Note: Transcripts may contain errors. If you wish to re-use all, or part of, a transcript, please contact CBC for permission. Please check the corresponding audio before quoting in print. Copyright © CBC 2023

The Current Transcript for May 29, 2023

Host: Matt Galloway

STORIES FROM THIS EPISODE

Prologue

[Music: Theme]

SOUNDCLIP

JONATHAN STOKES: We're dealing with a pathogen that's inherently resistant to most, if not all, current classes of antibiotics that we deploy in the clinic.

MATT GALLOWAY: It is one of the most dangerous and hard to treat superbugs called public enemy number one by those who study it. Finding an antibiotic that could potentially attack it would have been an epic task for human researchers. Instead, Canadian scientists used artificial intelligence to find the answer in under two hours. In the face of doomsday warnings about the technology, we'll talk about using AI for good, especially in the world of health care. That's in an hour. Also this morning.

SOUNDCLIP

FEMALE SPEAKER: Mental illness is a medical condition that is diagnosed by a doctor. This is when our brain gives us challenges with our thoughts, feelings or behaviour over a long period of time.

MG: Teenagers are suffering. A mental health crisis exacerbated by the pandemic has left them searching for supports that can feel like they are impossible to access. In 30 minutes, the role that family doctors could play in getting those teens the care they need. We begin, though, with the wildfires sweeping through the outskirts of Halifax. Good morning. I'm Matt Galloway and this is The Current.

[Music: Theme]

HALIFAX FIRES

Guests: John Barrington, Tina Simpkin, Scott Tingley

SOUNDCLIP

FEMALE SPEAKER: We just heard on the radio that there was a fire. And then as we got closer down, we could see the smoke And it's, you know, credible. And it's across the road where, sort of the subdivision across the road.

MG: It is a fast moving fire just outside of Halifax that started yesterday afternoon. And by evening, thousands were already evacuated. Dozens of fire crews are struggling to contain it. The Halifax Regional Municipality has declared a seven-day state of emergency for affected areas, and schools are now closed in places where the fire is burning. At least 10 homes have been destroyed in the Westwood Hills subdivision in Upper Tantallon, Nova Scotia. John Barrington is a roofer in Dartmouth. He was at the scene yesterday. John, good morning.

JOHN BARRINGTON: Good morning.

MG: Tell me what you were doing yesterday when you got wind of this fire.

JOHN BARRINGTON: We were putting a roof right next door to where the fire started. And basically, Matt, we were on the roof and started smelling smoke and it kind of peeked our interest. Looked up over the peak of the roof and the backyard was on fire.

MG: What did it look like? Tell me what you saw.

JOHN BARRINGTON: Well, it happened so fast because it was quite windy. It was just a small fire at first, and then all of a sudden, there was 50 foot trees on fire.

MG: So what did you do then?

JOHN BARRINGTON: Well, we ran and told the home owner that there was fire next door, called 911 and people started seeing the smoke themselves around and we were telling everybody, you know, get out and do what you can and basically just tried to get fire or keep the house from burning down and get the people out of it.

MG: So it sounded like it moved pretty quickly. As you said, there was nothing, then you smelled smoke and then suddenly, these 50 foot trees are on fire.

JOHN BARRINGTON: Absolutely. And it was jumping was the worst part. We were there trying to do what we could do and then all of a sudden, there was a fire started behind us. And then all of a sudden, there was a fire started 20 feet the other direction. And it was jumping and it was moving very fast and very hot. Yeah.

MG: I was going to say, what did it feel like?

JOHN BARRINGTON: It was very hot and very smoky.

MG: And so when this is happening, as you said, there are people who are there in those homes, the home owner’s there. You tell that home owner. What else did you do to try and let people in the neighbourhood know?

JOHN BARRINGTON: Well, it was me and my crew, the house that we were working on, I went in there and kind of forced the lady to go outside and get out of the house, her and her child and her cat. And then the other guys in the crew were on the street and, you know, telling everybody that they could, you know, hurry up and get out. Yeah.

MG: And then how quickly was it spreading after that? Where was the fire going from that point?

JOHN BARRINGTON: Well, I don't know, probably about 10 minutes or so after we called 911, the fire truck showed up and it was spreading so fast that they pretty much immediately evacuated us, told us we got to go.

MG: And so before you smelled that smoke, you had no sign that there was a fire that was that close.

JOHN BARRINGTON: No. No. It happened so fast, Matt. It was pretty intense, really.

MG: What do you think has happened to those houses?

JOHN BARRINGTON: Well, apparently, the house that we were working on, the fire department did get it stopped right just before the house caught fire. But I'm led to believe that there was a bunch of other people that weren't so lucky.

MG: Hmm. And you were evacuated out of the area.

JOHN BARRINGTON: Yeah.

MG: So what was that like? I've seen some of the footage of people trying to drive down the road and you see like the bank of smoke and you can see some of the fire kind of, you know, popping through some of the trees and stuff. What did it look like when you were trying to get out?

JOHN BARRINGTON: Well, yeah, that was, you described it pretty good, except for at the start, we were right next to it. Like I said, it was pretty hot. We were getting little embers and stuff hitting us, and that's how close we were.

MG: Wow!

JOHN BARRINGTON: Yeah.

MG: that sounds terrifying.

JOHN BARRINGTON: Yeah, it was pretty freaky at one point in time, for sure.

MG: Yeah. Have you ever seen anything like this before?

JOHN BARRINGTON: No. And I've been doing this kind of stuff for a long time and no, no, nothing like this.

MG: No. So having done this, I mean, what happens now, do you think? These fires are still burning in a bunch of parts of the area outside of Halifax and you have the state of emergency that's there. What's your sense as to what's going on right now?

JOHN BARRINGTON: I'm not really sure, to tell you the truth. I'm just getting up here not too long ago myself and just meeting up with the fellows and, you know, learning what's going on myself, right? So I don't know. I know that a lot of people are hurting right now. That's for sure.

MG: Yeah. I'm glad to talk to you about this. This is quite a scene, not something that you would expect to have happen on a Sunday afternoon.

JOHN BARRINGTON: Absolutely not. I was looking forward to sitting down and drinking some beer, but that didn't happen.

MG: And suddenly, you were fighting a fire.

JOHN BARRINGTON: Suddenly, I was fighting a fire. Exactly. Yeah. Yeah.

MG: John, take care of yourself and thank you very much.

JOHN BARRINGTON: Not a problem at all. Thank you.

MG: John Barrington is a roofer in Dartmouth, Nova Scotia. Tina Simpkin is a CBC meteorologist in Nova Scotia. Her family was forced to evacuate their home yesterday. Tina, good morning to you.

TINA SIMPKIN: Well, good morning to you.

MG: How are you doing?

TINA SIMPKIN: Well, you know, I'm doing okay. Maybe it hasn't set in just yet about the events that just have happened over the last 24 hours, but right now, I can tell you and I can honestly say that I feel very grateful. I have my kids that were with me yesterday. I got into my house in time to get the pets and we got out of there. And I have a place to stay at my mom's and not everybody can say the same thing.

MG: So John was talking about what he saw. He was on a roof and the fire appeared in the backyard. Where were you when you got word that there was this fire?

TINA SIMPKIN: Well, we were actually in a ball field. We were past the airport about 35, 40 minutes away from my home. And so every ball player on that team was from the Hammonds Plains and District. So Hammonds Plains covers Hammonds Plains. It covers Upper Tantallon. Tantallon also covers Timberlea. So kind of a circle around the outskirts of Halifax and very close to the city. So we knew that there was a fire going and we were all kind of monitoring social media and seeing what our neighbours were saying about this fire. So the ball game ended and I've got to tell you, you know, we were all pretty excited about this weather because it's been so cold in Nova Scotia for the month of May. It's also been very, very dry. And in fact, tonight, there are frost advisories in effect for parts of the province. So the weather has been all over the map, but what has been consistent is the lack of rain. And we haven't seen a whole lot. So we are all driving home and as we’re, I mean, you could see the plume of smoke from where we were. And again, we were about 35, 45 minutes away on the highway, away from where our homes were. And we started to get word that one of the players’ homes had caught on fire and probably is not there. And then we were, oh my gosh, this is really serious. As we're getting closer and closer to Hammonds Plains where I live, the fire moved out of Upper Tantallon into Hammonds Plains, and they are adjoining communities. Our winds yesterday were southwesterly winds, 40 gusting up to 60 kilometres per hour, and we had hit a high of 31 degrees. So it was hot, windy and dry. And of course, that fanned the fire into the surrounding neighbourhoods. As we're going on the Hammonds Plains Road, and if you're familiar, you know that this road is two-lain, back and forth oncoming traffic. It is a very congested road on the best of days. No matter what time of day, you're always going to get stuck in traffic on the Hammonds Plains Road because there's only one way in and one way out. So as we're driving and the traffic is coming in both directions, it went from daylight to darkness into the smoke as we're going in. So my two kids were in the car with me. And of course, we're just trying to get home to get the kids. I call them the kids, my four-legged kids, the dog and the cat. And so we're trying to get into our subdivision, the traffic is horrific trying to get in. And at that point, our particular subdivision hadn't been evacuated, but we had just heard, we got the alert that the Westwood Hills subdivision was evacuated. So at that point, we were not evacuated or mandatory evacuation, but it was very clear that we all needed to get out of there. So I saw as I was pulling in my subdivision, the trucks, the Nova Scotia power trucks. They were taking all the power down, getting all the power down. So while we’re two minutes into the house, the power went out. And again, it's day, it's the middle of the afternoon. The sun is still high in the sky. It doesn't set here until after 9:00. So there should have been plenty of daylight, but it was dark in the house because of the smoke, how much smoke that was there. And then we just found the cat, got the dog, grabbed a few things. I'm thankful that I had a dentist appointment on Friday because I have a toothbrush in my purse. So we just grabbed a few things and we just got out of there. Normally, it takes me about a minute to get out of my subdivision from my house to the main road. It took 15 minutes to get out.

MG: There's some dash cam footage that's floating around online from Hammonds Plains and it shows, this is not you, but it shows somebody driving out of the neighbourhood and there's fire on both sides and there's embers blowing across and heavy smoke. Was there anything like that as you were trying to get out?

TINA SIMPKIN: Oh, for sure. And I mean, I was driving even though we were pretty much stopped for a while. My daughter has equally the footage that was going on. And, you know, people were panicking. They were upset. I want to say that, and I don't mean to make this sound mean or anything, you have to understand that there are firefighters, police officers on the scene that, you know, they're putting their life at risk to help, but there wasn't a lot of organization to get people out of the subdivision. And again, there's one way in and one way out. And out of the subdivision I live, there's two ways out on to the main road, but one was blocked off because it was worse on the eastern side of the subdivision than it was on the western side of the subdivision. So people trying to get out of a very packed subdivision that continues to grow like so many communities across this country. You know, there's so much expanding going on and people are moving in, but there's not a lot of foresight on how to get people out in case of an emergency.

MG: Did you say that you got an emergency alert? This is a question, obviously, in the wake of the tragedy at Portapique where people were saying, you can't just tell people that there's an emergency on social media, that you need to have an emergency alert on your phone.

TINA SIMPKIN: That's right.

MG: So you got something like that.

TINA SIMPKIN: So we did. We got an alert to say that Westwood Hills subdivision was being evacuated. But the subdivision that I live in is called Highland Park, and we at that point had not gotten the alert. It was when we were in the house, the power had been shut off and then we got the alert to evacuate. But I mean, it was clear when we were pushing in and trying to get into the Hammonds Plains area that we needed to get out. It was really, really bad. Seriously, it was just like a movie, but you were in it and it was real. The scenes, I'm a meteorologist. I'm constantly looking at the sky. This is part of my passion. I have never seen the sun like that. It was red. It was when you could see it going into the smoke. And then when you couldn't see it in the smoke, what kind of sky? And just to see the convection with these storms and how the smoke was swirling, being fuelled by those winds, again, southwesterly winds, 40 gusting 60 kilometres per hour, it was something else, something I'll never forget.

MG: So I'm just back from Alberta where the fires there were roaring. And what people were saying is that the conditions are such that these fires are burning earlier, they're burning hotter, they're burning differently because of what's in the forest, but also of a climate that's getting warmer and warmer and warmer. Are you seeing, I mean, this is in many ways the other side of the country, but is that what you're seeing as well?

TINA SIMPKIN: Yes, absolutely. And we have to recognize that. And I'm glad that you brought that up because climate change is here. We are living it. This is just the beginning. May in Nova Scotia in the Maritimes is usually wet. We're not coming out of a drought-like conditions in May. It's never usually this cool, at least in the beginning part of the month. And at least, by, you know, the holiday weekend, usually, most people are in their pools or at least opening up their pools at this time of year. Nobody's got their pools open because it has been so cold. And then it switched right away into the upper 30s, not upper thirties, upper 20s, lower 30s and it's been so dry across this province. Normally, we're harvesting strawberries. Well, there hasn't been enough water to make great strawberries, so we're not getting the strawberries. And this is all attributed to climate change and it is real and it's just the little things that we can do. One, talk about it that it's happening and plan for the future.

MG: We're going to talk more about this in just a moment. Just before I let you go, do you know how the subdivision is, how your home is? Have you heard anything?

TINA SIMPKIN: Well, I heard most of what I know from my son and from his baseball team because they're all very active on social media and chatting amongst their circles. And I know over three players that don't have homes this morning from our team. So if I know of three players and now I know of about five or six people that I know just directly, not indirectly, directly that do not have homes, as I'm speaking to you, I have no idea if my house is standing. I'm going to be finding out as the pictures come in throughout the day today. But again, I truly feel grateful. I have everything that I need. My kids are safe, my four-legged kids are safe and we have a place to stay.

MG: I wish you the best of luck, Tina. It's been quite a day. Thanks for talking to us.

TINA SIMPKIN: You're welcome. Bye.

MG: Tina Simkin is a CBC meteorologist. She was in Halifax. Scott Tingley is manager of Forest Protection at Natural Resources Nova Scotia. He's in the Provincial Fire Coordination Centre in Shubenacadie, Nova Scotia. Scott, good morning to you.

SCOTT TINGLEY: Good morning, Matt.

MG: What do you know about the state of the fire right now?

SCOTT TINGLEY: I know the state of the fire's out of control. We're still waiting for an updated size. You know, department of natural resources officials are on site now. They're arriving through the night and into this morning. We're working very, very close with the municipal officials, RCMP, Emergency management office, trying to get an update on what's going on in terms of size and where we think this fire might head today. So.

MG: What are you up against in terms of trying to fight this fire?

SCOTT TINGLEY: Yeah, it's a challenging situation. Obviously a terrible news about how it's been impacting the community already. So the challenges are it was extremely dry, extremely windy yesterday. You know, it was some of the worst fire conditions we've seen in a while. And so, you know, these fires grew quickly yesterday fast. And we had a wind shift start overnight with a cold front passage. So winds have changed direction today and so that can present a challenge. So active fire perimeter that what we would call was the flank yesterday could become a hit fire today. So officials and crews are going to, that'll be all part of their assessment in how they attack this today, you know, with that wind shift. But things are still very dry. We didn't get any measurable rain or precipitation with that passage. And so things are still very dry.

MG: Talk about the dry conditions. Tina was talking a little bit about that. And you'd been saying as well that, I mean, it's only May. There was a fire, a wildfire earlier this month, I guess, on the other side of the province. And there are concerns about the conditions. Again, things being drier, but also, I gather because a lot of trees were knocked down or perhaps not cleaned up yet from Post-tropical Storm Fiona, that that has led to more unstable conditions.

SCOTT TINGLEY: Yeah. As your meteorologist had been mentioning, it's extremely dry. We've had well below normal amounts of precipitation for both March and April from what I understand. We do have a few bits of rain kind of come through there, but it's very dry. In terms of damage from Fiona, certainly, it's going to impact it. We're mostly seeing those impacts, you know, kind of Halifax to Truro and East. The western end where some of these fires are weren't as impacted. But anywhere we were, we've got heavy amounts of downed trees, particularly softwood trees, but downed trees that adds fuel to the fire. And so starting last fall, you know, mapping had begun, you know, to scope out how, you know, where all this downed debris is. And that work continues and that's incorporated into how we respond. So, you know, we know fires, you know, can spread a little quicker and become more intense in, you know, where those, you know, heavier pockets of downed trees are. And that risk is going to, you know, we'll be facing that for a few years to come as we wait for those trees to decompose and then be less susceptible to burning. So particularly in the western end of the province where we've had a lot of these fires this past spring, it hasn't been as impactful. As I said, there wasn't as much Fiona impact in that region, but it’s definitely encompassing.

MG: It definitely is a complicated thing.

SCOTT TINGLEY: Yeah, sure.

MG: Just before I let you go, I mean, as I was saying, your colleagues in Alberta and British Columbia are already saying as we make our way through May, that this is going to be a particularly challenging fire season. How worried are you about what's yet to come not even in the summer, but for the remainder of the spring?

SCOTT TINGLEY: It's been a very challenging season here at home and across the country. We work very close with all the firefighter agencies across Canada and we all try to help each other out. And that work continues. So we're all, you know, monitoring the conditions very closely. And I know a lot of work's going in to try to allocate resources and prioritize nationally to make, you know, to make sure that we're putting resources where, you know, they're going to be most effective and can mitigate, you know, the biggest impacts. So it's going to be challenging. We need, you know, we need the weather on our side, but we've got a lot of hardworking, dedicated people out on the ground and behind the scenes supporting all these incidents. And so really, hats off to all the first responders and volunteers that are helping out and we're doing our best.

MG: Scott, thanks for this.

SCOTT TINGLEY: Thanks so much.

MG: Scott Tingley is manager of forest protection at Natural Resources Nova Scotia. You're CBC News with the latest on these fires is coming up next. And then, 40% of people who have depression have their first experience with it before they're adults. Teens in Canada often go undiagnosed and untreated. And a new study finds that health care providers in this country don't feel confident in helping teens with depression. We'll hear from a teen and a doctor when we come back in six minutes. My name's Matt Galloway and this is The Current on CBC Radio and the CBC Listen app. Stick around.

[Music: Theme]

TEEN DEPRESSION

Guests: Katie Foley Tremblett, Dr. Anna Banerji

MG: Good morning. I'm Matt Galloway and you're listening to The Current.

[Music: Theme]

MG: Artificial intelligence has been billed as the great problem solver, a balm for all of society's aches and pains. It's also the subject of doomsday scenarios where humans become obsolete. Now, a group of scientists have made a major breakthrough using AI to attack a deadly bacteria. Coming up, should AI really be lauded as a saviour when it comes to health care? That's in 30 minutes. But first, the state of teen mental health.

[Music: Theme]

MG: Over the pandemic, the rates of depression in Canadian teenagers grew. And now, even as the pandemic wanes, those rates remain alarmingly high. With long waitlists to see psychiatrists and other mental health professionals, many teens are simply going undiagnosed. And a new review released today in the Canadian Medical Association Journal says we are overlooking a critical part of the solution, family doctors. Katie Foley Tremblett is a 16-year-old high school student who has struggled with depression and anxiety. She is a youth advocate with Children First Canada. Katie, good morning.

KATIE FOLEY TREMBLETT: Good morning.

MG: When did you start to think that you might be struggling with depression and anxiety? Take me back.

KATIE FOLEY TREMBLETT: I don't remember when I started struggling with depression specifically, but I know that anxiety, I think I've been seeing some type of counsellor for probably since I was like eight or nine.

MG: What was going on when you were eight or nine that led you to have to see a counsellor?

KATIE FOLEY TREMBLETT: Some family issues. My dad and his then girlfriend who was really close to us were breaking up and there was just some family issues then.

MG: How was that impacting your life?

KATIE FOLEY TREMBLETT: It was destabilizing probably would be the word. Yeah, that would be the best descriptor. It was just destabilizing.

MG: And so what was, I mean, if you can describe it, how did that anxiety manifest itself? What was going on in your life where you would notice that anxiety?

KATIE FOLEY TREMBLETT: I used to get panic attacks pretty often. I would like pick at my skin, my cuticles a lot and be really, really vigilant about like making sure before I go to sleep, I would double check the locks and all of that type of stuff. So yeah.

MG: Is that something, I mean, when you talk to friends, is that something that you saw other people your age going through?

KATIE FOLEY TREMBLETT: At that point, No, no, not really. Once I got a little older, I’d seen people around me struggling with the same issues that I was with the anxiety and the depressions.

MG: You mentioned going to see some people who could help you at that early age. Tell me a little bit about the kind of support you've been able to receive over this time.

KATIE FOLEY TREMBLETT: I remember that I was taken to, there's a program here. I don't know if it's nationwide, but it's called Doorways and they do like triage for, like mental health triage kind of. And once they assess your situation, they'll put you on a waitlist to go see a counsellor after that. It'll take like between three and six months usually to get into something like consistent, but I've gone there I'd say three or four times now and started seeing counsellors and I guess that was the first time that I went.

MG: So the delay is about three to six months.

KATIE FOLEY TREMBLETT: Yeah.

MG: Yeah. What has that support meant for you? What has it done for you?

KATIE FOLEY TREMBLETT: There were a couple times it was really helpful, but a couple times where it wasn't as much. I found that it's harder to find someone that you really, I'm trying not to sound like I'm criticizing the program now because I'm not. It's an amazing resource, but I found that it's really easy to get put with somebody who maybe doesn't click with you or understand what you're saying as much maybe.

MG: Because it's a huge trust thing, right? Like you're really opening yourself up to somebody that you’ve just met.

KATIE FOLEY TREMBLETT: Yeah. Yeah, it is.

MG: Did you ever see a family doctor for what you were going through? Did you talk to your family doctor about any of those?

KATIE FOLEY TREMBLETT: No. No, I didn't. Here, I don't know if it's all over the country, but here in Newfoundland right now, we’re in an awful family doctor shortage. Almost nobody has them. And I think that I got my family doctor about four years ago, and I’d seen her once when I first got on her list. And then COVID happened and then I've only seen her once since then because I had to start a prescription. That's another story, but yeah, no, I've only ever seen my doctor twice and she just left actually. I don't have a family doctor right now.

MG: Oh wow! Because I was going to say, I mean, one of the things we're going to be talking about in the next part of this program is just whether family doctors could be a way to help, you know, intervene in these crises early on. But you need to have a family doctor in the first place for that actually to work.

KATIE FOLEY TREMBLETT: Yeah. I think that the statistics here are something insane. Like 50% of people or like something close to that right now like don't have them.

MG: We talk a lot about this idea that there is a youth mental health crisis or a teen mental health crisis. When you talk to your friends, how are they doing right now?

KATIE FOLEY TREMBLETT: I think that a lot of, like a significant amount of my friends are in what I would call a mental health crisis. Like with COVID happening recently and really messing with everyone's heads and just school, all of it, I would describe my friends and myself as in what I would call a mental health crisis.

MG: And so you say yourself that you're in one.

KATIE FOLEY TREMBLETT: Well, not like in this moment, but I guess a couple of years ago and more kind of recently. But I'm pretty okay right now. I'm okay right now. I am. Yeah. But yeah, yeah, in the greater picture or whatever, if you put all of us together, I'm not exempt from what I'm saying.

MG: What do you think we don't understand about that? What aren't we seeing? Because people will talk about that, but maybe they don't really understand what it is that people like yourself and your friends are going through. So what would you tell us about what it is that the people of your generation are in the midst of right now?

KATIE FOLEY TREMBLETT: I would say that it's just people think that because we're young and we're young people, that the problems that we're facing and the way that we're feeling aren't quite as valid as if we were older because the issues that we're having are like, quote, adult issues or whatever.

MG: That you'll just go through it or something like that and that this will work itself out.

KATIE FOLEY TREMBLETT: Yeah, exactly. But just because we're younger doesn't mean that the way that we're feeling or the consequences, the possible consequences of not dealing with and not getting access to proper help from a mental health issue or an elongated time of mental unrest could have, it doesn't make those any less real or serious.

MG: What would you like to see in place to better address the crisis, as you say, that many people of your generation find themselves in?

KATIE FOLEY TREMBLETT: I think that schools, not that schools are doing anything bad. I love my school. They're fantastic. But I'd say that put more resources in schools, like easier to access because that's where you'll find all of the young people almost. That's the common denominator. We’re all going to school. And I would say make it an easy conversation that's had in classes, not just something that's like posted on a bulletin board. And put in place things that make it easy for students to access fast mental health help because it's rough to access it right now. It's kind of hard.

MG: You're turning 17 next month?

KATIE FOLEY TREMBLETT: Next week. Yeah.

MG: Next week.

KATIE FOLEY TREMBLETT: Yeah.

MG: It's a big birthday. How are you doing?

KATIE FOLEY TREMBLETT: I'm good. I'm really good.

MG: This isn’t easy to talk about. You and I have never met and here you are spilling your guts about what you're going through. But it's important and I'm really glad that you'd be willing to do this. And I hope you have a great birthday, but I also hope that you are getting the help that you need as you make your way through this crisis. Katie, thanks a bunch.

KATIE FOLEY TREMBLETT: Yeah, thank you.

MG: It's good to talk to you. Thank you very much.

KATIE FOLEY TREMBLETT: You too.

MG: Katie Foley Tremblett is a high school student in Grand Falls-Windsor, Newfoundland and Labrador. Dr. Anna Banerjee is a pediatrician, public health specialist at the Faculty of Medicine and the Dalla Lana School of Public Health at the University of Toronto, and she's with me in studio. Good morning to you.

DR. ANNA BANERJI: Good morning, Matt.

MG: So we talk a lot about this idea of a mental health crisis among young people and you just heard Katie talking about that. She uses that word and says that many of the people in her generation, her friends are in that. How do you see this? I mean, is it a crisis?

DR. ANNA BANERJI: Yeah, it is. I think that we have not had so many adolescents as stressed or depressed as we've had in the past.

MG: So what does that crisis look like? I mean, we'll talk about some of the ways to tackle this, but from your perspective, what's going on and why is this happening now?

DR. ANNA BANERJI: So they estimate the numbers are about a quarter of the adolescents have had mental health disorders during COVID, so depression, anxiety, etc., and we just don't have the resources to deal with it. So they can't. There's not enough psychiatrists, family doctors, many of them don't feel very well equipped to handle it. So there's really not a lot of places for these kids to go. And sometimes, the consequences can be, you know, dropping out of school, substance abuse, etc. But right now, I don't think we've ever had it quite so bad.

MG: What did the pandemic do? Because the assumption is that this existed before the pandemic, but that the pandemic, you know, put fuel on the fire.

DR. ANNA BANERJI: Absolutely. The pandemic resulted in isolation, uncertainty, stress. A lot of kids did not adjust well to online learning. It was the are we going back? Are we not going back? Are we staying? Are we not? And post-pandemic, with the price of food going up, rent going up, there's a lot of family stress as well, and that is compounding the stress for kids.

MG: And anxiety about their futures as well.

DR. ANNA BANERJI: Absolutely.

MG: These kids are going to graduate, they're not kids anymore, but they're going to graduate into a world where life is expensive. Jobs may or may not be there for them. Who knows what they're going to be facing?

DR. ANNA BANERJI: That's right.

MG: What are the difficulties in detecting this in young people?

DR. ANNA BANERJI: I think that first of all, a young person has to either talk about their symptoms and let people know or display symptoms such as, you know, not eating or eating too much, sleeping more or less, crying, etc., not participating in things. But with social media now, a lot of kids can be isolated in their grief or in their anxiety. And people, kids just don't hang out anymore. They don't go out, they don't talk, they don't reach out to each other so much. So I think social media has also had an impact as well. Social media is like road rage. You're anonymous in a way. You can say things that could be hurtful that you wouldn't say to someone in person. And so there's a lot of factors contributing to this. Now.

MG: Katie said something interesting, which is that in some ways, it's as though people of a different generation may not really understand what it is that people of her generation would be going through, that we might dismiss it because we think that kids are resilient and they'll just work their way through this. If you're talking about depression in particular, how different is depression among younger people than it might be in adults?

DR. ANNA BANERJI: For younger people, it depends on the age. The younger children, they may not come out and say that they're depressed, but it might come out as, you know, bad behaviour, anxiety, crying, physical symptoms, tummy aches and that. And so we're more likely not to recognize it. In teenagers, it may present itself through again those symptoms like anxiety or physical symptoms, but also withdrawing that behaviour. But I think that we tend to minimize the symptoms that youth have and we tend to think, oh, they'll just get over it, it's a normal part of being adolescent, when a lot of these kids are suffering. And they're suffering a lot more than we've seen because, you know, kids have always had a hard time during that transition phase. But now, on top of COVID and post-COVID and the uncertainty, I think there's a lot more stressors on children.

MG: Well, again, it goes back to what Katie said. And you can imagine there's another generation that's listening saying, yeah, of course, you know, when I was going through adolescence, it was a tough time. Everything's changing, guts, you know, churning because life is upside down and you don't know what's going to happen. You're saying that this is a different time, that something different is happening right now.

DR. ANNA BANERJI: Absolutely. Yeah.

MG: How difficult is it for young people to get the care that they need in this country?

DR. ANNA BANERJI: It's very difficult. I mean, there's only a certain number of psychiatrists and there's long waits. Family physicians, a lot of them don't feel comfortable in dealing with mental health. And perhaps, the funding is not there so they can spend more time with an individual patient. I think that if we can get primary care providers more educated as far as depression, how to manage it, then there's less stress on the limited number of psychiatrists. And it's also very bad. You know, we have a significant problem in a place like Toronto, but even the more remote places, it's more difficult in smaller towns. But even with indigenous communities in remote communities, it is extremely difficult to get adequate psychiatric or psychological help.

MG: And again, it sounds obvious, but the consequences of that, of not addressing this early on are what?

DR. ANNA BANERJI: So it can impact a child's life or a youth’s life for the rest of their life. It could cause, again, they may have problems at school, they may end up dropping out, there could be substance abuse. You know, in the worst case scenario, depression can lead to suicide. But then, we know that depression, anxiety, it can become chronic. It can impact someone's ability to, you know, if you don't go to school, then you don't get a job and you have that cycle of poverty. We also know that there are physical risks such as increased risk of heart disease, hypertension, diabetes, that are related to depression, you know, in in youth, in adolescents.

MG: The suggestion in the absence of adequate care is that primary care physicians could relieve some of the stress on the mental health care system and help treat kids earlier. How could we go about doing that?

DR. ANNA BANERJI: I think education, having medical students trained more in psychiatry and mental health issues, but also continuing education on mental health issues. But there are also other ways, like not just family physicians. It could be nurse practitioners. You know, their salary is different and they can spend more time with someone or even having psychologists. A lot of them are not, they're private. And so only certain people that have insurance or who can afford psychologists can see a psychologist, but maybe we need to have something in place so that everyone who can't afford it has some kind of insurance or some kind of policy so that they can access mental health services so it's not on a two-tiered system.

MG: There is, as I mentioned, this review released today in the Canadian Medical Association Journal. One of the things it says is that clinicians constantly and consistently report a lack of confidence in their ability to comfort adolescents with depression. Why would that be the case? Why would primary care physicians not feel confident in their ability to comfort a young person who, I say young person, I mean, this is a cohort that goes across the spectrum, but not feel confident in their ability to help comfort somebody who has depression?

DR. ANNA BANERJI: There's a lot of reasons, I think, especially now since COVID. A lot of primary care physicians or nurse practitioners, they're very overwhelmed. They have so many people to see and there's a time crunch. And I think a lot of primary care people have not had training or adequate training with the drugs and what to watch for. And in some ways, it's easier to refer to a psychiatrist when there really aren't a lot of psychiatrists around. And so I think better preparation, more support, more education for family physicians or nurse practitioners would be part of the solution.

MG: I mean, we’re also up against a situation where, as Katie said, her family doctor just left. There are a lot of people who simply do not have a primary care physician.

DR. ANNA BANERJI: That's right. Yeah. Which compounds everything. Yeah.

MG: This is your professional work, but this is also personal for you.

DR. ANNA BANERJI: That's right.

MG: Do you mind me asking about this?

DR. ANNA BANERJI: No, go ahead.

MG: You're writing a book about your experience with your son, Nathan. Can you tell us about Nathan?

DR. ANNA BANERJI: Yeah. So I adopted Nathan from the Arctic when he was four months of age from a foster home. He was just the most wonderful, beautiful child. I wanted him to be a proud Inuk, so I used to take him to the Arctic, you know, every couple of years to see his foster family and biological family. And he became very close to his older brother, Corey. And then when Nathan was about 10, Corey died. Corey died by suicide. And when we went back to the Arctic when Nathan was on the cusp of 12, he saw the abject poverty, the food insecurity, he realized that many people had addictions in the community. And then he went to his brother's grave and that was a moment that really changed his life. He came back depressed. And we were lucky to get him into a mental health facility for two years, but then he was declining. And I went and had an emergency meeting. I went through his whole history, explained his risk factors. He had every risk factor for suicide out there. His brother had died by suicide. He was turning, my son was turning 14 the next day and we were basically dismissed. All of the concerns that I had were dismissed. And then I find him in my basement dead 12 days later.

MG: I’m sorry.

DR. ANNA BANERJI: Yeah. So even when you have psychiatric care, it may not be culturally safe or people may not be aware of the additional risks and trauma for indigenous populations or other populations, marginalized populations or refugees. So it's a real problem right now.

MG: But if you a pediatrician, public health specialist, media savvy, somebody who is plugged into the system can't get the care that you need, where does that leave the rest of us? Do you know what I mean? There are families who are listening to this who are saying that your experience, this tragic experience maybe is their experience, or they worry that it will be their experience because they're hitting those roadblocks and they can't seem to get the care for the young person in their life.

DR. ANNA BANERJI: Yeah. And also, I've been working with indigenous population for 30 years and I run the Indigenous Health Conference. So I'm very well aware and I was very articulate in my concerns. I went through a list. That's why I'm writing my book because my book is, you know, talking about the need to do more, the need to do better because Nathan shouldn't have died. He shouldn't have died at that point. And we need to, that's right. If someone like me can't get the help that's needed, then how does an indigenous person that may be on welfare or a new refugee that doesn't speak the language? They may not feel empowered enough to say, you know, this is not right and I need help. So that's one of the reasons why I'm writing the book.

MG: So what do we need to do to address the crisis before I let you go?

DR. ANNA BANERJI: I think as individual people, kindness goes a long, long way. Recognizing that someone's suffering and just to, you know, try to be gentle and kind because there are a lot of people suffering out there. There's a lot of pain out there. Validation. You may not agree with their reasons, but you can say, I understand that you're feeling this way. Those things go a long way. I would say limiting social media because social media really disconnects you from reality and from the outside world and nature.

MG: But we need the services to treat those people who are going through those crises.

DR. ANNA BANERJI: That's right. And so we need to educate primary care people on mental health, get more people trained, whether it's family doctors, nurse practitioners, psychologists, have a different funding method so that anyone can access this. We need to increase the pool of people who are able to manage these people with mental health issues, especially adolescents.

MG: This is important and as I say, it will resonate with people across this country. Your own story is key to this. And thank you for sharing it, but thank you for being here.

DR. ANNA BANERJI: Thank you, Matt, for the opportunity.

MG: Dr. Anna Banerji is a pediatrician, public health specialist at the University of Toronto. If you have a young person in your life who is in the midst of a crisis like this, we'd love to hear from you. What have you run up against in terms of trying to have whether it's depression or other mental illnesses diagnosed? But also, what have you run up against in terms of treatment for those conditions that the young person in your life might be experiencing? You can email us. The email address is thecurrent@cbc.ca. This is The Current. My name is Matt Galloway. I’m just going to tell you quickly about something that we have coming up this week on the program. It is estimated that something like 20 million people are displaced in the Americas, some within their own countries, others on the move. And many, as you might know, are headed for the U.S. border. The Cinchado border crossing sees many of those migrants who are heading north. That border crossing is in southern Guatemala on the border with Honduras. Many of those migrants have come all the way from Venezuela. The Current’s Liz Hoath had a chance to meet with some of those migrants and she will bring us their stories from the border crossings and beyond. Coming up later on this week. This will be the first in a series looking at how migrants are travelling through difficult and treacherous routes through Guatemala and Honduras. You hear the stories of people who are at the U.S. border. These are the stories of those people as they are trying to get to that border. Those stories coming up in the days and weeks to come here on The Current. Your CBC News is up next. And then.

SOUNDCLIP

JONATHAN STOKES: I am of the mindset that these AI algorithms can be used to augment the work we were going to do anyway.

MG: Artificial intelligence is being sold as the future of everything, including health care. Scientists have now used it to attack a previously unkillable bacteria. In the face of these breakthroughs, though, there are still very big questions about the speed at which this technology is moving. We’ll talk about the role and the future of health care for artificial intelligence when we come back in 90 seconds. I'm Matt Galloway and this is The Current on CBC Radio and the CBC Listen app. Stay with us.

[Music: Theme]

AI AND MEDICINE

Guests: Jonathan Stokes, Rahul Krishnan

MG: Hello again. I'm Matt Galloway. You're listening to The Current. Artificial intelligence is taking on a deadly bacteria. A group of scientists used AI to find a new antibiotic that can finally kill a deadly superbug that was until now resistant to nearly every treatment. Jonathan Stokes is an assistant professor of biochemistry and biomedical sciences at McMaster University. He helped make this happen. John, good morning.

JONATHAN STOKES: Good morning.

MG: First, I want to talk about the work that you did, but tell me about the superbug that you are trying to tackle. You describe this as in some ways, public enemy number one. What is it?

JONATHAN STOKES: The bug we were interested in addressing with this project is a bacterial pathogen called Acinetobacter Baumannii, which is a mouthful, but it's a hospital acquired pathogen and it lives in hospital settings on surfaces and it hangs out on, you know, doorknobs and hospital equipment, whatever, for very long periods of time. So it's quite common to find Acinetobacter isolates residing in clinical settings that are resistant to many antibiotics. Given that it's such a clinical burden, we were interested to see whether, you know, artificial intelligence methods could help us find new antibiotics against this bug.

MG: What are the problems that the bug creates? I mean, part of this is, as you say, it's resistant to just about every antibiotic. Given that it's, you know, we've heard about superbugs in hospitals, in long term care homes, for example, what are the issues that it can lead to?

JONATHAN STOKES: Yeah, a diverse array of issues, pneumonia, meningitis, wound infections, sepsis. It can be a deadly pathogen, especially when you're talking about patients who have perhaps prolonged stays in hospitals and may be, for example, immunocompromised.

MG: The World Health Organization has said this is a critical threat. When something like this appears, what are you up against? Before you get to the artificial intelligence piece, what are you up against in trying to figure out a way to tackle it? How difficult is that?

JONATHAN STOKES: It's remarkably challenging. And that's because we're dealing with a pathogen that's inherently resistant to most, if not all, current classes of antibiotics that we deploy in the clinic. So when we go to search for new antibiotics, it necessitates that we start looking for chemicals, antibiotics that have brand new structures and brand new functions. You know, we have to develop a fundamentally new treatment.

MG: And so why did you turn to artificial intelligence to try to answer this question?

JONATHAN STOKES: It's a very good question and one that we in fact ponder all the time ourselves, given that we do this every day. At a high level, the application of artificial intelligence approaches allows us to explore chemicals, potential new antibiotics very rapidly and inexpensively. On a typical drug discovery setting, if we're looking for a new antibiotic, we might have to go in to the laboratory and test, you know, hundreds of thousands, a million chemicals more to see which ones of those work against the bacteria in a dish. But you can imagine, you know, if we want to test 100,000, half a million, a million molecules, that's 100,000 experiments that we have to perform in the laboratory. And that's, you know, remarkably laborious and time consuming and expensive. So the idea is can we leverage artificial intelligence algorithms to do a lot of that work for us in the computer? And ideally, by leveraging these artificial intelligence algorithms, they can look at these chemicals much more rapidly. And by looking at a broad array of chemicals very rapidly, they can help us prioritize which experiments to run in the laboratory.

MG: And so with this superbug, you trained an artificial intelligence network, you trained it to do what?

JONATHAN STOKES: We were looking for new chemicals with new structures that could kill Acinetobacter. That's what we set out to do. In order to train a model to predict new chemical structures with this activity, we have to first teach the model what chemical features are associated with the antibacterial activity that we want. So what that looks like is we physically tested in the laboratory about 7500 chemicals looking at which ones inhibited the growth of Acinetobacter and which ones did not. And through training a model on these 7500 compounds, as well as their inherent antibacterial activity against Acinetobacter, that model could then learn the chemical features that are associated with antibacterial activity in this context. So once we trained that model, we could then start showing the model a whole bunch of new chemicals that it had never seen before, and then it could make predictions about which of those chemicals it thought were antibacterial and which ones it thought were not antibacterial.

MG: How many compounds did you feed into the machine?

JONATHAN STOKES: We trained the machine on 7500 about, and we found this one molecule, Addison, that displayed, you know, quite remarkable biological activity against Acinetobacter.

MG: So this has been described as a game changer and by you, was described as validating the benefits of machine learning. Tell me more about this and why you think that this is so significant what you were able to do with the velocity that you were able to do it at.

JONATHAN STOKES: I'm going to take a step back. So my first, you know, the application of artificial intelligence in antibiotic discovery was back in 2020. Me and a remarkable team of colleagues and friends wrote a paper where we leveraged artificial intelligence to predict new antibiotics against typical laboratory strain E. coli. And that was sort of initial validation, I'll say, that these algorithmic methods can work for antibiotic discovery. But again, we were using laboratory strain E. coli. We were just sort of dipping our toe in the water. But with this project, we said, can we leverage these artificial intelligence approaches to go after a pathogen that is remarkably challenging to address? You know, we set a really high bar. So when, you know, we completed this project and came to the conclusion that yes, these AI models, when appropriately trained and appropriately applied, can discover new chemicals that target such a burdensome pathogen, I feel like we're entering an era where AI approaches can meaningfully influence how we discover clinical medicine from the earliest stages of discovery.

MG: Tell me more about that because that's been the promise of AI. And we'll talk about the doom and gloom in a moment, but that's been the promise, is that this, because again of, I'll use that word, the velocity in which you were able to make those calculations, that it's going to lead to discoveries in a much more expedited way than if humans had to do those experiments themselves. So what do you see as the promise there?

JONATHAN STOKES: The promise, like you alluded to, is that humans might not have to spend so much time and effort performing these large scale experiments in order to find chemicals that might do the thing they want them to do. Meaning that, you know, we can train these artificial intelligence models to help us make predictions on enormous chemical data sets that we couldn't otherwise test. So I'm going to put this in perspective. At the largest scale using conventional, you know, high throughput chemical screening approaches, we can test, you know, on the order of a few million chemicals. And those are at the largest scale. But in using an artificial intelligence algorithm, we can relatively easily test on the order of hundreds of millions of chemicals, even billions of chemicals in the computer. So we're now talking about number of molecules at a scale that is infeasible in the laboratory. And I like to think about this as, you know, shots on goal. If we can look at more molecules, we have a higher probability of stumbling across those that have the ideal features, the ideal characteristics to treat the disease we want to treat.

MG: So we've been talking about finding an antibiotic to tackle this superbug. When you say that you will be able to make those calculations to treat the disease we want to treat, what specifically do you think the artificial intelligence and using artificial intelligence would lead us toward? What are the diseases that we might have a better chance in tackling if we can get a handle on this technology?

JONATHAN STOKES: I think that AI technology is in a position and is sufficiently advanced that it can be implemented now. What is currently lacking across many disease areas, though, are sufficient quantities of very high quality data with which to train these models. These AI models are, you know, we call it data hungry. They need to see a lot of examples in order to make robust predictions. So I think the acquisition of data with which we can train these models needs to be at the forefront of all of our thought. What does the best training data look like and how could we acquire sufficient quantities of that best training data in order to train these models to output predictions that we can trust as the human scientists doing the experiments?

MG: But people see this as a breakthrough. I mean, people have talked about different forms of cancer, breast cancer, colon cancer, what have you, that if you have the data, that this will allow us to tailor more specific treatments to those diseases. Are those the things, the kind of things that are possible, do you think?

JONATHAN STOKES: I believe so. You know, instead of training a model, for example, just on, you know, I want to find a molecule that cures some cancer, right, well, we know that lots of chemotherapy agents, for example, come with terrible side effects. One can imagine a world in which we have artificial intelligence models that help us predict new anti-cancer agents that are very specific to treating the cancer and thus, minimizing all these side effects. But again, it boils down to can we acquire the training data to train these, you know, sophisticated models?

MG: In the midst of doing this work, what do you make of the pushback against the speed at which AI development is underway? I mean, you have, you know, people like the godfather of artificial intelligence, Geoffrey Hinton, leaving Google because he's worried that he's created a monster that will eventually make us humans irrelevant, that the machines will push us out of the way, that we are developing this stuff too quickly and we don't have a handle on what it actually is capable of doing. What do you make of that?

JONATHAN STOKES: I understand the trepidation. This is clearly a transformative technology that is here to stay and has significant scientific and social implications. I am of the mindset that these AI algorithms can be used to augment the work we were going to do anyway. I don't see them replacing humans. I see them augmenting human productivity and augmenting human creativity, augmenting the work that we're interested in to begin with. You can imagine cases where, you know, these powerful and dynamic technologies can be abused, but I don't necessarily know if that's any different from any other transformative technology that has existed previously.

MG: Just finally, I mean, as a scientist, what is it like to be in the midst of this? People have described this, you know, as kind of like it's a piano rolling down a hill and it's getting faster and faster and faster and people are just trying to hang on to it as it's hurtling along. What is it like to be in the midst of this as developments are moving at such a pace?

JONATHAN STOKES: My goal currently is to discover new antibiotics to save people's lives. So if there are, you know, more robust, more powerful AI technological developments that help me achieve that goal, I am going to embrace them. That might not be the case in other domains of life. But when it comes to the work that we do and the ultimate goals that we are trying to achieve, you know, I look at these technologies as beneficial helpers to achieve those goals, you know, more rapidly and less expensively for the people who are ultimately going to benefit, which are patients.

MG: John, this is fascinating work. I'm really glad to talk to you about it. Thank you very much.

JONATHAN STOKES: Thank you.

MG: Jonathan Stokes is an assistant professor of biochemistry at McMaster University. He and a team of scientists along with the help of AI discovered a new drug that can be used to kill a deadly superbug. Our next guest studies the consequences of implementing artificial intelligence in health care. Rahul Krishnan is an assistant professor in computational medicine at the University of Toronto. He's with me in studio. Good morning to you.

RAHUL KRISHNAN: Hey, good morning, Matt.

MG: You were listening to that. How much of a game changer do you think is this when we're talking about using AI to crack a superbug that people have not been able to figure out how to attack otherwise?

RAHUL KRISHNAN: I think it's an incredible achievement to be able to build something like that using data. I think I agree with John when he says that, you know, it's a tool that we can think of as accelerating the workflow that we as scientists do on a daily basis. And if it helps us get to discoveries even 10% faster, that's a huge win for society as a whole because we can start making and discovering these drugs at a much faster scale.

MG: So you study this. In the simplest terms, what do you think artificial intelligence can do when it comes to changing how we think of medicine and what happens in medicine?

RAHUL KRISHNAN: Right. So, you know, when we think about artificial intelligence, I think as a field itself, on its own, it's rapidly evolved. And I like to give the story of 10 years ago, I was in graduate school, and the best thing that we could do with AI and machine learning was simply take a digit, take an image of a digit and predict what number that digit actually was. But 10 years later, we’re now able to have machines that create freeform text that are virtually indistinguishable, indistinguishable from what a human might write. And I think that class of methods has grown in nature in and of itself. But to get back to your original question, what can it do in medicine? I think the key idea in for AI in medicine is really to think about how we can help clinicians make decisions faster, more safely and more efficaciously.

MG: What does that mean?

RAHUL KRISHNAN: So it means taking a look at a patient's clinical data and using them in conjunction with a predictive model to assist in clinical decision making. So let me give you an example. Whenever a patient interacts with a health care system, there's a certain amount of that data that is collected. So if you or I were to go get a lab test, a panel of numbers corresponding to each of the values of the lab that the clinician ordered would be recorded on a database somewhere. And for people who suffer from chronic diseases or even those who occasionally see the doctor, their history, so to speak, or their medical history is sort of recorded as this long sequence of data points. And a lot of machine learning in health care, this intersectional field that I work in, is about taking that data and using it to develop tools and predictive models that can assist clinicians in making better decisions.

MG: So if something is happening now, it might be able to, based on what we've already seen in the data that's inputted, say this might happen next, this is what you need to be aware of.

RAHUL KRISHNAN: That is correct. That's right. So there's a lot of diseases, for example, diabetes, which if you were able to predict early, you can have a clinician prescribe early interventions. So things like exercise, modifications to diet that might significantly reduce an individual's downstream likelihood of developing more severe forms of the disease. And so from a public health standpoint, having the ability to have good predictive models deployed at scale might actually help individuals make better downstream decisions about their health.

MG: What are the diseases that are kind of like the Holy Grail when it comes to using AI to get closer to a treatment of or a cure for in an aspirational kind of way?

RAHUL KRISHNAN: I think, you know, there's a lot of, so AI in medicine is a topic of huge interest across the board in medicine. And if you were to go to some of the big clinical conferences around North America, what you would find is that each clinical subspecialty is incredibly interested in using this tool to accelerate both the prediction, diagnosis and treatment recommendation. I'd like to point out that I think the clinical subdomain where AI has sort of seen the most rapid translation has been radiology. And so some of the earliest FDA approved tools for AI in medicine have been in the context of radiology.

MG: And looking at images and decoding what's happening in the image.

RAHUL KRISHNAN: That's exactly right. So this machine learning model would take and input the radiology image and would put out a series of numbers that correspond to either severity or the prognostic indication for what might happen to a patient in the future. And a clinician might use that as an assistive tool to aid in writing a radiologist report.

MG: The concern initially was that that would be the end of the radiologists. We don't need them anymore because the AI will do it for us. John was saying that in some ways, the work that he is doing in his lab is the AI is an assistant to what it is that they're doing. It's not a replacement for what it is that they're doing. Are the health care worker jobs that we see now in radiology, are they, you know, in jeopardy? Are they facing a threat because of AI?

RAHUL KRISHNAN: My personal view is that I do not think that they're in jeopardy right now. What I think will happen in the future is that we will have tools to make clinicians’ jobs just a lot easier, but differently. If you have a clinician who is capable of seeing 10 patients now, what would happen in a few years is that they might be able to see 100 patients because what the AI might be able to do is automate away a lot of the simplistic cases that they can sort of check mark and approve and they can spend their cognitive effort and the cognitive cycles on the much more complex cases that demand their attention.

MG: Tell me about the robot doctors that you say are coming here within a decade.

RAHUL KRISHNAN: So I think that there's a lot of interest, particularly as we've seen the evolution of tools and generative modelling, to think about how we can take all of the clinical knowledge that's been curated through human history and compress them into a machine learning model. And so if you've been following the news in the context of tech and medicine, Microsoft and Epic recently announced a partnership to put GPT-4, which is one of the most advanced large language models within electronic medical records. And what that means is a clinician might soon be able to have access to this chat bot that in theory, they could ask it free form questions. Now, I should say that this is still in trial phases and we haven't yet seen the scientific results of what this is actually going to do and how this might be presented to a clinician. But to think about how this might be used in the future, you can think about it as a compressed store or a repository of clinical knowledge. And if this was done correctly, then it can be a mechanism for clinicians to write free form questions to this machine.

MG: Which would know everything.

RAHUL KRISHNAN: Which would know everything that we know. So this is an important distinction to make. A machine is only so good as the data it's actually trained on. When we train this model on everything that we as humans know, we can ask it to create responses that are in line with the data that it was trained on. It in some sense can often hallucinate, and this is one of the failure modes of large language models.

MG: Just make things up.

RAHUL KRISHNAN: Which is just make things up. And obviously, that is a huge concern in the context of health care.

MG: There are also issues of bias, right? I mean, that you're only as good as the information that you put in. If there is a bias that is inherent in that information, that's going to come out in what the machine spits back.

RAHUL KRISHNAN: That's precisely right, Matt. So bias is another big concern that people are incredibly worried about because we know from a lot of studies that have been done over the decades that the health care system that we have in North America is incredibly in some ways unfair. It is not equitable across the board, and those inequities are often translated into the data that are then fed into these algorithms. And if not corrected for at the point of training, these biases could get encoded into the algorithm and every subsequent output that they put out. And so I think it's incredibly important to be both aware of the consequences in terms of the pros of using this technology in terms of accelerating the ways by which clinicians can receive decisions faster, but at the same time, be cognisant of some of the limitations of these tools and know when not to use them.

MG: Do you think we’ll recognize the health care system in five years based on this technology?

RAHUL KRISHNAN: You know, I think that the health care system is one of the systems that is slow to change. And I suspect that a lot of the advances that we're seeing in machine learning will take a bit of time to trickle into the health care systems. So in five years, I think we'll still recognize it. Perhaps in 10 or 15, it might be another story.

MG: It's exciting.

RAHUL KRISHNAN: It is exciting.

MG: It's quite something. Rahul, thank you very much.

RAHUL KRISHNAN: Thanks so much.

MG: Rahul Krishnan is an assistant professor at the University of Toronto, researcher investigating the future of AI in medicine. He was here in our Toronto studio. That's The Current for this Monday. Q is up next on CBC Radio One. We are going to leave you today by remembering a Canadian theatre giant. The director, Daniel Brooks, died last week after being diagnosed with lung cancer. He was 64. He was a titan in Canadian theatre nominated for Governor General's award, won the first Simunovic prize for his decades long contribution to Canadian theatre. He influenced generations of people who work on the stage in Canada. And he worked with some incredible artists, including the songwriter Justin Rutledge. Justin Rutledge was initially asked to score Daniel Brooks version, the stage version of Michael Ondaatje's novel, Divisadero. Brooks said as they were working through this, the story goes that he needed another male character. Do you act? He asked Justin Rutledge. Justin said no, but he cast him anyway and he performed on stage in this production. It was beautiful and performed a number of songs as well, including this one. And we'll leave you with this this morning in memory of the great Daniel Brooks. It's Justin Rutledge and Lay Me Down Sweet Jesus. I'm Matt Galloway. Thanks for listening to The Current. I'll see you tomorrow.

[Song: “Lay Me Down Sweet Jesus” by Justin Rutledge]

Monday May 29, 2023 Episode Transcript

Social Sharing