Experienced Corporate Recruiter with a demonstrated history of working in the computer software industry and healthcare. Feel free to connect and ask any questions you might have!View the profile
About the talk
This video is part of the R/Medicine 2020 Virtual Conference.
Thanks for waiting everybody. I'm stuck in Cudahy and Ice my extreme pleasure to introduce. Patrick messiah's mdpc, who's the vice chair of clinical operations of laboratory medicine, at the University of Washington, School of Medicine. And one of the leaders and driving automated large-scale. Covid-19 testing in his lab alone has done more than 500,000. Is that right? Patrick 500,000 test test, which laboratory medicine an enormous scale from starting with nothing and he is going to show you how he used our to help orchestrate. This just ginormous effort,
take it away. Patrick. Sorry for the delay switch computers. Hopefully, I'll be okay and no technical issues. So I just I'm talking about Ashley, this is not a Are eccentric talk. This is really an open-source Center talk that we use a variety of Open Source tools and infrastructure to help us expand the scope of a million in Laboratory Testing at LabCorp in Madison and kind of this pandemic are the five increasing our capacity of testing. We're really increasing. Are there
are severe limitations as I think everyone is aware. It's rough country brought the world with providing timely testing so we spent a lot of time and effort using some of the infrastructure. We've actually developed over time to rapidly deploy applications that can help us do some of the gas that we see if I'm going to spend a lot of time talking about out of the general problem and solution. I'm not going to go into too much technical detail about anyone pollution is Ghent. There's a lot of python and a w s in this talk but, sir. There's some are that I will cover at the end. Really
are has been has helped us out the Whipple operation but I think it's interesting to go through the various Solutions. This is really a lot more of us and a clinical informatics perspective on how to make things work under difficult circumstances. I'm so I can give this talk without really acknowledging the folks in you to be so doctor graninger and his team really over saw one of the earliest covid tests by country. And it's my opinion that would take a lot of strong arguments, that convinced me
that our UW biology laboratory was, it is not that was not the, most are laboratory in the US. So as soon as the genomic sequences were, we're out in the wild for stars, Kobe to. They were on a test of element at the end of January. We're pretty close to it. Having a test available and then it really spent the majority of February going back and forth with the FDA to try to get this test available in and out there I am because of regulatory and other issues were not able to push out as early as we like, but it's really their effort and
your folks, like red peppers. The manager got to keep Jerome who's the director for a very large family, medical testing and really love the role but I T lab analysis. Have helped help, make sure that their effort is is maximal utilized to get testing out there. I just kind of frame, you know our perspective in in the pandemic on on the importance of testing we have are symptomatic, high-risk populations. Will we have folks who are also asymptomatic and really not the goal is to identify as many folks who are who are tearing cars. Kobe to the virus
test, test many populations as we can reasonably with Jamar Supply, constraints, and identify those cases isolate them and then you'll Cascade is process of understanding who they been in contact with and identify other. Folks, you might be infected so that we can go spread and I'm contain as much as general framework. That many countries have you stopped texting is not the only piece of the of the solution but it is a critical piece of the solution.
As laboratory ends when we think about the soil testing process, we kind of think about this. The schematic here at least many of us. Think about the schematic here, which would she want to be familiar with the brain drain? The general idea is there is some action from a patient, some issue with a patient position works with them, two to figure out problem and that triggers are allowed for the order. We have our kind of total testing process going through ordering, you have to collect a sample. Make sure it's properly, identified transport. That's the laboratory with, in
the laboratory, we need to prepare that for analysis, which included critical step of accepting the sample into our log information system. Analyze that we were Port that. And then, you know, whether it's a complication where there's an actual conversation, or a validation, which the laboratory and taken their understanding of of the test and injected that into a dealer, can be an automated report that goes out to the This is this is kind of General workflow that we work within every day for medicine and from
informatics electronic fan point we will think about our laboratory information system which really coordinates all those activities in the last in our case study medicine and then we are very often and 2020 we're now working with the electronic health record and so are our organizations run Friday or other electronic health-record and those are not involve both of those systems are not involved in every step of the process of critical critical role that. So, typically the way that we move data from one system to another
call hl7 interface, Health level 7 is a standard of moving around. I m near will receive And that is, that's kind of the ideal state within the laboratory where we get electronic order as we send electronic results back and that's where we want to be. But hl7 can be challenging in that it is a disaster Solutions between these pressure systems typically of wire lot of expertise to build these pipes in the system. So he's messaging standard but it's not completely standardized to the point. That you can't just say I'm going to build an interface from this system, the system, and
I can just apply a standard in and be done with it. Every institution has got there, different flavors out of the oven and in some cases customization it was critical for some of the cases. It's just a result of dorkly how they've implemented the sander. And so what we see in the attic space on a day-to-day, Send a message to system. Is that building these types of each other messages or are we a high overhead activity? Most of our interface project, take weeks to months and we
need to respond to the pandemic on a on a shorter time. So, the default mechanism by which we receive boarders is actually in 2020 is still a piece of paper so it will get us a sample. And a piece of paper, is example, of Our Stars, Kobe to requisition. And you know, this will be filled out with a lot of fish. Sometimes they populate something filled with with with data from a PDF and a very often are still lots of written elements that form and there's this critical step on intake. In the laboratory of it can be a bottleneck of
getting the information from your piece of information system. Analytical team in place. If you don't have a huge fan on a staff and the bass off of data analysis which are to understand through put, we we know we have a general idea, depending on different flavors of interphase orders, a 1 ft. The process of 250, 250 orders for shift and it's a manual orders as much lower somewhere between 15 and 50. So whenever possible, we want to shift manual order the paper requisitions, into an electronic process. The largest grouper accessioning. We want
to think about that total testing process. We have a number of people fix them out of analytical equipment. So we want to think about what our approach has to take every step of that total testing process and make it more efficient, so that we can get more test through. And even if we, even if we can't get a week, we can't steal all of our analytical asking, if we want to do what we can, to make sure that we can get results out in a timely manner, that contact, tracing your testing in tracing methodology. Relies on us, being able to get results.
One of the think about this whole testing process, one of the areas where we have applied up, some soft words around ordering and that exception aging process, bro. Initially early on in the, in the Fantastic, we were working with a Skilled Nursing Facility to really identify that you'll be identified that the guy from the virus and so many of our local residents were not already registered, UW medicine, patient. And then I overheard the kind of doing that on a on a mass population of
jail. So while we were developing the the kind of the physical infrastructure to go in and go in and help screen these these locations, how can we do this quickly with everybody copies caliper Staffing, we don't want to deal with. April, requisitions by the many of these abilities. Even if they don't have an electronic health record, they do have rosters available in the structure for Matt. So are on a starting point for a lot of these improvements were, can we take a spreadsheet? Get orders into our Electronics in into our lap for information system, the spreadsheet.
I am so just to briefly touch on some of the resources that were help on doing this week. We have been investing in expertise and time and developing an Amazon web services resources. So don't have a place can do in Warfare securely within 8 of the US and have that hooked into our dedication. Our organizational authentication methods that was one key enabler and then another enabler was really are dokis back, which is Have a great day briefly it's just that
the general principle is a platform as a service so we can we can you package all of the components that go around the point weather application? In a way, where you can focus really on the application and the application logic and can I have a template for all of these other pieces that allows you to just focus on your application? Have Panama, a template that will you do to two-and-a-half quickly and then have that hooked into our authentication. So, that's kind of our starting point. For a lot of these Solutions is really having this ability to, it's been up a nap, right? The
logic and then deploy it on AWS secure manner. So the our kind of first round at at using this type of solution was to work with Dr. Arm and his team and Post Acute Care where we would schedule drops. You get spreadsheet ahead of time and then we developed Call up files or upload application. That would allow us to take a roster do data validation and understand your what day is missing what's going on. Break. If we try to send data to SunQuest and really pay some the place of what of the
role that the electronic health, records are electronic health record, does allow data validation for you and so when you send the transaction from a DHR to your Lis, you don't have to worry about, you know, about missing data or data display that that is not the right format for your field and you can collect missions as though. I'm so we felt pretty light weight. To replace that data, validation function, and then have a workflow to get that through interface engine generator strain of hl7 messages, based on that spreadsheet. And then and then
send us some Quest. So this is just a quick look at what the what the app looks looks. Like it could very quickly took like a date and to develop this and then we've been working on the logic for the validation since as we add additional Fields like Insurance, he's like that, Pretty simple and pretty, pretty straightforward, use case, but I think the reality is, if you are working kind of an institution, it teams that are stuck with electronic health record, or the lack of information system, is not really this wrapping solution to help us together. And,
in many organizations, at least knock knock knock this over all can help us through these efforts. I'm an heiress election site processing or not manually typing information and allows us to capture really that full information, straight off of out of the record from from ability. We have deployed as not only for our skilled nursing. Facilities is actually came into play as Sunnah play with, with doing screening in other settings like a, we have a lot, a large number of a Alaskan fishing boat that are based in Seattle. And so, we've been able to deploy it rapidly deploy
screening for the boat by I'm working through this, this mechanism orders in and bring them quickly. So far we've been able to scream 30,000 people based off of this roster upload mechanism. We take what we were thinking about this might as well that wasn't nice solution. And, you know, that's deployed at the end of story. That was kind of kind of thought that it was a question that will do and that will be yet, but actually a couple weeks of applying this at the city of Seattle came to us and and asked
whether we could help support reproductive. There was a federally supported testing site that was going to be going away in three weeks and we needed to have a replacement for that really quickly it. So I really want to capture all of the day and then minimize our lab for South Apartments over at all cost, large volumes of paper that we can't. I can't login. I'm so in light of these challenges, we quickly came up with a with, now, your solution to do our spreadsheet solution, using a song about call Windsor company
self-scheduling. But after the city was to make it as easy as possible for someone to get a task part of that is, can you provide access or can you provide a channel for them to go on a phone? Say I know I'm such an attic, I need to schedule something I installed and rear is really slow. Wrong traditionally they were working with physician Health Physician Offices. That would kind of lightweight front end to get people sell scheduled to tour schedules but we found with that sentence of the u.s. digital response
team and a rapidly doing everything to do a full rrfp on the time frame. What time do the rapid evaluation said? This is a good solution to help us get structured data that we can feed into this date of allegation and Order pipeline. So just thought I'd walk you through the process. Looks like we have a very prominent city of Seattle covid testing website, you can go through and it gives you information of who's eligible for testing. You can text you that and then find one of these sites, you know, whatever is most convenient for you. You can go to that, you can
go to the go through that front, end up, schedule, yourself in or time when you enter required information about the station and then you're one of the lucky things for for the city of Seattle sights, as we had a large number of these emissions. Testing site that were used here they were they were frequently use until December 31st the regulations Around Mission in Washington State. No longer have a perfect start to setting up the covid, testing sites, repurpose decide, and then
patients, go through drive-thru. We have registration text to her kind of doing some training up front. You see the gentleman here putting a sticky note on the car and so they're different colors. Take note that indicate the level of completeness of the information of the patients as put into their selves record for the downstream registration check to review. And then help make sure that we have a complete contact information in insurance information and things like that. And then they're Seattle Fire Department actually stopped collection that for this.
I'm so your house that physical workflow tie into our electronic workflow. We have our samples with barcodes. We scanned those into a field in the solve application and that That's a payload, can be carried across to a file validation, and then we set up Sunquest. But importantly, we have a stamp light on a fire like that, actually followed the whole process and generates that order. So that when the sample does his efforts collected and transported the lab, we can actually just scan that into our system and retrieve the order. And so that
cuts out, all of that cuts out, a huge amount of work login, paper, prepositions. But so far with a system in place we've we've performed a hundred fifty thousand test for the city of Seattle site before the site just opened up yesterday. I'm also working with Kane County area, not testing desert down in South King County to take the same system and apply it to underrepresented populations as well. Another couple of some of the writing solutions that we've deployed for for our Prien local phase of testing getting samples and into the into the lab. And again, we have a
disparate systems that are that are that are that are casted for their purpose. But there is not really a great, a simple way to tie them together in the way that we needed it because that's where we use are custom software development of the court that you actually think about with in the laboratory, when you think about those profits in the same way. So we do have automated testing platform, but I told it had severe Supply constraints. So many labs are still using a variety of manual processes with with instruments that are kind of tied together with our. I actually tried the other
very well with the data flows. And so we kind of have a similar type of a issue. analytical side within the lab so you don't stay the same general solution of have a locally about assets blue, so that we can send data back and forth between the manual transcription and Daniel are So I can dream in the lab. As you have this, your beautiful automated line and you your samples come in the lab, they magically go on the line and you'll get all these Robotics and take care of all the work. And then do it will issue a result and you have very little
process or hyperbole That's not the case in them like your lab. So this is the one of the things that we need to think about is your lab and particularly after worried about contamination, but for these rest for samples in particular, you're dealing with samples of a swab that can't go directly on an instrument. For there, has to be some manual processes to transfer your media. Liquid, it is in to another two and then go through a series of other instrument, for example, of the PCR, the thermostat on the right, that's really the reality. For most molecular labs are probably come
out there that have more automated Solutions on main line, but in general most molecular, they're not doing so you have glue these these bushes together in front of your data, back and forth. So, Jessica at a high level two to think about what happened for a star's. Kobe to viral test. You process, the transfer that to a, to that those tubes have identifiers on the alarm system. The physical work for the samples. So, you're picking me up to go from these these tubes for not writing an automated enshrined. If you go from these tubes, that have a sample to 96 Weller
384-well play and then left the truck preparation by Liberty Handler for lucky to have a automated Handler and then you move that to an extractor. And so that's a that's a instrument at a purpose-built for getting you play Acid which were trying to detect, Audi samples. So you can increase your sensitivity and then you're from the other some preparation and then it goes on to this thermocycler, which is that? That PC I'll process that may be familiar with where you cycling temperature up and down and then your exponentially amp amplifier.
Many of the defrost, many of the processes, particularly on the, on the back end of this were very manual. So one of the challenges that we identified early on where that to get the to get the sample, the sign to the right. Well in the software the bathroom and standing in the the two are the labels on the containers have to constantly. And there go, they were taught to do the racks and 92 at a time to go onto the plate. So this is, if you think about
taking that workflow, that might be manageable for a few plates today for a technologist. But when you have to multiply that 2000, then Another issue that we identified early on. Was that the results transfer was again when you're working at Molly volume. You can't you you can manually review the output from from the results from the thermal cycler. They were kind of have a manual workflow of putting a ruler behind under each result and a manly transcribing that to the laboratory information system. Do I deal with something that we've been nice to know about prior to
the pandemic, but your was manageable with with free fire volume, when your scaling-up 2000, today's that's just not sustainable. So those are a couple of the issues identified early on and I'm using the same kind of bottle of Transformer cars and moving around data to to reformat things to get them in the appropriate for myself. As an example, we might have a XML file that output from our Hamilton liquid Handler. I am we need that course that so that it is so that it can go on to
our upload it to our PCR. Instrument is a experimental, layout is accurate. And the right samples are reserved that mapping of the what's the well on the plate. Ample, that would you not? Well, So you're very pretty pretty straightforward operations here. See here is a python script in the application and we talked with Zach and it's really that operation. You take your output file, you drag and drop it on the light app and then if it produces the outfit
that you need, that, you can go through review and then you can download the files that you need to go onto the instrument. So that's that's just that that actually, that addresses, that that 92, the scanning 92 samples at a time portion, the workflow and then address the resulting piece of the wharf can pretty straightforward challenge of of just transferring data that you have data on a certain formats year on the very far left. And then you want to toss that in a way that Our Generations, which is our Middle where that used to, to
talk to the insurance and 11 for information system. Did that solution actually has a has a standard format for four flat files to basically send the flat file today in animation. And then file is the Dolphins. A lot of information to the pretty straightforward. Onset of operations began, having your nice web app deployment, I can help us take that that script and then turn into something that that the Astrologist on the bench to interact with quickly. I get the outfits they need and send it to the lab. So it,
we think about this whole process. Basically, you're pretty early on the response, really? I would say, within the first week of the response, we make sure that this whole dataflow was was much more automated in, and then you helped us really know we were talking to, for early on, in the pandemic response in the water class. You was weird. We are thinking about 300 to 500 path based off of Ali's manual processes and then this really allowed us to scale up to 2,500 per day per day, based off of
How important was that ability to scale for for the response for this is just showing very early in the pandemic based off of publicly available data. This is showing on the percentage of testing performed. By UW virology Lab First against you, look at Blue Line at total testing in Washington state. A washer that point we were really, it was awesome. The state line that was performing, testing a state public, health lab, and then. And then the red is a kind of testing for the u.s. do in the first couple weeks were
actually doing 25% of the testing for all of the US initial and supports my contention. I think we were really having the it support helps jail on Elsa Border Collie. So I won't spend too much time on it, but what this is actually also enabled up to do very quickly is to as the pool samples. So please think about that, custom dataflow and do it and the ability to transform data in the various formats. We can take a cooling protocol. So ignore liquid Handler
and instead of putting a single sample in a single, well, in the in the microwave we can put for samples in the microwave plate and we have a mapping that we can ask for from the handle, it says, yo, look here at the east line as well and then you can see the self-lighting associated with that. And so you can take that in that information that mapping and then generate your what we need to go on to the PCR instrument. And in this case, we just concatenate the the individual samples to make it fully transparent.
Thermocycler. And then, once you have this, your those samples when you're ready to upload, results of the hell asked, very simple operation in this break apart, is this nation? And then you can explode out of school and do it. And then you offer detected pool, we don't send anything to the lobby close in that rises from releasing a result. But then the negative pools, go to the information system output from the page that I can print out. And then go retrieve the samples, and
reflects those two to the, to the knothole platform. So, using this overall solution, we actually pull is now have our default method for running samples in for outpatient. So you'll disability to do the transformation has been able to Take us from your babysitter. Using the existing laboratory-developed test methodology. But I'm really thinking about the analytical side. I think another aspect of the response is really getting results to patients at Soca in this day and age.
Not only are we communicating results from the laboratory to to the provider to the doctor. We are also providing information to a patient patient patient portal but in the settings that I've described so far is a lot of the settings are not traditional. UW-Madison traditional Healthcare settings broke in the challenge of how we deliver those results to Patient. You really need it would you want the patient's Nellis results in more on those results? Also have a
subset of healthcare employees whose data is not an electronic health record. So you really want to be able to Glide with all the portal, patient portal, do have some overhead and no phone calls for every single results are really when you're typing answer. So we are solution was to rapidly develop but application. That really had the right lab dataflow so that we can pour don't read one couple that physical work watching sample. Are our general solution was to kind of have a pair of codes that travel along with, with the sample.
Do we have a basic, we have a QR code on a 16 digit code that we kind of, we both have one Deanna to debark a code based off of the 1D barcode. Go on a laboratory requisition. The 2D barcode decode, a link that has the unique know that each of these approaches you need. So unique, retrieval code, embedded within, just scan, the barcode and access the result. So, we know we also have some other physical details around how the labels are printing. Help, help, keep things straight with
the right there, but the general workflow for this is you have your laboratory requisition code, we develop basically, Fields in the land formation system, to support scanning in this code. For this part of our routine process for exceptional sample in the lab and into the field generator file that contains both the results and the retrieval code. And then using AWS back, we, we generate objects for a GS3 office for each of those results. And the key is something you have and something that, you know, so in this case, we have the beauty bar, code is the physical
Unicode and then the Piece episode date of birth was all. So, this is what the lightweight portal looks like you. You'll go to go immediately and door code and then you ask your date of birth, and then you get your results back and we can also provide turn the first level of guidance based off of, The results predict populations of folks and make sure they understand what it means for them. Are So that solution is a little bit more involved than some of the other kind of lighter-weight, solutions are solutions that I described but I am requires a significant amount of support
for printing and distribution distribution because we need that ensure that each of the keys or each of those QR codes is you need for the really the only one source of Truth. For those end up having a field, support half the field for called, pick up people can retrieve the results and then also involves close partnership with a colleague to make sure they understand how the workflow work and ensure that every pair of QR codes goes to the right places. We had a
look at some of the data from this, and tell me some interesting Trends in in that, on meeting time from home to first, visit on the site from collection, was about 6 to 7 hours. And then, in general meeting time from when the result is actually available to one the patient and retrieve the result in the round 2 to 3 hours at a time. Obviously longer for overnight. We don't have, we didn't have the data from before, July 1st, and easily, easily accessible fashion. But
since July 1st, more than 35,000, generally have had limited down times, are other issues using deployment. So far, talk a lot about the custom applications that are mostly written in Python but no. But I think the important aspect of a lot of this response is if it's really taken some expertise and some development over time, working with these Solutions, taking things that we've developed, they are honestly in the lot of a contact myself together and form a stocking of development Solutions. One after the research but then how to repurpose
them and rapidly deploy them or lab operation. So I haven't talked a lot about our that far for the last Harvest talk. I just want to talk about really are and you're his role in supporting our operation, until just for some contacts about about the volume tests are typical test volume. Four-volume laboratory was about $5,000 per month and currently we doing 5 to 9. Yesterday in the process of doing that, we had to move to a 24/7 operation. We've hired a large
number of Staff both on the analytical. And then we had the, we basically a tide that kind of adopt a mindset really going to start of this and make decisions on a day-to-day basis. So I think it is harder to make decisions on a day-to-day basis. If you can't, if you don't understand what's actually was fundamentally happening with the flow of samples to the lab and how you're delivering on turnaround times. And you know what kind of response so far has really been critical to driving our dashboarding original reporting, and
despite having a lot of expertise throughout our staff python Fargo to, for whenever we're going to ask a question, analytically is it is our For some contacts about what our infrastructure looks like. So, I'd mentioned Sunquest before. That's really not completing the center of our universe on the lab side. But it's really important for us. But this is just kind of shows you the flows of data into some of our reporting and analytics at the structure that way, throughout the Overtime, Bill
sunglasses are allowed information system. Underlying Sunquest is a cachet. I don't have these other data sources that we coordinate the series of a daily extract, kind of a traditional, you know, ETL extract transform load type of work flow into our database server. So we run postgres database and then we rent again some pretty light weight Man, line functions. Our warehouse package, it allows us to, to import your basically run some important schedule, things set up a database schema, and things like that. And then on the kind of receiving end of all that data are are
python. We have some used to have a little, but you're really are go to for all things for overtime was kind of build a workflow using Docker and get loud. Really not unlike what you can do, what you can get in place with rstudio snacks. And some of our work on this created awareness of that Lucien on, but we've kind of build a solution that is that provide multi-language support and really contain arises individual reports. And then you can supply it with a figuration that that not only includes some meditate, like name of the report, the closed on our website,
also, allowed access to stuff a bar on commissioning in our groups that are available within within some other infrastructure, at least we've felt. So this is a nice word flow of you can work with it in it. And so we used to get Robinson branches, Bad things and then when you when you merge things to the master that can trigger the automated Machinery, so that the are your report on the website is updated. And we found this, this works, this is like a perfect match with,
with flex dashboard dashboard. You can quickly, kind of spin out a flex dashboard, and then deploy it, you know, within minutes and then just kind of set it and forget it. So far we pretty, we have 20 reports. We have many more reports than that. We have 20 covid-19 4th and they hadn't been involved in the belt in a motel in general. Are turnaround time for requested is use one to two days. After I have to figure out design going back and forth and then publishing I'm going to jump out of presentation, real fast. And I
See if I can. Hopefully you can see screen start taking some select selections from our operations dashboard. Now this is not a full dashboard, has some information that we review on a daily basis terms of our operation. So you can see here that you are very we're very focused on providing great turnaround times process. You can see our overall statistics are median overall turnaround time for yesterday percentage of samples. Resulted in 24 hour, First 48 hours. And then
this is our is our overall turnaround time by date. So this is I think in addition to flexdashboard another key component is just expensive piece of plywood. Using plotly makes it so easy to take a lot and turn it into an Interactive. Nashville line of code, we use that extensively. And then you know, this is just showing 5th percentile mean. Green line, median, and 95th percentile turnaround time, this is overall picture is helpful in with more important process. We tear our
sample based off of wood, medical input, input from clients, so forth from the criticality. So we are to 1 samples. You want to try to deliver that the best around time off. Or I'm going to have these other and other categories based off of your patient outpatient, other different settings. And then we can also drill down and make sure and understand r e d and impatient around time is because these are not actually are rapid, tests are wrapped up. We can get out the door much more quickly, the Builder on site
Hospital, either for samples that are sent to the Central Neurology Laboratory. Your wee-wee review these daily and then when we see Trends across the bridge at their, we have more specific, use the database application and clients and so will look at that and then provide will make adjustments when we see that that he'll come on turnaround time fall outside of their 2-year kind of operational decisions like that. Then we have a selection of some processing data, hear that kind of show the flow of samples and different sections of our
of our health system and then going to overall samples flux to see, you know, how many we got in the door. How many or resulted on? We have other room, detail information about volumes and tears. And this is, this has been really helpful for Staffing to hear leave early on in one week, with the changes, this represents for each each of these vertical set. The panel is a day and then each of these on the x-axis is the hour of the day and we have different categories so we can take a look at when are when are large, shipments are
coming when that changes and then make sure that we're just stopping based off of the expected. Number of samples coming in for shift. So there's some more detailed drill down. Information on kind of sample by our day and required on that are similar to this, but we also among the staff. And so we made some changes and stuff from Target for improving our ability to log city of Seattle, samples, particular. And so, you know, we have staged a 401 order per minute on The Daily scouting. The average amount of time, it takes to log in the sample. And, you know,
we have a gold one point, two orders per minute for some reason changes, we're getting closer to hit. We can look at your probably Health standpoint look at population Trends. And so, luckily, the great news for washing date is that numbers are coming down across the boards. This just shows positivity rate by day for Washington State and then we have positivity rate by county. In settings, do inpatient, vs. Outpatient across our different counties, Most counties are looking are looking really good in terms of race
down to the lower Township. And then I will. Perfectly absurd, some information from this. One of the key aspects of maintaining a pulling workload, is making sure that you don't have samples come through with to high rate of positivity. You at if you're pulling for samples to a well, and Austin rate is 25%, then your chances are every single well on. Your plate is in deposit, then you're just going to have to do a bunch of extra work to reflex, all of those Wells to the
individual tests. So we use on a day-by-day, do you are a sweetheart in Spanish? You can see I've taken out locations here but can businesses or actually organize each row here is ordering location until we can base off of their positivity rate for the last week we can basically exclude or include different locations in our sample flow from going to the pool waiting for flow. then I'll just mention quickly that In addition to, to monitoring her this more operational
aspects. We also supply chain is a significant issue with all of our Laboratory Testing, and so we have a flex dashboard driven by an inventory spreadsheet process that we have stopped War, monitoring inventory, daily, making updates, and that really drives another dashboard in supply chain as an indicator that out. You're starting to run low on a platform. In that we use this information out, the shift on testing from one place to another. When needed Last I won't, I won't jump to the to the public
dashboard by the same like that for that they were looking at Dr. Publix has here in overall summary of total testing numbers on the TV, right? The website, the car wraps in Rochester. You can look at the numbers are here and then we can leave by yesterday's basic kind of collectively. All of the things that we put in place to increase the efficiency of testing, we tested, more than half a million patient and detected no more than 25,000 infections and you are continuing to scale up, our are testing really on. Based off of Tyne Daly, Adoration of the stricken by
by laboratory data largely display. My r Don't turn the wrap-up. I'll just share a kind of, a few of the Lisa. The lessons that we've learned is, are within our group. I think we have like a USB and recognition of this important role in supporting actress the testing. But we have these kind of really an integration Solutions. And so about being able to develop some open source, software, ducks and geese, and direction for Section 8 of the US and help us fill out, and we would have driven a lot of
operations position. And I think I just want to emphasize that Increase, increasing laboratory capacity for any, anyone who's working with inline medicine or Jason medicine, and that combination of increasing laboratory has to be partnering with our, with our colleagues outside, the lab and helped fuel a recession. Music knowledge, the three ends here, Noah and Nathan's and knows the director of Leventhal again for Maddox and has been critical in helping. Our group move forward on
the Tennessee, supposed to know who's been critical of identifying, just stepped in to do application development, as well as Fortune. Have a picture of Caitlin. She has been also critical to allowing us to Taylor are testing and other members from our lab. It team are out and on the top right corner there, I can't say enough great things about you. We've been able to hear we're just providing them and helping them increase their efficiency, but they're really doing all the work that you out there. I'm finally not typically we talked about
funding and support in this case, not supported by grants that supported by our great leadership in the department, dr. Rod project aired, our Sheriff has really said your weed up. I was on my way with you. I will support you doing whatever you can to testing. The Ramses has been the same things I do. You need to spend to bring in as many as you need to support this response. And then he has provided funding as well as a bomber because she made a large donation, the UW-Madison for testing specifically
and I do also want to thank all the donors. We've had lots of people who who donated funds and food and other goodies to WR ology. So there's a sword is always appreciated and I don't know we have a lot of time but I will be happy to take you questions. We have we have two minutes. Thank you so much for this amazing presentation. The first question that was posted most, I'd be up for it. It is specifically what AWS Services did you use, where they any issues with HIPAA compliance will have two more for all questions, so maybe we can fix
it quickly and I feel a lot of our time invested in a w. S was focus on the security model and how will you make sure that we plug in our authentication access management systems in a in a HIPAA compliant with both Amazon and Ric security team to make sure that we're doing everything aboveboard and I don't have any plans for a different solution that we shift. But I got my head, I don't know all of the full list of them, but we basically anything that Amazon provides that that is kind of automated, a large chunk
of the deployment. We try to do. And then just one more question, how much will what you have learned be applied to other kinds of lab tests, things to paper and hands getting to modernize your work. Those father tests over the next 5 years. Yeah. So you are our typical kind of agility a response or other things are is is off of much lower. I think I have no doubt that for for some of these Nursing Facility in order, testing situation that work will continue to heal utilize on things at the roster upload
generator orders at you. I don't know how much that's in demand out outside of the pandemic, but with your work, we're doing in parallel to a lot of these Solutions is deploying kind of the more production, traditional hl7 interface is alongside those. I'm so I think I've kind of change my opinion. I thought you know, there's no way we could go live with it. He's upset with it without an hl7 interface for a solution might solve. And we had to build an hl7 interface for that. And based off of is off of what we've done with this, I think this is actually
more valuable than something area out to get something out there. Quickly that you have dedicated staff underrated app ever. You deployed Lucien, you have to have the right Staffing level to support them if they're being used to crush your clothes. I think my opinion has changed a little bit. I think I'm planning to do more. And round software development in the longer term that I was originally thinking about firing the pandemic. So I don't know if I can't say that, you always wish it was going to be used to pay off all lab for testing. Certainly, for our PCR work, clothes were
going to deploy that automated. We're working on the point of automated data transfer Solutions across other Great. And that's all the time we have unfortunately. So thanks again. Patrick and thanks for a fantastic presentation.
Buy this talk
Interested in topic “Medicine, Health and MedTech”?
You might be interested in videos from this event
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.