Learn Rust, train doctors – Interview with Caroline Morton

Introduction

[00:00:00] Tim: When you’re new to working in an emergency room, it’s hard to avoid getting overwhelmed by the cognitive overload of a continual stream of patient after patient. Device after device and beep after beep..

This episode follows Dr. Carolyn Morton’s career from medical doctor to software engineer.

It’s a fascinating story that involves learning to code, learning Rust, starting a company, starting a new career, and having the ambition and determination to know that it’s the right thing to do.

​

Yellowbird Consulting’s product Clinical Metrics

[00:00:34] Caroline Morton: Uh, yeah, brilliant. Um, so I am running a company. It’s called Yellowbird Consulting, but it’s soon to be Clinical Metrics. It’s a medical simulation platform, um, and the back end is built in Rust.

What it essentially is, is a virtual emergency room.

So, um, I’ve. Well, we’ve created a way to make an unlimited amount of virtual patients that are based on the underlying epidemiology features of-

We’ve got currently 52 diseases in it and patients arrive. You start with three, and then they just keep on arriving, much like real life, except you’d never start with just three.

And you have to take a history from a chatbot.

You can order any investigation, that you might want to in a real emergency room.

You can order blood tests, you have to wait for them to come back. Writing notes.

The aim of the product is to improve clinical reasoning, but it’s also to test the cognitive load and to really train people about that cognitive load, which I found in my experience is just not well taught or, um, it’s not even taught.

It’s like experienced as a junior doctor or as a, as a medical student in particular, where it’s not just one patient at a time.

You have to, you have to think, Oh, I’ve got a, this patient, they’re here. I’ve got to remember to chase the bloods for this other person.

Or this person I’ve sent for an x-ray.

And you’ve got to be able to keep it all in your mind and develop systems that you can keep track of everything.

And to be honest, that’s, what’s really challenging when you’re a junior doctor is not necessarily that the content is so crazy difficult. It’s the cognitive load of trying to remember all of the things you must do.

[00:02:20] Tim: Do you also integrate aspects like, Oh, the Charge Nurse is like demanding something, or the human dynamics of essentially being in ED and being distracted, or like someone,

Like, my experience as a patient. So, um, I happened to spend several hours at our ED with my girls. My, um, uh, my six year old is asthmatic and, um, and I’ve also come off a mountain bike a few times.

And so I. I’m aware of the kind of when you’re stuck in the waiting room, you spend a lot of time observing the practice, or at least I do. And, um, yeah, and I have both sympathy and extremely high regard for, uh, doctors in an emergency room because, um, Essentially [it] is just a large series of difficult choices.

At least that’s what it seems to me. Is that accurate?

[00:03:21] Caroline Morton: I don’t, I wouldn’t say necessarily that they’re difficult choices because a lot of it is protocolized, so, you know, there is…

[00:03:28] Tim: No, right, right, right.

[00:03:29] Caroline Morton: That you do, but it’s, you know, I think it’s the volume. There’s definitely a volume problem. You’re also managing not only people’s like physical health complaints, you’re also trying to like manage their emotions of they’re frustrated, or they’re drunk, or you know, there’s all sorts of problems.

And also…

[00:03:49] Tim: They’re sore and tired and…

[00:03:52] Caroline Morton: Yeah.

[00:03:53] Tim: Totally it’s, um, it’s, and would you imagine a junior doctor essentially playing? Their free time Is that the,

[00:04:08] Caroline Morton: Yeah.

[00:04:09] Tim: this essentially a game like

[00:04:12] Caroline Morton: Yeah, it’s a gamification.

[00:04:12] Tim: Sim for example? Okay.

[00:04:15] Caroline Morton: It’s a gamification.

So we’re in two medical schools in Germany at the moment, hoping to expand to the UK, maybe the US market.

I’ve just taken the leap to go full time on the company, having run it part time for four years, with, two other co founders, including Maxwell Fliton, who you know, he’s quite big in the Rust community.

Um, and we, yeah, so we just, we’ve taken the leap, uh, going to try and move it into the UK market.

But in Germany, the way that it’s, it’s used is it’s embedded within the curriculum.

So, um, they have a slightly different system to, I know the New Zealand system is quite similar to UK, but in, Germany, they seem to have these very quite long terms where you might do cardiology and you have teaching at the same time. You might have cardiology lectures at the same time that you’re also going to the cardiology ward So it’s in a way more

[00:05:08] Tim: So that, that is, if you’re going through becoming an emergency specialist, that pathway is quite different, is that what you’re saying?

[00:05:21] Caroline Morton: Yeah, it’s really different in Germany.

At least this is my understanding. So, with the medical schools, they might have, the way that it’s used right now is, or our tool, Clinical Metrics, is that they, they do it every week. And, um, over like, say, six weeks or eight weeks, and at the same time they’re receiving lectures on, you know, various heart conditions, say if they’re doing cardiology, so they might see lots of cardiology patients within our simulation.

Um, but they don’t, at least this is my understanding, they don’t really have emergency medicine as its own specialty in the same way that we do. We have it in the UK. It’s like you’re, you’re a, I know you’re a respiratory doctor and, but you work in the emergency room. So it is, there’s

slight variations.

[00:06:12] Tim: Yeah. And I, I, I assume-

there are both cultural and, uh, practical, sorry, like the, the pathways must be quite different for how we train doctors, you know, the must be quite different all across the world, or I guess there are probably similarities, but yeah, I, there’s also, I happen to know this because I, did pre hospital emergency care, the essentially the start of ambulance training a while ago.

And, uh, the thing that was really interested me was that on the continent, there is much more onsite care, uh, if you’re very, very serious, if you’re very hurt. Uh, whereas the other philosophy is that you basically scoop and run and you try to get to the hospital as quickly as possible.

Like especially for, uh, let’s say trauma patients, you know, there’s kind of the big debate whether or not you kind of patch. on site or you kind of pick them up and get them to the hospital, get them to ED. Um, anyway, it’s probably,

[00:07:16] Caroline Morton: Stay and play, isn’t it? Stay and play or run like scoop and run.

[00:07:21] Tim: Yeah, yeah, I don’t actually remember any of the terms anymore.

I should, um, but at one point, yeah, I was, I was, uh, an aspirational, uh, ambulance volunteer, but, um, the Uh, however, so I’m, I’m kind of inherently kind of interested in this,

Why Rust?

[00:07:41] Tim: But, um, the thing that I’m curious about, like more from, I guess, the Rust perspective is why would you pick kind of a crazy, difficult programming language to implement something which could be probably implemented in something that is much more conventional?

Mm

[00:07:57] Caroline Morton: Oh, well, it’s gone through various iterations. So it actually started initially with a Python back end. Um, and a, what did we, I think we had a React app to start with. So Python back, uh, Python microservices back end and a, uh, React app.

Um, which, you know, that was like the first, first year. Um, I wouldn’t say we didn’t know what we were doing, but we were, you know, kind of, we didn’t really realize it was going to become the business that it has become, if that makes sense.

I always say that we sort of accidentally set up a company.

It was very much a side project.

Me and Max were really into Python at the time, um, and, yeah, and so, and then, you know, two things happened at the same time.

One, Max, who’s the other business partner, um, he was getting really into Rust.

He was writing this Rust and web programming.

I was quite reluctant initially to learn Rust because I just didn’t feel that I had a hugely solid grasp on Python and I was also like, a junior doctor.

So it was, I was like trying to manage various different, I was trying to finish my postgraduate exams and COVID was also happening.

Um, so it was, it was kind of a busy time, but essentially we, the sort of final thing which happened was we went into production.

We have this chatbot where you have, you know, up to three patients, per student, and you might have, say, 120 students in a class.

What happens at the start of the class is they all log in at the same time.

So you’ve got this five minute window where the server’s getting absolutely smashed. And, uh, basically it fell over in a pretty major way. The first time we used it, um,

[00:09:43] Tim: Right, and sort of quote in anger or what have you.

[00:09:46] Caroline Morton: Well, it just, it just, well, we sort of initially didn’t know what was going. It just like the sort of user experience was, it got really slow, but luckily we had lots of metrics on the backend.

We’ve got this amazing DevOps engineer, um, who helps us, Harry. And, um, so we’re able to pinpoint it was coming from the chatbot and essentially we just started ripping stuff out of the Python backend and turning them into Rust until we just had the skeleton of the sort of Python. It was more like an orchestration server in the end.

It just like sent. Stuff out to various Rust microservices, and then we’ve just switched over to Rust, completely.

I’m not sure I’ll ever write a sort of big project in Python again. Um, I still love it for various things, you know, like more fancy scripting. But, yeah, Rust is yeah, it’s amazing.

So that, that, that’s really our origin story. It wasn’t, you know, the sexiest of

[00:10:41] Tim: No, no, I assumed, like,

[00:10:43] Caroline Morton: sort of just need to move to

[00:10:44] Tim: It is interesting to hear that essentially you kind of , We’re forced to choose something, you know, like essentially the application had reached its limits and whether or not it was, you know, there’d be people screaming at us, um, through the headphones saying, Oh, it was the suit.

It was the architecture. Da da da da. Maybe it was this, it was that, you know, nothing should fall over with 120 people, da da da da da, whatever. Ignore them because in practice, what you decided to do is say, look, we’ve hit the limits. Now we’re essentially forced to try something else. And I think it’s really nice to hear that Rust and, uh, it’s community hopefully was accessible enough to enable you to.

Kind of get going, even as quote, not a, let’s say expert Python developer or what have you, because by the way, you’ve got this entire other career that very soon within 24 months, you’re kind of waving goodbye to, at least that’s my perception of what happens next. Is that correct?

Kubernetes to the rescue? Until you need to pay for it

[00:11:56] Caroline Morton: So just with the, the 120 users, I just, just to go back to that, um, we, so we did have a, we, it was all running on a Kubernetes cluster. So we, our initial sort of patching problem or patch was that we massively scaled. The servers. So when we knew we were going to have a class, we just, you know, went up from say like three nodes to like eight nodes, spun up loads of pods.

And, uh, I had to learn how to do that, which was really interesting. So like essentially like really scaling it, which did. alleviate a lot of the pain. The problem is if you’re a tiny like startup is paying for it.

[00:12:35] Tim: Yeah, it’s awful.

[00:12:36] Caroline Morton: yeah, so it was like, yeah, this, this might solve our problem. And we did some stuff where we, we made the chatbots overnight.

We like cached them in, in what do we, I think Redis, can’t remember, but we cached them. And then we, so we have sort of various different solutions. , but, um, ultimately, it becomes a financial choice because you are, you’re like, yeah, this would work. But equally we’re like, we don’t wanna run out of money.

And so just these chatbots, they’re presumably, they are doing a lot of work.

[00:13:11] Tim: Are they, so is this sort of large, large language models? Are they, do they have personalities? Are they. Sort of trying to mimic being an individual with a condition who can’t really describe exactly what’s going on. They just know that they’re having difficulty breathing or what have you.

[00:13:30] Caroline Morton: Yeah, so that’s, um, so what’s been really interesting for us is like how ChatGPT has sort of forced us to innovate in many ways.

I had a prototype of the chat bot probably back in 2018, which was very much a command line tool in Python, you could ask a question and it would give you some response. Um, which wasn’t very good and that was just using a Python library, open source library.

We initially were using that, and, um, it was, it wasn’t like brilliant, I would say. Uh, we tried sorts of different ways of making it better, but the main thing, people loved the chatbot.

Um, and this is obviously before ChatGPT. We didn’t really, we had no idea that was coming, gonna, But people were like, oh, this is amazing.

Even if it wasn’t very good and they were very tolerant of the fact it wasn’t very good because they got to speak to this patient. Um, then we’ve moved to, yeah, we moved to sentence embedding. So basically, I was a bit unhappy with the Python chatbot. I took apart the open source library to find out how it worked.

Learning Rust

[00:14:36] Caroline Morton: Um, and, And what was amazing to me was that it had ever worked at all, um, once I actually got into like the nitty gritty of how it was, how it actually did, um, like responded. So I rewrote it in Rust actually, that was one of the early things we did.

I know it sounds a bit insane, but that was like one of my early Rust projects was rewriting it.

Um, and then we eventually went to sentence embeddings.

[00:15:01] Tim: I have a very small, uh, point to make about, or at least people often ask the best way to learn Rust. And I usually say the best thing to do is to write something that you have a good understanding of the domain, like essentially rewrite something smallish, not trivial, but I think it’s a really good approach because like it’s very health, because you don’t want to be learning a new thing.

Like I need to write a chat bot. Like how do I do that? And learn Rust at the same time. Like we really, like you were talking about cognitive overload, kind of need to retain a little bit of working memory in order to continue to move forward.

[00:15:49] Caroline Morton: Yeah, I totally agree with that. I think it’s so important, uh, to not, you want to get that sweet spot, where it’s difficult enough that it challenges you and is interesting, but it’s not so overwhelming that you just want to give up.

I think, you know, the major advantage that, I had, um, well, I had two huge advantages.

One, I had Max. And so, you know, if you’ve got a buddy, I’d highly recommend you always have a buddy who is better than you at whatever you’re trying to learn, because you can just ask them, how would I do this?

Or just watch them code something up. And you just get a lot of understanding about different design patterns and different approaches to how they break down a problem.

That was the sort of. Huge advantage I had.

The second advantage though was, which I do not recommend to other people, is we had customers. So it wasn’t, you know, it wasn’t like, oh, this is a toy project. I’ve, oh, it’s got a bit hard. I’ll just give up and maybe start again with a new repo. It was like, oh, we’ve got customers.

So,

[00:16:54] Tim: And actually they’ve got needs.

[00:16:57] Caroline Morton: They’ve got needs and it’s 2 a. m. Um, I’ve this has got to be written. So, um, that’s just how it is and that really accelerated my learning curve, but it was I would say a slightly painful experience at the time

[00:17:11] Tim: As a doctor, do you have any tips for, uh, essentially long extended periods of work? Uh, this seems to be something that’s drilled into everyone that junior doctors have to work. Lots of hours and do lots and lots, essentially many, many rounds. I don’t know what the metric is, like what the unit of time, I suppose it shifts.

Um, but it does seem to me that, I don’t know if there is any way that there’s this kind of problem of like grinding people, like grinding the soul out of them. You

[00:17:48] Caroline Morton: Yeah, I think I do agree I saw I can see it from both sides and I think you know, when I was like in F1, which is like the first year you’re a doctor, uh, Foundation Year One, um, it was just a miserable experience to be totally honest with you. Um,

[00:18:07] Tim: Yeah,

[00:18:08] Caroline Morton: You do is sleep and work pretty much. Um, but then, you know, you, you also need a system which can output, you know, you know, it’s a production line that people forget this.

Like you want output thousands of doctors who are qualified in being GPs or being cardiologists or surgeons. And that production time needs to be as short as possible. for it to be effective and essentially you can’t, you know, if it takes seven years, or say it would take five years to be a GP, um, you know, unless you want that to be 10 years, you have to really,

[00:18:51] Tim: Do the hours, right?

[00:18:52] Caroline Morton: Have to do the hours.

There is a volume thing, again, where you just need to see, like, I’ve seen thousands of patients and that’s just, you know,

[00:19:01] Tim: Yeah. I, so there’s a couple of things I should pause there just because there’ll be some people that don’t understand the terminology.

So in the British system, which I’ve, I guess I’ve inherited the big GP stands for general practitioner. So this is essentially a family doctor would, I

[00:19:15] Caroline Morton: that’s what I

[00:19:16] Tim: be a term. Um, that’s quite a difficult specialty. As I understand it, is that correct?

[00:19:25] Caroline Morton: Yeah,

[00:19:25] Tim: I don’t know if there’s any such as a

[00:19:27] Caroline Morton: yeah,

[00:19:28] Tim: Because I,

[00:19:29] Caroline Morton: think so, it’s one of those,

[00:19:32] Tim: mean, I was going to say, I guess It’s all hard, but I think if you’re dealing with people and we call them primary health organization. So if like, if you’re in a practice talking to patients who, you know, everything is different, there’s no continuity.

You, some of them you have just met for the first time, others you have dealt with for the last several years and they trust you and you’ve got to do all of that within a 15 minute consultation. Um, strikes me as particularly challenging. ,

[00:20:10] Caroline Morton: It’s one of those things where it’s really, if you ask lots of doctors who are not, who are not GPs, um, this is kind of, I mean, hopefully it’s got a bit better now, but it used to be like, oh, they’re just a GP, it’s like, you’re not smart enough to do, you know, one of the proper professions, like anaesthetics

[00:20:27] Tim: yeah, yeah, you’re not in a hospital, right? You’re a community, community practice, right? It’s like, it’s like a serious, serious medicine. Um,

[00:20:36] Caroline Morton: but it’s

[00:20:37] Tim: could even be worse. It could be dentistry.

[00:20:40] Caroline Morton: Yeah. Although, you know, you’d be rolling in the cache. So, um,

[00:20:45] Tim: That’s true.

[00:20:45] Caroline Morton: praise and

[00:20:46] Tim: I don’t know, but the distinction between, uh, medicine and dentistry is the weirdest thing, by the way, I just find that the most bizarre historical quirk. ,

[00:20:57] Caroline Morton: We shared a medical school with, or we shared a physical building with the dental school, uh, and. You, I think I met, I was there six years, I met one dentist, like it, they sort of, you, you just never saw them and they presumably never saw us. It was just a very strange sort of, like, I know they’re here somewhere, but who knows where.

Um, so

[00:21:21] Tim: Yeah, well, um, the other term that I wanted to just to clarify was junior doctor, because it sounds like something, it’s like, oh, you must be 19, but no, no, no, no, no, the term junior doctor is reserved for someone who’s gone all the way through medical school, gone then through specialist training and is then going through the process of essentially becoming the next step after that, I think is consultant, correct?

Or like senior doctor, then consultant or

[00:21:50] Caroline Morton: no. So you’re

[00:21:51] Tim: so, but

[00:21:52] Caroline Morton: you’re a junior doctor from the day you graduate medical school until you become either a fully qualified GP. or a fully qualified consultant, which is a bit of an

[00:22:05] Tim: that would be the difference in the hospital.

[00:22:06] Caroline Morton: Yeah. Yeah. So,

Why choose to become a software engineer?

[00:22:08] Tim: Um, yeah. Okay. So we’ve got, we’ve gone through many arcane details of like, so this is actually one of the things I think I think that I was mostly curious about when I, um, when I was thinking about this interview, which is that you’ve devoted a huge amount of your life to the profession.

And, if this thing that you’re building works, then it could maybe change Outcomes for patients worldwide, let’s say, just, just, just steer the needle ever so slightly towards like, you know, whereas as an individual, you’re sort of limited to those Just this quote, small number of thousands of patients.

Um, I don’t know if that was, uh, was, was, was that the kind of calculus that you were going through?

[00:23:12] Caroline Morton: Yeah. So I would say so it’s definitely played into it. The simulation program came from something that’s just used to, basically just used to annoy me so much whenever I did simulation training, because you would typically go into a room and there would be a simulator and you would say, you know, the, the simulator would say, or the simulator patient would say, I’ve got chest pain, and then you would say, Oh, I’d order this, an ECG and a troponin, which is a blood test, and then you would be handed those results immediately.

It’s like, this is not realistic. In real, like, in real life, my bleep would be going off, which is like the little pager, which we still use. And, um, all of, you know, and, but you’d also have like 30 patients and you’d have, you know, ward rounds to compare, like, and handover. And so, really, it very much started as a pet project of mine because I was like, this is so annoying.

It’s like, I’ve got to solve this problem. Um, and then, you know, it’s slightly gotten a bit out of hand and now it’s my full time job.

Um, but yeah, the idea is to try and get, um, People to take this training and then they can also have some way of standardizing the sort of, like, it’s not trying to replace like going into hospitals, talking to patients.

That’s obviously a key part of training, like still training. It should still be an apprenticeship model in that way. But, um, you know, you can’t guarantee when you’re doing your, uh, I don’t know, neurology block that you’re definitely like in that two week period, you’re definitely going to see somebody who’s got status or which is an epileptic, you know, you know, unresponsive basically to typical things.

Anyway, um, the point is, um, you can’t guarantee that you’re going to see a patient with that condition in your two weeks that you’ll happen to be there. Especially the serious ones, because they should be stopped or, you know, like stroke, um, patients get taken straight into the cath lab now, or they get straight into like, and have the clot removed.

That might completely remove the medical student from, from even seeing that. And so, um,

[00:25:27] Tim: right. The, so the, so, so what you’re

[00:25:29] Caroline Morton: It’s like trying to standardize it. Yeah.

[00:25:32] Tim: is through some sort of the, the apprenticeship model. Ideally, we would like to expose students to as many patients as possible. However, there are some of them who are so severe. So a very, very sick that they really need help right now. Which actually means that they basically get, they bypass, these, these critical cases essentially can’t be trained on, which is-

[00:25:58] Caroline Morton: Yeah, they bypass. And also you might, they just, random chance, you know, they just might not, you know, have turned up. I did a week of ENT in my whole medical, like medical school training. I obviously didn’t see the vast majority of ENT conditions, uh, during that time. Um, and you could say that about lots of specialties.

So I think, you know, you want to be able to standardize the sort of, you know, learning opportunities that people have had and this is part of what we’re trying to do, if that makes sense.

Ethics of educational technology

[00:26:32] Tim: I think so. I think that there is a difficulty with kind of edutech in the sense of it can be. Because it’s so good at being instrumented, it can kind of be weaponized against you in a little bit, at least. There’s two sides of the coin. If I wanted to learn, if I wanted to do extra work, I’d go on to Khan Academy, for example, and I would like learn mathematics and I would kind of push myself ahead, or I could go via classroom.

All of these EduTech platforms essentially provide very specific metrics about individual performance to educators. And I sort of have this personal wince about the idea that you’re actually exposing people to, uh, I don’t know, humiliation, although I guess in the space of like medicine, learn very quickly, presumably that everything is observable. Essentially, like all of your decisions are, that’s why you’re copiously writing notes because there are reviews and essentially everything that you decide to do, uh, you know, every diagnostic tool that you asked for, or, you know, did you ask for too many?

Were you, uh, all of your clinical interventions, when were they discharged? All of those decisions get checked. And maybe, so maybe it’s not appropriate. I don’t know. I just have this kind of personal, uh,

[00:28:15] Caroline Morton: point. I, uh, so, I mean, one, I, something I thought a lot about, um, especially because you don’t want to discourage people from, um, you know, you don’t want to create an environment in which the the, say, if we, you know, quote unquote, the poor performers are, you know, worried about being punished in some way, so they stop engaging with it.

So the separation between the top performers and the poor performers just increases. Uh, that’s definitely not what we want. So it’s something we thought a lot about the other, um, You know, we work very closely with the universities in Germany, and Germany actually has some quite interesting sort of data protection laws.

So one of the caveats was you couldn’t pick up individual people and say, you know, um, you know, Sharon is doing really, yeah, is doing really badly at this. And so what the teachers get is a broad summary view of how their students are performing per disease. And so, um, one of the things, so for example, we, I’m going to talk about cardiology because that’s just recently what we just did, so the, like, You might have a cardiology block, and you know, it’s very tempting when you’re a teacher, or even you’re a junior doctor, junior doctors do a lot of the teaching of medical students, like, okay, cardiology, the most important thing to think about is, like, myocardial infarction, which is like, you know, you’ve got a blockage in your heart, you get a sort of heart attack.

Um, And,

[00:29:47] Tim: feel like just pausing there and just saying that myocardial infarction is because so the myocardium is a part of a part of the heart, which, um, and infarction is, yeah, like a, yeah. Is it Greek or Latin for stoppage? I’m not sure.

[00:30:03] Caroline Morton: oh, I don’t know. It’s bad. You haven’t got enough blood supply going

[00:30:07] Tim: Yeah, that’s right. Yeah. Essentially like life is getting, um, very short for you.

[00:30:12] Caroline Morton: and it’s painful. Um, and so that’s, you know, one of the big kind of ticket items in terms of cardiology. You always want to make sure that you are, you know, capturing that and, you’re not missing any, but it’s also so typical that the medical students know it really well.

They know chest pain. They know it’s, that’s the thing you need to look out for. And what they’re less good at is dealing with, um, and I think genuinely they do receive less teaching on is things like heart failure, which, you know, kills hundreds of people, thousands of people a year. Um, some of the arrhythmias, which are perhaps less common, but maybe do present.

in the emergency rooms in slightly atypical ways. And so

[00:30:53] Tim: Oh, so there’s another point that we should possibly raise for a generalist audience is that the, this flatlining kind of cardiogram is, is not what happens in real life. If someone’s, if someone’s heart is distressed, the, the rhythm that the cardiogram has will look very odd. Um, and there are different patterns which require different interventions.

[00:31:18] Caroline Morton: Thank you, that’s useful. Um, I, so I, what teachers can do is that they can then look at, say, I’ve included six diseases in this class. Um, and even though all of the students have played, they play the role of the doctor and they’ve seen different patients with these different conditions because they’re all unique patients.

They can say, oh, broadly speaking, my patient, my students are not good at, um, you know, treating heart failure. And so I’m going to give some specific teaching about heart failure because that’s an area that they don’t know as well, or they’re not picking up as well. And we often see when we look at these, maybe you have six diseases within a certain class, we often see that they’re doing really well with myocardial infarction that we talked about, um, but they’re doing very poorly with things like heart failure.

Or, alternatively, they get the diagnosis right, um, of the, you know, terrible chest pain that you’re having, but they don’t give any pain relief.

That’s a very common thing that we see. And that’s very true to life. You do see that in, you know, you get people in the emergency room who, you know, they get the right diagnosis, but they don’t then end up having any pain relief, which is the thing that they care about, is, you know, to get rid of this terrible pain.

Um, and so we are picking up those things. Uh, we are picking up those things and. The fact that it does reflect like what I’ve seen in real life and what other people have seen is, is reassuring.

Um, so yeah, so, but back to the question, which was, you know, are we picking out individual students? We’re not, uh, they can see their own performance, but teachers can’t see anything other than the sort of metric at a class level.

Expanding the scope of the simulator

[00:33:10] Tim: Ostensibly, this is a Rust discussion, but this is fascinating to me. So I have one other thing about planning this. Do you ever have a very, like, I don’t know what the, uh, I always try, I always get stuck on the pronunciation of this word. Is it pedagogy, pedagogy?

[00:33:32] Caroline Morton: Oh yeah, yeah, it’s not, we don’t really use it as commonly in the UK to be honest, but other people use it.

[00:33:38] Tim: Okay, um, right. So if from like an instructional design point of view, if you were to just give a boring day in the simulation, like, is that something that’s of like. Worth? Do you ever just kind of not have really tricky, odd kind of test cases in there? Or is that kind of silly?

[00:34:04] Caroline Morton: No, I think it’s a really good idea. I think, so we don’t necessarily have boring patients in the sense they all do have something wrong with them. Um, I have

[00:34:15] Tim: You don’t have someone coming in and asking, essentially that, actually paracetamol or what have you is fine. Go home.

[00:34:23] Caroline Morton: we do have a couple of patients. like, um, scenarios or you’re not patient. So diseases where, um, essentially the treatment is like go home and rest and take ibuprofen, um, things like musculoskeletal pain. Um, and we do have like a, you know, cold. Um,

[00:34:46] Tim: do turn up to the ED with cold.

[00:34:48] Caroline Morton: Oh yeah, I mean, I used to, I mean, I, when I was working in general practices, I used to get a cold literally every two weeks because I saw so many sick kids for, for winter.

Um, so yeah, it was, yes, I, I don’t miss that aspect of it. Um, But yeah, we do have them. We don’t do purposely boring, uh, shifts, but

[00:35:14] Tim: No, no, no, it was just, I, more, more out of curiosity. I don’t

[00:35:17] Caroline Morton: I think it’s a good point, and we do have, um, we, I would say the vast majority of the, the diseases, they do present in the sort of classic boring way, you know, the chest pain patient, uh, and then there’s sprinkled in, there are a few, you know, um, You know, and more unusual presentations of the same condition and so there is you have to make sure that you’re still looking out for those Those you whatever it is that’s presenting as something slightly unusual, which which happens

[00:35:49] Tim: Do patients ever get transferred to the doctor in a way that either misses or misrepresents? So it’s essentially the handover. Um, and they have to go through the process.

[00:36:05] Caroline Morton: we don’t have that but we are just gonna have handover happening. Uh, from, you can arrive in the emergency room via the ambulance or via just walk in.

And so there’s, um, in the new version, hopefully which will be out in Autumn time, there will be a, uh, capacity to have like a triage nurse give a hand over.

And an ambulance, crew give a handover and I think I might add some variation into that.

So sometimes they’re just missing, um, you know, maybe not from the ambulance crew, but certainly from, you know, maybe the, the nurse hasn’t got to them yet. Um, and so there is no sort of handover. Um, so yeah, there’s, there’s

[00:36:48] Tim: Right. So you arrive before the handover does. And so the patient’s waiting there to be, and you, okay. Yeah.

[00:36:55] Caroline Morton: to decide from the, it’s sort of the, you’ve got to decide who you take first. So everyone starts in the waiting room. So you’ve got to make your decisions about who you take first, and that can also cause problems, because you might be very tempted to take the older person, who seems like, you know, they’re really sick, because they’re, I don’t know, they’ve got chest pain, it turns out they’ve just pulled a muscle, wear it, and you’ve missed the 44 year old.

who’s got, who’s actually having a serious medical complaint. Yeah. So, um, there are, yeah, there’s sort of, you make decisions about, you know, who you should take

[00:37:32] Tim: I am laughing to myself. Yeah, no, I was going to say, well, do you have staff walkouts and how much of other kind of complaints? Like what if someone forgets to put the T on? Um,

[00:37:42] Caroline Morton: yeah. I mean, the only sort of distraction we have at the moment is the bleep, which is the bane of my life when I was a junior doctor. Um, I mean, I, it was, I had such a, I think it’s a response to it that any time I heard a beep noise

[00:38:00] Tim: beep, beep, beep, beep.

[00:38:02] Caroline Morton: years, I used to look down at my right hip where my pager was or wasn’t and, um, and look for the bleep, even though I was on the tube or somewhere, it was just someone’s phone.

Um, so yeah, that was a big distraction. Um, but yeah, we, we have capacity to add in more distractions.

[00:38:20] Tim: Yeah, I’m sure they’re just like, uh, it will be hard to try to make sure to kind of stop building new stuff. Cause you could always add more complexity. Um, and

[00:38:33] Caroline Morton: learning point for me is it’s very tempting to add more and more features, but feature creep really is a. Is a really big problem. And so, um, it’s really important to just make sure that you are just making sure that the things first are things which people actually want and, uh, that you have been paid to do or someone is willing to pay for.

Um, and. Yeah, so that’s, that’s a, it’s definitely a learning experience running a company and having to start thinking about these things.

How do you introduce yourself? As a software engineer?

[00:39:07] Tim: So, how do you describe yourself to, let’s say a friend of a friend you, uh, like how, how do you introduce yourself?

[00:39:18] Caroline Morton: Oh, that is a, that’s a difficult question.

[00:39:22] Tim: I know it’s, it’s awful. Sorry. It is, it is a challenging one.

[00:39:27] Caroline Morton: so I’ve, I’ve adopted the, I’ve adopted the handle of like, uh, I’m just going to call myself a software engineer now, which feels kind of still slightly uncomfortable. Um, but I think. You know, I’ve been writing, I’ve been writing software for like three and a half years full time and You know because I wasn’t even before I went to the company full time.

I was writing software and not working as a doctor and so it’s yeah, that’s that’s what I’ve decided to go for and that might annoy people but that’s what I’m going for

[00:40:05] Tim: Who would it annoy?

[00:40:08] Caroline Morton: Yeah, I think It’s difficult.

[00:40:10] Tim: Sure. So, so, uh, may I, oh, let’s, let’s ask that again. So if I could probe slightly. I’m curious as to, are you saying that you haven’t earned your place to call yourself a software developer? Is that what you’re saying?

[00:40:28] Caroline Morton: I sometimes feel that, um, because I haven’t done sort of the equivalent of junior doctor, doctoring in software. ,

[00:40:37] Tim: it helps, I don’t know if this does, people fly me around the world to listen to me speak.

[00:40:44] Caroline Morton: that

[00:40:44] Tim: I have

[00:40:45] Caroline Morton: great gig if I can get it. So that

[00:40:49] Tim: I will look, if I can figure out a way to afford to bring you back to. a country with very, uh, if I can afford to bring you to, to, to New Zealand next August, I’ll try the, um, but I also have not.

So one of the things that software engineering, or even the term software engineering is difficult, right?

That the software industry does not have centuries of practice.

It doesn’t know what it is. It means to be in the profession. It isn’t a profession. Is it a vocation? It’s really hard to know. It’s primarily filled with people who are, most people describe themselves as self taught. There are very few people that start at university and go through a tradition, let’s say a conventional education.

Or educational pathway. There’s a lot of frustration that people need to bulldoze through, I think, when you’re learning to code. You’re continuously fighting syntax and then semantics and you’re always trying to use it. slightly away from what you want. You’ve got this conceptual idea about what you want to construct, want to build, but you’re actually, you don’t have the cognitive, you don’t have the intellectual ability to actually implement it.

And that’s extremely frustrating.

[00:42:16] Caroline Morton: Even when it works though, and you’ve implemented the thing, I often find myself thinking, you know, “Is this the right way?”

Like, have I, should I have used a different design pattern? So I’m a bit obsessive about reading like software engineering books, like trying to upskill in that area.

I’m now going, I mean, ChatGPT is actually a great resource for, you know, asking questions like, what, what is this thing?

[00:42:42] Tim: Yeah, no, I agree. I actually..

[00:42:45] Caroline Morton: Yeah.

[00:42:46] Tim: I think it’s very nice from the, Although I’ve kind of defected slightly to Claude, um, which is, and the other one, um, although there are several other ones, but yeah, I, I think that these AI assistants are quite nice for being able to give you like a condensed Wikipedia page of a topic.

Um, and I ask it silly questions, like, um, and I think that one of the nice things about software. Oh, so, so actually, no, I don’t want to, I want to. Try to figure out If there’s anything more that software can do, do better at allowing people who are changing careers to feel welcome? Um, is that.

[00:43:39] Caroline Morton: So I actually think the software engineer, the software engineering community does this very well. Um, because, and, you know, the fact there’s so many, you know, all you need is a computer and an Internet connection and, you know, a bit of willpower and you can probably teach yourself, you know, and that’s what we did.

Like we’ve all pretty much, well, not all of us, obviously, uh, but, you know, Most of us have self taught, used books, used videos, whatever it is. Um, and so you can get into, I think you can get into this profession. And I wouldn’t say, I don’t, I mean, there’s obviously some parts of the community which are a bit unfriendly, but, um, I think it’s also, you know, we also have to look within ourselves.

You know, I, no one said to me, oh, you’re not a proper software engineer. I’m, I feel that sometimes. And so, and I think if you speak to lots of people, that’s what they often will say. Um, And, you know, especially junior people, they might say, Oh, you know, I’ve just write a few lines of, of code, you know, or they’ll have some way of couching it without using that label.

Uh, so I decided three years ago, I was about, I was just going to start using it. And it felt really uncomfortable to start with, but now nobody’s sort of really questions it. Uh, but it is, Yeah, it’s something to have a think about, but it, you know, it’s your own, it can be your own insecurities. It’s not necessarily a problem, I would say, with the community, because actually I think the community is way more accepting of you being self taught than, I can’t think of another profession.

where they’d be like, you know, they’re not exactly going to be like, I’m a self, I’m self taught civil engineer. They’d be like, you’re not building a bridge, like get out of here, like, or I’m a self taught doctor, like there’s not going to be that. So I think, yeah, I think that’s a really positive aspect of it.

You can change career. It’s so interesting how many people, how they’ve got into software engineering.

What about the term “entrepreneur”?

[00:45:36] Tim: Yeah, I agree. I actually expected you to say something slightly different, if I were to look at what you’re building, I would describe you as an entrepreneur who is using software, essentially. Like I’m surprised that your personal emphasis is on the code rather than the company, let’s say.

Um, however, it’s fascinating to me that that’s how, that’s kind of the mantle that you’ve chosen to wear.

[00:46:08] Caroline Morton: Yeah, it’s the one which feels most comfortable. I just can’t ever imagine or describe, I just, I just cannot imagine ever describing myself as an entrepreneur. That seems, I just cannot. Um, but, but yeah, it’s, uh, I, I, yeah, I’m enjoying wearing the software engineer mantle for now. So, or

[00:46:28] Tim: Yeah, superb. No, it’s absolutely, again, I, I don’t know whether anyone will do, does one wear a mantle? I’m presuming, I presumably

[00:46:38] Caroline Morton: I don’t know. It’s why I changed it to label. I was like, do you wear a I’m not sure actually. So, um,

[00:46:44] Tim: ..but, um,..

[00:46:45] Caroline Morton: No doubt there will be comments correcting us.

Women in Rust

[00:46:47] Tim: Yeah. I, I, I, I’m looking forward to it.

I heard you are involved with Women in Rust. What is Women in Rust?

[00:46:56] Caroline Morton: So Women in Rust is a community group that I set up with a, uh, another person called Lizzie Holmes, who is head of operations at SurrealDB, which is a database company that’s built in Rust, multi modal database company. Um, and so basically we’ve set up, um, this Women in Rust group. It’s currently a meetup group, and we just started in, I think, April.

We had our first meeting. meetup. Um, and we’ve already got 190 members. Um, and so that’s been really positive. We’ve had some education, uh, lectures. The first two were me and then we’ve now, I’m delighted to say, we’ve got external people now coming in. These amazing female role models who can come and talk about how they’re using Rust to build their business.

And we’ve, you know, we’ve had people who Rust is their first programming language and then people who are super experienced in other and they’ve come from sort of different backgrounds. Um, so yeah, if you are interested in Women in Rust, you can join the meetup group and there probably will be Slack or some other form of electronic communication coming soon.

Uh, once I figure that all out. Um,

[00:48:10] Tim: Is it specific to London or at least the south-east of England or?

[00:48:15] Caroline Morton: so it’s an online, so we always run, um, yeah, we always run online versions of the event. So actually we’ve only had one in person meet up, everything else has been online. And so we run this monthly lunch at, uh, what’s it called, Learn at Lunch, and, um, that’s, you know, an hour over the lunch break every week, and not every week, every month, um, and that’s been really fun, really friendly community.

Um, and that’s recorded and is put on YouTube and sort of various different places for people to watch. later if they want to. Um, and then we also have just started like a community kind of catch up, um, as well, that’s new for us just to sort of try and encourage people to introduce themselves, talk about who they are, why they’re interested in Rust.

Um, and we’re going to be running some workshops in September on APIs and hopefully some other stuff next year. So yeah, very grateful for Surreal Database Company, a DB, Surreal DB for sort of being, providing such a lot of, um, support to us and, and Lizzie’s amazing. So you should definitely talk to Lizzie one day.

She’s a real advocate in the space.

[00:49:26] Tim: I shall. Well, at least I’ll go and nag her .

I have a question

The PhD

[00:49:31] Tim: Are you doing a PhD?

[00:49:33] Caroline Morton: I am doing a PhD, a very part time PhD, but don’t tell my supervisors. Um, so I’m doing a PhD in, we haven’t talked about this, um, but I’ve also had a bit of a career as an epidemiologist. I actually did it before I trained as a GP. I was, did. epidemiology. And so I, yeah, I’ve been involved with it for sort of 10 plus years, um, or 10 Um, and eventually I’ve taken the, um, sort of bit in the bullet, I suppose is the phrase, and, um, doing a PhD, um, at the London School of Hygiene and Tropical Medicine in medical statistics.

My PhD is on creating synthetic data. Um, for realistic synthetic data for electronic health record research, which is pretty niche, but really important.

[00:50:26] Tim: No, I, I want to kind of scream because this is sort of slightly incredible. I’ve, I’m, it’s unfortunate that we’re right at the end. I, um, I want to talk to you about record linkage actually.

[00:50:39] Caroline Morton: Okay, it’s something I’ve thought a lot about and I’ve been very frustrated with in the past. And I think it’s easiest in the UK, actually, of anywhere. Um, you know, I think we’ve got a very good setup for linking records. Maybe not the easiest, but it’s sort of up there. And even then it can be a frustrating experience.

So

[00:51:00] Tim: I bet, I bet, I bet. Well, I shall let you go and have a really, uh, and please do have a pleasant day. Um, how has summer been, by the way? Has it been heatwaves and disgusting heat? Or,

[00:51:13] Caroline Morton: it’s been distinctly not that. It’s been a lot of rain. Um, but it’s. Yeah, I mean, we’ve had a few days of sun, so it’s been, been nice to have that. Unfortunately, I managed to get COVID during the time I was meant to go on my summer holiday. So I did not get to go to Portugal, but nevermind. Um, we have had a few sunny days since then.

So,

[00:51:36] Tim: okay. Hey, well, it’s been a very sincere pleasure. And, uh, thank you so

[00:51:40] Caroline Morton: thanks for having me.

[00:51:42] Tim: Cheers. Um, I shall push stop.