Glen Ansell, Intelligent Automation Lead, and his team at Tangent Solutions worked on a project whereby they used artificial intelligence (AI) and robotic process automation (RPA) to help speed up x-ray screening for potential COVID-19 cases. In order to build an effective solution, though, Glen’s team had to make sure their solution was non-invasive, rolled-out quickly, and built robustly without room for error.
In this conversation, Glen outlines the project’s problem statement, the technical requirements, the unique challenges and constraints his team faced, and how they approached their solution given their specific context.
Check out some of the key takeaway lessons below, and follow along our conversation using the full transcript — also included below!
TL;DR: Some key takeaways and advice
On building ‘non-invasive technology’:
Make it as easy as possible for a customer to adopt your solution, irrespective of their technological ability.
“I'd say the main principle that we focused on in this project was, number one, make this technology noninvasive. That meant: If somebody is OK with being able to drag a file into a folder, keep it as simple as that. Do not go in and inundate them with new systems that they need to learn, where they have to understand the underpinnings of machine learning or tech jargon or what’s underneath the hood that does the heavy lifting. For this project, if we could make the report as simple as an Excel spreadsheet output, we did that. Without having to understand the inner workings of machine learning or robotic process automation, they should be able to fundamentally get a result without having to work outside of the water that they swim in.”
On dealing with novel problems and difficult constraints:
Disregard assumptions, and don’t build technology for technology’s sake.
“It was really important that we find and kill assumptions. I think we as technologists often try and go in like a bull in a china shop and throw technology at every problem without fully understanding: What is that problem? Even though you have the technology to solve it, it doesn't mean it will actually land within that area that you're trying to solve for.”
On building solutions in the world of artificial intelligence and machine learning:
In AI and machine learning, you need to work with high-quality, large databases.
“We fast-tracked our process, but still followed the normal processes we would do: We kicked off with a discovery session to understand the process of somebody going through X-ray screening. And then started mapping out where all the data exists — in the world of machine learning and AI, data is everything. Without it, we can't build these models. We had to make sure that we not only had sizable, concrete databases to work off of, but also that they were quality databases.”
On building tech solutions in an industry where tech is under-utilised:
Know the context you’re building solutions for: The way tech builds solutions, and the way that industry understands solutions, will likely be completely different.
We had to find the intermediate way of bridging these two worlds: Good clinical practice, and how you do clinical research, is fundamentally different to how you go and do research and build solutions within technology. And technology has to take that into account. It’s also about educating the healthcare world as to what is possible out there right now, and ensuring we connected these two worlds, and created collaborative initiatives as opposed to building technology for the sake of technology, without understanding the context in which it needs to operate.”
Podcast chapters:
[04:50] Rapidfire ‘warm up’ questions
[08:04] Introduction to Tangent Solutions: What problems do they solve?
[13:15] Defining AI and RPA in the context of this project specifically
[16:21] Project outline: The problem statement, constraints, and challenges
[20:21] The project’s technical requirements and “main goal”
[22:16] Contextualising the project’s goal within the technical constraints and challenges
[25:42] High-level project approach: Critical considerations, and step-by-step process
[32:34] Guiding principles: What does “non-invasive” tech mean?
[37:06] How unique challenges and constraints affected workflow and development
[41:00] Closing thoughts: Potential of this tech going forward
Transcript
[04:50] Jomiro: Cool, Glen, welcome to the podcast. It is such a pleasure to have you.
Glen: Thanks for having me.
[05:00] Jomiro: Before we jump into our conversation, I have prepared a set of questions which you haven't prepared for, and I'd like to rapid-fire them to you now. It is just to get some insights into who you are. The goal is to not think about them too hard. And there's also no follow up questions. So, short answers, no follow-ups. Just a little back and forth rapid-fire.
Glen: Nothing like being put on the spot. So fire away.
[05:23] Jomiro: Absolutely. First one, after a really tiring day, how do you spend your evening?
Glen: Oh, wow. That's a tough one. I spend it by working even further most of the time, following up on research, on trends in technology around the world. And in between that, try and cook dinner, get my daughter ready for school and the normal stuff, right?
[05:50] Jomiro: And what's a really good book that you've read recently?
Glen: I'm actually currently reading Bill Bryson's “The Human Body”. And I am really finding that fascinating. I loved his book “A Brief History of Everything”. Phenomenal book. He's just a brilliant author that really knows how to dumb down a whole lot of concepts for us mere mortals.
[06:16] Jomiro: Yeah, he's a brilliant mind. I must agree. And then tea or coffee?
Glen: Coffee - a lot of it.
[06:22] Jomiro: And how do you take it?
Glen: Black, one sugar, and strong.
[06:29] Jomiro: And in a sentence, how would you describe your feelings towards superstitions?
Glen: Nonsensical, that's what it’s about - nonsensical and no place in the real world.
[06:48] Jomiro: And what about luck?
Glen: This may sound pretty contradictory now because I do believe in luck. Luck has a role to play. But yeah, probably no real basis behind it.
[07:02] Jomiro: And what is something you're glad that you said “no” to recently?
Glen: Yeah, probably mass consumption of the media. I have been saying no to it a lot, but I jumped onto the bandwagon by watching statistics daily for about a week and kind of got over that, not knowing what to believe, and really just avoided it at all cost right now.
[08:04] Jomiro: Yeah, that's a really cool set of answers. Thank you so much for participating. And I think with a book recommendation, Bill Bryson, at the top of the episode, it can only only continue getting better from here. So thank you. Just [to get] some context into who you are and what Tangent Solutions does... Could you just start with some insight into how big your company or your team is? And as a business, what problems do you solve?
Glen: Sure. Tangent Solutions is a division of the bigger group Juramani. We're approximately about 600 people strong, and the organisation focuses on building digital transformation and helping organisations move towards digitally transforming themselves through the use of various technologies, predominantly leading emerging tech that you'll find in the market. Tangent, specifically in that division, focuses on custom development, helping customers move workloads into the cloud and consume the cloud better, specifically Microsoft Azure, and AWS or Amazon. In the area that I'm the practice lead-in, and we focus on helping organisations automate, automate processes – we use technologies like robotic process automation, or artificial intelligence, machine learning or just advanced statistical techniques – for lack of a better way of putting it – Internet of Things and just areas where we're helping organisations consume technology. And also democratising it you know, democratising it not only for large organisations but also for smaller ones.
[09:57] Jomiro: And your team sizes… where are you in your life cycle?
Glen: Our team is just close on 25 people at the moment, a mix of engineer and industrial teams. We've got project managers, developers, business analysts, process analysts.
[10:25] Jomiro: How does Tangent Solutions approach digital transformation and helping your customers where they use technology in various ways? What are some of the ways in which you deliver that?
Glen: So, I think it is inherent in our DNA and how we approach things to see the world as a process. Before we even look at technology, we understand what the process is in organisations, what the limiting factors in those processes are, what the strengths of some of those processes are, how they impact those organisations before we look at any form of technology. Once we've obviously mapped those out, we typically go into discovery sessions with the customers. And then if we find that there are areas of improvement, we'll go into a deep dive session with them really mapping out those processes, who the role players are, what the impacts are, what the risks are, of those processes falling short and deviations on those processes and how to mitigate those. And once we've done all of that, we would then start looking at, well, if there's potential for potentially automating – process reengineering – and start looking at the types of technologies that could facilitate improvement in those organisations.
We typically look at areas, especially in the economic times that we find ourselves right now, where automating processes or improving them – where there's going to be some demonstrably, financial or customer experience or efficiency improvement.
So the other thing about Juramani and Tangent Solutions is – we consider ourselves to be technology agnostic with an opinion. So, we don't believe that there is necessarily the same tool for every problem, and we evaluate every problem on its merits and the merits of the technology that can potentially go and solve that. When I say with an opinion, we obviously believe in certain software, in certain areas that are fit for purpose and are going to, in our experience, have worked best within organisations, that we solve problems or help digitally transform.
[12:39] Jomiro: And what about your role in terms of the things you do on a daily basis? Could you also give some context on that?
Glen: Okay, so I'm practice lead for the Intelligence Automation division within Tangent. I suppose that pretty much means my role is sales, marketing partnerships, mentorship, commercial and legal decision making, and I really wear whatever hat I need to on the day. But in general, it's around defining a strategy and aligning the individuals and making sure that we have the right resourcing to deliver on that strategy.
[13:15] Jomiro: Awesome. Diving into our main conversation for this episode, your team recently worked on a pretty big and a pretty important project using artificial intelligence and robotic process automation, so AI and RPA — all to help speed up X-ray screening in this COVID-19 global health pandemic. But before we get into that, I hear these words thrown around a lot and I'm curious to hear from you... AI, I think, is quite a catch-all phrase, and I don't know how many people have heard about robotic process automation.
So, just in terms of what AI and RPA look like, can you illustrate it with a specific example of how that has looked in a project?
Glen: Sure, so let me let me tackle RPA quickly. So, you know, robotic process automation, taking away all the technical acronyms and jargon around that, it's just really a new way of helping automate processes within an organisation that are highly repetitive. You're looking for high accuracy, and in support of reducing high volumes of processing – the kinds of things you can imagine that space is dealing with when it comes to applications – anything where there is documentation involved. And if anything, really, to try and liberate human beings to do things that we're really good at, and that computers are not good at, and go use computers to focus on these highly repetitive, high accuracy requirement volume repetitive tasks, and things that are not necessarily intuitive. And then let humans focus on intuition and experience and those kinds of tasks.
[15:19] Jomiro: Awesome. And then AI?
Glen: Yeah, so AI — as you rightfully said — is a bit of a catch-all term. There's so much that falls underneath it, you know. Our view of the world of AI is anything that gives the perception that it behaves somewhat similar to a human being. It makes decisions that one would expect possibly would have followed similar formats to the way a human being would have made decisions. I mean, the underlying technology around all of these things, such as, you know, machine learning, building deep neural networks and data science, those are actually the technologies that underpin what the AI is. AI in our world is just really how you interface with another human being in a way that seems seamless. It seems that it's constructed in such a way that we would expect a human being to interact and use the same sort of language.
[16:21] Jomiro: I really liked the way you introduced AI and RPA there – this idea of seamless interface and liberating human beings from this high process and mass amount of repetitive administrative work, because I think that links quite nicely into this problem you were trying to solve with this particular project around helping speed up X-ray screening.
Can you outline what that project was all about? Maybe in a sentence or two, tell us what was the problem there? What needed solving?
Glen: Sure. So I mean, we felt that dealing with high volumes of screening x-rays within your radiology department is time consuming, and it obviously would be increased because of the high prevalence of pneumonia and looking for other factors that influence the outcomes within COVID like the other comorbidities that exist such as TB, pneumonia and other factors.
Because of that, we felt that this influx of patients coming through with potentially COVID related features was going to put a lot of strain on the radiology department and in general medical practitioners that need to deal with testing and screening of which X-rays and hence imagery would play a major role in that.
We have extensive experience in computer vision and felt that that was an area that we could make a significant impact in – helping support health practitioners in dealing with these high volumes that were expected based on the statistics and research that you know, we found relative to this global pandemic:
[18:29] Jomiro: And just quickly, because it's a new concept to me... What is computer vision?
Glen: So, computer vision is just really the concept of utilising statistical tools and machine learning tools to analyse images in a way that for computers it is not that straightforward. They see pixels and bits and bytes, which is not the way we necessarily view imagery. So, computer vision is just that arena that allows computers to basically analyse and interpret what they're seeing in an image or on a screen or whatever they find.
[19:15] Jomiro: And is that something that is currently quite a slow process or what makes that such a slow process?
Glen: Well, the thing is, it's not necessarily that it is a slow process in normal circumstances. The challenge is that you want to catch the results as quickly as possible, and a lot of the infrastructure is already struggling relative to the number of practitioners out there — how many radiographers you have, how many radiologists, the number of radiology departments, issues such as people out in rural areas that have to travel far, and with the actual global pandemic hitting us... These volumes are just going to put that under-resourced environment under even more strain.
So, less so than in a normal circumstance is it a problem, but rather that this pandemic is precipitating the fact that that environment would, under these abnormal circumstances, be under tremendous strain and hence increased volumes.
[20:21] Jomiro: So then, given that, what was required from a technical point of view to solve this? I think it would be cool maybe to speak to what you were talking about: Getting computers to do the repetitive stuff so humans can do what they're best at. What was the goal there? What were you going to try to automate using AI and RPA? What are you trying to enable radiologists to do more of or better?
Glen: So really what we wanted to do was not to diagnose certain features found in the X-rays.
The main goal was to create a prioritisation queue that supported medical practitioners in looking at the most likely cases of positive COVID patients, so that they can move to a treatment state as quickly as possible.
And in turn specialists could go and find whether they do or don't have COVID based on those X-rays. And we felt that utilising concepts that we are very familiar with like computer vision based on machine learning models, to go and analyse these x-rays number one and number two, automating the output of those analytics as seamlessly and easily as possible, but also in context of the medical environments which which within which they would need to operate, which are often quite low tech.
[22:16] Jomiro: So what you were saying about trying to automate and make those outputs as easy to process as possible... In terms of this project as a whole, what were some of the biggest challenges to getting to that point, or solving that problem? What were some of the biggest blockers or limitations that you had to think about and really consider?
Glen: I would say, trying to make sure that this entire project took into account that we should not be using technology for the sake of technology and focusing on making it accessible to the layman.
Without having to understand the inner workings of machine learning or robotic process automation, or even have to hear any of these acronyms, [the customer] should be able to fundamentally get a result without having to work outside of the water that they swim in.
Excel is not a stretch, but machine learning models potentially are. Dragging, and dropping files into a folder is not a stretch, but understanding concepts of robotic process automation, maybe. So we wanted to make sure that we bring this high-end tech in a very low-end interface.
[23:15] Jomiro: Yeah, and because you weren't sure of the experience level of the whoever the radiologists might be, you had to really develop it for for anyone, any layman, to be able to use and benefit from.
Glen: Precisely and also to move away from… We don't do any kind of heavy integrations into the X-ray systems or any of the medical records or any of that. It needed to be really straightforward and something that we can hit the ground running and try and support the fight against a global pandemic as quickly as possible.
[23:50] Jomiro: So just out of interest, because I know you said normally sit down and brainstorm with a client... I highly doubt you were able to sort of speak to every radiologist in the country, I don't actually know how many there are, to be honest. But how did you think about who the client was, who the customer was? And how did you bring that person in?
Glen: So we had a couple of discussions with some heads of Radiology departments, in looking at the feasibility, understanding with our implementation of low tech for literally just dragging and dropping x-rays into folders was even viable, feasible at all, and more importantly, would have been practical. So we had those sorts of discussions.
We were also very fortunate that one of the software partners that we work with, specifically in robotic process automation being UiPath, had a couple of people building solutions in the space. We had also done some research as to the global importance of being able to solve problems such as prescreening and pre-vetting of x-rays within these radiology departments.
So it was a mix of research that said, obviously, because of the timing and how quickly COVID hit, it would not have taken our normal path around research and solving and finding a business case. And unprecedented times, unprecedented pandemic. So we also took a different approach to this and said, “Look, if we can solve problems that can support this fact, we're going to get involved as quickly as possible.”
[25:42] Jomiro: Given those challenges and given those unique constraints around it being during a global health pandemic, how did you go about this project? At a high level? What were some of the first steps you took from a technical perspective in terms of starting to problem solve this? Maybe you can also compare it to how you would normally go about it. What was that approach? How did you manage to start working on this and get it off the ground?
Glen: Sure. So I mean, we absolutely fast-tracked it for obvious reasons, but still followed the normal processes that we would with any other customer.
We kicked off with a discovery session that involves just really understanding the typical process of somebody going through X-ray screening for, in this case, obviously, pneumonia, but we're looking at it now as well for tuberculosis and other areas that computer vision can pick up.
And then started mapping out where the data exists.
In the world of machine learning and AI, data is everything. Without it, we can't build these models. We had to make sure that we not only had sizable, concrete databases to work off of, but also quality databases.
We tested what sort of medical departments have been using them. Where are they from? Where the images are from? And confirming that we had just a quality set of data.
From there, we obviously went into a deep dive understanding of how we bring together this process that we are looking to implement — but also in context of the technology and what we could build.
Naturally, then we went into a build state — testing, building up marketing collateral from there, and now we're starting to move into a go-to-market strategy to look at how we get this thing propagated into the various departments where in which they can obviously add value.
[27:42] Jomiro: That's interesting. I find the the discovery thing you mentioned quite cool because I were to paraphrase, it's almost trying to see, as you said, where data exists, and where you can plug-in AI and RPA into the system, and whether it's surface area for you to really work on and unhinge or unblock some of the limitations that are currently there, get a lay of the land… This is obviously not... We don't really interact with x-rays and radiology on a day to day basis, so to figure out also for yourself where things fit in and what it looks like.
Glen: Absolutely, I mean, it's really important we find and mostly kill assumptions.
I think we as technologists often try and go in like a bull in a china shop and throw technology at every problem without fully understanding: What is that problem? Even though you have the technology to solve it, doesn't mean it will actually land within that area that you're trying to solve.
It seamlessly conflicts with all the stakeholders and context of any barriers that may exist that you may not have been aware of, you know, albeit that you've got this wonderful technology.
[29:00] Jomiro: And like you said, not using technology for technology's sake, sort of being very intentional where you put it and how you use it, which is quite cool.
Glen: So, in this case, as well, being really sensitive to issues such as privacy. I mean, we're talking about people's health records, so things like anonymising the data, anonymising it while still making sure that it is accessible, and for lack of a better term de-anonymising it when a practitioner needs to obviously then action against that data.
[29:35] Jomiro: And were there other triggers for you when doing that discovery. I know you said data was one of them looking for mass amounts of data and quality of data, but were there other triggers that you noted as places where you could use AI or RPA?
For example, one that comes to mind, you can correct me if I'm wrong, but if there are touchpoints where someone needs to send something to another person, that is a potential surface area where you can use AI and RPA.
What were the other triggers for you that you used as markers?
Glen: Sure. So there were a couple of other triggers. But I think most importantly, what we found is that dealing with patients with these sorts of symptoms and these sorts of challenges, is dealt with in a very multidisciplinary kind of way. You're not just talking about a radiographer and a radiologist; you're talking about a whole lot of disciplines involved within the medical fraternity, that all needed to receive this information. So naturally, you know that that can obviously be a cumbersome process. With RPA, we will go “Hey, let's push this out to everybody that's involved that needs it immediately as we've got that kind of information.”
The other side of it really, as we were doing our research that was relatively evident, is that there is some incredible technology out there. And that can heavily influence the way in which we do screening and utilise technology to support practitioners in their efforts. But it is very underutilised.
Computer Vision is not a new concept. The fact that we're applying it to x-rays, for that matter is not even that new. But it just doesn't seem to have propagated into the healthcare sector.
And we feel that over and above dealing with the COVID pandemic. And while we initially built this project, there is scope to be able to utilise this sort of technology in a myriad of areas within the health sector to support various different problems, and I mean I've touched on one being tuberculosis as the next project that we're looking at right now. For screening, we're especially considering in many rural areas within the context of South Africa, availability to facilities is not readily available. Even practitioners and I mentioned there's multidisciplinary practitioners that are looking at these kinds of symptoms, may not be available on a continual basis as you would find in some of the higher private healthcare environments. And being able to move information around rapidly, analyse it and support these practitioners in a digital format can dramatically improve on their efficiencies and support a lack of resources in the healthcare space.
[32:34] Jomiro: So, in retrospect, what were some of the main principles or tactics or strategies that helped you succeed with this project?
Glen: So...
I'd say the main principle that we of focused on this project was, number one, make this technology noninvasive.
[32:54] Jomiro: And what does that mean?
Glen: What that meant was: If somebody is able to drag a file into a folder, keep it as simple as that. Do not go in and inundate them with new systems that they need to learn, and understand the underpinnings of machine learning or tech jargon or what’s underneath the hood that does the heavy lifting. If we could make the report as simple as an Excel spreadsheet output, we did that. So sort of two areas you've got high-end heavy lifting technology, doing a whole lot of stuff, while making sure that we provide a very simplistic noninvasive interface for practitioners to utilise.
If somebody is able to drag a file into a folder, keep it as simple as that. Do not go in and inundate them with new systems that they need to learn… If we could make the report as simple as an Excel spreadsheet output, we did that.
[33:43] Jomiro: I don't think we've actually mentioned it yet, but with that principle in mind, could you outline the high-level process or solution? What are the touchpoints? Drag-and-drop as one of the elements, and Excel sheet is another. But what is the high-level start to finish view of that?
Glen: Sure. So I mean, underneath the hood, you've obviously got machine learning that will go and take an image, process that image and evaluate with the confidence level of whether that image shows signs of pneumonia or not. In the same breath, once it's gone and processed that there is a robot that goes and takes that data opens up a spreadsheet, inputs the data within that spreadsheet and makes sure that it updates any individuals that need to know whether a specific document, in this case, an X-ray image has or has not been processed. From the radiographer's perspective, they literally go and drag the image into a folder, and the robot takes over from there.
[34:54] Jomiro: And then they get the results outputted to the Excel spreadsheet.
Glen: Precisely.
[34:59] Jomiro: So that... I mean, that's really cool. As you said that emphasises the noninvasiveness — where the radiographer need not understand anything about the technology behind it, other than if you put this file into this folder, you will see the information you need on the spreadsheet. And they can trust that and it runs as is and when they need it.
Glen: Exactly. Yep.
[35:26] Jomiro: Awesome. And what were the other other tactics or strategies or principles that helped you succeed or win?
Glen: So as I've mentioned, our major sort of principle and how we were delivering this... I mean, medicine is a low-tech environment in many ways. And we obviously needed to deliver a solution that would seamlessly free up time relative to these increased volumes that they would be seeing.
Strategically, we needed to align our solutions to global trends and forecasts that we were seeing. There was obviously going to be a lead-time to go and build this, and we would hate to have built something that would be useless relative to a non-event or a pandemic that no longer exists by the time we had completed the solution.
So we were putting a lot of effort into looking at the trends, looking at the movement of data, looking at what one can expect going forward relative to volumes, and always aligning the project to this on a continual basis — which is quite a different perspective to work from. It's quite a different way in which we work, where you typically have a customer with a business problem, and you have a time frame, and you know that in their time frame, it is still going to dramatically impact their organisation.
> This is a moving target on a continual basis. It is unprecedented. We don't know what we don't know. Always making sure that we're staying in line with what we're with those global trends was was quite an important strategic goal that we had with within the project.
[37:06] Jomiro: And how did that difference impact your ways of working? Did you have to have someone assigned to continually check trends? Did you have to have regular status updates? Did you have to increase the number of meetings? How did that that difference actually look on the ground?
Glen: I would say all of the above.
I think the project took on a different form. It was incredibly Agile for obvious reasons. We needed to get this out as quickly as possible, but in the same breath be as robust as possible.
We are not dealing with business data per se — not that that is any less important — but we are dealing with human lives. We needed to be very sensitive to that and ensure that we could build something quickly with incredibly high levels of accuracy. It really cannot falter.
So naturally, we had people researching what is happening on the ground with COVID, looking at the stats and making sure that we would still be relevant in bringing a solution like this to market while in the same breath, a lot of research going into how to ensure the accuracy, quality of outputs from from solution.
So, yes, there were stand-ups on a continual basis to go and build the solution out, and a lot of individuals involved in researching areas of COVID relative to the project, but they would not necessarily or directly be involved with the delivery of the tech on the project. So it posed quite a unique, challenging way in which we needed to go and deliver on the solution.
[38:43] Jomiro: And do you have any advice for that? I can think you know, if for example a piece of data came in from the research being done, that was quite a dramatic shift in trend, to just sort of derail your roadmap and jump onto that is not a great productivity hack, by any measure... But also, you need to make sure that the trends coming in are not just a once-off occurrence.
So, do you have any advice for how to stay up to date with this kind of a moving target like this, but to take it in in a very sane and collected way?
Glen: Sure, I suppose... Difficult question. But the easiest way for me to answer that is it's really about attitude, and how you approach these kinds of things. I think the bottom line is take nothing for granted.
Assume nothing, and know that you're going to need to roll with the punches. Things will shift — deal with them.
It's the attitude of the team, and how you approach these kinds of things. There's a lot of luxuries on other projects that might exist, that you don't necessarily have over here. But nobody asked for this pandemic to hit, and nobody expected it. So you roll with the punches and take it as it comes.
[40:07] Jomiro: I think that is great advice for life in general; it is also a very good point. And what has been one of the most surprising things about this project for you from a technical perspective.
Glen: So I'll probably be repeating myself but it's mostly around how much tech is available and not only in the computer vision space but for healthcare in general. That the medical fraternity just doesn't seem to have adopted at the sort of pace at which it is being developed. So I definitely feel that there is a bridge to be crossed and a connection to be made between technologists and what's being built out there and healthcare practitioners and finding a good medium of how we consume these things but in context of what health practitioners expect from us, as technologists.
[41:00] Jomiro: And you're alluding to the potential that even just a project like this, the groundwork that you've laid, the potential that it can have for medical practitioners across the board… I think you said that TB was the next area of focus for you guys. Correct?
Glen: Yeah. So that that is the next area of focus for us. But I mean, if you go and do the research, and you're going to look at what computer vision is doing in general, and for that kind of machine learning, looking at any kind of markers that can help indicate areas of disease. Areas, you know, symptomatic type evaluation of any kind of disease that may exist within a patient. There is so much out there, and I mean, we're just scraping you know, the top of the surface at the moment.
So I'm sure it would be a very, very long list where to go and take you through all the tech that we see out there. But in general, I will say machine learning, being a very strong proponent and leader right now in helping medical practitioners... or potentially, rather, helping medical practitioners evaluate problems and symptoms rapidly and, in some cases, you know, more effectively.
[42:27] Jomiro: What do you think is the biggest thing holding the medical industry back from embracing tech like this? Is it just an unknown thingm or what's the… In your opinion and experience from having worked within that space now, what are some of the biggest limiting factors for tech projects like this to proliferate?
Glen: I would say probably finding the intermediate way of bridging these two worlds:
Good clinical practice, and how you do clinical research, is fundamentally different to how you go and do research and build solutions within technology. And technology has to take that into account.
In the same breath, I would say it is also about education. It is educating the medical and healthcare world as to what is possible out there right now, and ensuring that you kind of connect these two worlds and create collaborative initiatives as opposed to as I said, you know, technologists building technology for the sake of technology, without understanding the context in which it needs to operate. Which is a very different world in the world of healthcare? So more collaboration, more communication, a stronger understanding by technologists of what good clinical practice, good clinical research looks like, and what does that process look like so that it can conform to that.
[44:07] Jomiro: Yeah, I'm really interested to see that, because projects like this do put a spotlight on the potential and illuminate some of the ways in which people can really make a difference using — whether it's AI or RPA or computer vision, whatever the tech is — these projects can show people that there is a lot of work and a lot of potential, a lot of opportunity.
I think that this project, as you get ready to go to market, will be quite interesting to see and watch bloom/blossom. I don't know if that's the right word. But if people want to get involved or if they want to follow up with this project, where can they find you, find Tangent, find the project online?
Glen: So absolutely, they can go and have a look at https://automation.tangentsolutions.co.za/ and coupled with a couple of other projects they will find the x-ray solution there.
They could happily contact me on glenn@tangentsolutions.co.za if they would like to... We would welcome any collaboration with anybody interested in working in this space.
[45:32] Jomiro: Thanks so much for chatting with me on the podcast and yeah, telling me about this project. I think it's fascinating and I wish more people could jump on opportunities to find limitations in the existing processes and systems that are out there, whether it is in the medical industry or not. Really excited to see this project take off, and hopefully stay in touch with you in any case,and chat and check in again in a couple of months' time.
Glen: Thank you. Thanks for your time. And thanks for inviting me through to the podcast. It's a really exciting project for us and always great to chat about it.
[46:15] Jomiro: Awesome. Thanks so much for chatting.
Glen: Great. Thanks, Jomiro. Bye.