from and sometimes it'll actually be the binary for a PDF and when it is that binary for PDF you can then do a spark write binary and actually have physical PDF itself.
27:39 Closing and call to action
So you could be surfacing there's the binary for the PDF you could actually be surfacing the actual diagnostic report itself as an image as a PDF image in your care management app. That's the power of what Fire is able to do. It's even able to take things like this binary here. Okay, I think that uh that concludes the demo. Awesome. And uh if anybody really wants to see that demo live, I know done it probably like what 200 300 times today alone.
Yeah, we needed to make it fit within the 40 minutes. So, uh I'll just I'll just mention a few things. So uh the workflow that we kind of breezed over really quickly I mentioned fully streaming uh using autoloader creates the bronze creates the silver tables. We offer a fire ingestion workshop now at data bricks where we will you can talk to your account team about that. We come on site with a lab workspace and we teach your data engineering team how to make that workflow.
So I go in great detail every element of that SQL. why I chose the things that I did. Can't obviously do that here. It's like a four- hour workshop. So, uh, and then in the industry booth, you were met with me yesterday or today or maybe even tomorrow. Um, we did cover it a little bit, but I didn't go through every element of that workflow, but we absolutely can come on site and teach you how to do that. Um, and then, uh, as far as things like the, uh, everything that you saw today is fully on my GitHub and open source.
You're able to go and use that and practice, including on synthetic data. We have ways of generating that for you as well. So you can practice adjusting as Fire JSON. Yeah, absolutely. And if you're interested playing around with the Redux portion of this that ultimately streams, we saw HL780T. We also have XMLbased CCDs that you can stream in a fire format into data bricks. And we'll soon be offering X12 uh payloads as well to be able to stream that.
That's all available completely free on Redux's site. So if you just sign up for a developer account, you can play around with this tooling immediately. Um we highlighted uh so highlighted the one capacity planning patient throughput use case you know how to route patients through the right location within the organization before digging into the demo. Then the demo focused on prior authorization that birectional rewrite and exchange uh all in real time.
There's a ton of stuff that you can do with data in real time both with the power of data bricks as well as redox powering that data feed both into data bricks as well as writing back into the ehr. So I include a couple in here that we've implemented for our partners already, but there's basically any scenario where you can imagine consuming data, maybe automating a query to pull a clinical summary as soon as that patient's discharged and you receive that ADTO3.
That's all stuff that can be automated directly within Redux to then land that data in data bricks and ensure that you have all the content necessary. So, if you're curious about any of these use cases or any of the capabilities, please don't hesitate to come up and ask us uh either right now or at the happy hour we're going to be doing right after this. So, if you're really looking to grab a drink, which I know I I am after today, um we'll be uh we'll be head just a block away and everyone's welcome.
Learn more about the Databricks Data and AI platform.
The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.