Optimized Caffeine Consumption with 2B-Alert

Welcome to the Data Science Ethics Podcast. My name is Lexy and I’m your host. This podcast is free and independent thanks to member contributions. You can help by signing up to support us at datascienceethics.com. For just $5 per month, you’ll get access to the members only podcast, Data Science Ethics in Pop Culture. At the $10 per month level, you will also be able to attend live chats and debates with Marie and I. Plus you’ll be helping us to deliver more and better content. Now on with the show.

Lexy: Hello everyone and welcome back to the data science ethics podcast. This is Lexy Kassan

Marie: and Marie Weber

Lexy: and today we’re going to be talking about one of my favorite topics – coffee!

Marie: And caffeine.

Lexy: So excited. The article that we are going to link this time is a study that was conducted by the U s army to try to identify an algorithm that could predict the optimal quantity and timing of caffeine for service men and women to stay at their peak alertness despite their lack of sleep. And now they have released this tool as a web tool to the public. I am so excited about this. It’s a fun tool to play with. We will link to it. What they did is they studied the timing and the quantity of caffeine across a range of beverages to determine the impairment level of those consuming it.

Marie: And so the impairment level of somebody consuming it when they’ve also been sleep deprived. So how could that caffeine help reduce their impairment versus just being sleep deprived and struggling through it?

Lexy: Yup. And it’s a bit of both. They do have the ability to predict the impairment level if you are not sleep deprived. Correct. And you don’t have caffeine if you’re not sleep deprived and you do have caffeine and all the permutations. The only factors that this tool takes into consideration is the amount and timing of your sleep and the amount of caffeine that you consume. And when the algorithm was specifically being assessed with soldiers who were in training, they were looking originally at licensing the technology for uses outside the military and now they’ve released it into the public. So what they found is they can improve alertness by 64% while consuming the same total amount of caffeine or alternatively a subject could reduce caffeine consumption by up to 65% and still achieve equivalent improvements in alertness.

Lexy: We played around with this.

Marie: Of course we did.

Lexy: Because of course we did and we had some really interesting conclusions coming out of our experimentation with this tool and of course some questions.

Marie: Absolutely.

Lexy: The first thing to note is if you do try this tool, the alertness level that it shows you is actually an alertness impairment and it compares your alertness impairment to a blood alcohol content of 0.06% which somebody would be intoxicated at that point. So if you pass that line, you basically are acting as impaired as if you had been consuming alcohol. And what we found is that each of us had our own times when we were at or above that level of impairment

Marie: just because of sleep schedules, work schedules, dogs, schedules, travel schedules…

Lexy: Yep. All of the above. And we found that if you put it in your naps, it seems to help. If you put it in your caffeine, obviously it shows a decrease temporarily. One of the more interesting things I found is that my two o’clock or so need for caffeine is justified because all of a sudden it looks like the impairment increases to above where it kind of peters out. So over the course of the day it has this kind of sine wave curve going on where earlier in the day it shows more impairment and then it drops and then it comes back and so forth. This is very interesting stuff. And of course because they did this based on the military, we have a lot of questions as to how applicable it is.

Marie: And so this just goes back to how the algorithm was trained in the population that they were training on. Yes. So when it comes to people that are in the military, they usually have more physical activity in terms of their normal job requirements versus what maybe a civilian would normally have. Their age is usually going to be lower versus what maybe an average civilian is going to be. So those are things to keep in mind as you look at this tool that this could be skewed a little bit towards certain populations.

Lexy: Absolutely. The other thing that came to mind is not only the activity level but also the nutrition in the military. They are given specific types of nutrients. They are given a different balance of nutrients. And so if you’re maybe eating more sugar and having more kind of peaks and valleys of sugar that can have some impacts. I mean the human body is a complex system. So sleep and caffeine are not the only variables but it’s still fun one to play with.

Marie: Yeah. And apparently they also have access to something called military energy gum.

Lexy: I want to know what that is.

Marie: And another thing that Lexy was thinking about as she was looking at her examples is that she knows that she needs caffeine because otherwise what happens, Lexy?

Lexy: Headaches and a lot of grumpiness.

Marie: So even if the, even if this algorithm comes up with an optimal schedule for her in terms of caffeine consumption, she’s probably still going to add in a cup of coffee a day. Just to maintain her current caffeine dosage?

Lexy: Yeah, that’s a good point.

Lexy: So, there are two different functions within this tool. The first is to show you your impairment level over the course of whatever days you entered in. The other is to predict your optimal caffeine intake in order to maintain peak awareness levels. So there is a tool in there where you can assign when you want to be awake and it will tell you when and how much caffeine to drink

Marie: when you want to be most alert.

Lexy: True – when you want to be most alert. Which… Maybe I was a little over ambitious because I put in like 12 hour stretches and I don’t think that it’s meant for that necessarily, but the example that they gave in one of the articles that says, “for example, if you pull an all nighter, you need to be at peak alertness between 9:00 AM and 5:00 PM and you desire to consume as little caffeine as possible when and how much caffeine should you consume? That’s the type of question that this 2B-Alert algorithm was designed to answer.”

Lexy: I put in an extended period of time, not nine, five, it was more like nine to nine or so in there. It has a maximum amount of caffeine and a desired alertness impairment, so if you use the number of milliseconds of reaction time, it gives you a different level based on what you’ve chosen. The other thing that it has though is it will show you smaller cups of coffee and bigger cups of coffee, so it gives you the amount as well as the timing because there’s a maximum amount of caffeine but not a minimum amount of caffeine that you can set as a parameter. It doesn’t give you the option like you were saying, to specify that you need a cup of coffee. Otherwise bad things happen because it’s a physical addiction and you can have withdrawal symptoms.

Marie: I have a brother who’s in the military and this is the cycle that he goes through. He will consume caffeine and then that won’t do as much for him. So he’ll need to consume more caffeine. And then you’ll get to the point where he is consuming, let’s say multiple Red Bulls and it’s not doing anything for him. And he’s like, Oh, I’ve reached my capacity where caffeine no longer works for me. So then he’ll take a weekend where he touches no caffeine and he is off caffeine for like a week. And then he goes back to caffeine and he can drink a red bull and actually has an effect for him.

Lexy: Wait, so, so you mean to tell me the caffeine works like a drug and that over time if you continue to use it, you may need more of it to get the same effect?

Marie: Yes.

Lexy: Say it isn’t so!

Marie: No, I’m going to say it is. Here’s the other example. I have my dad. Big coffee drinker. And one year for lent he gave up coffee.

Lexy: Oh that had to be painful.

Marie: He went through big withdrawals.

Lexy: Oh Gosh.

Marie: Headaches and everything.

Lexy: Been there, done that. Not Fun.

Marie: Yeah. So caffeine is a drug. Yeah, it’s on the flip side. I usually don’t consume that much caffeine and my brother looks at me like I’m an idiot. He’s like, caffeine is what has built the world. Why aren’t you on caffeine? I’m like, cause I sleep and I don’t need it.

Lexy: Some people just don’t have the benefit of having enough time to sleep, which is why this algorithm was developed in the first place because sometimes people in the military are not given a sufficient time to sleep or are woken by circumstances beyond their control and require caffeine in order to function optimally.

Marie: Yeah, and when there’s times when they need to be on duty and they’re only getting four hours of sleep or they need to be up for 36 hours at a time, whatever the the situation demands.

Lexy: Yeah, exactly.

Marie: There is a time and a place for it for sure.

Lexy: Exactly.

Marie: So there are different ways that people deal with their caffeine consumption. Sometimes you try to maintain it and sometimes you do want to decrease it or maybe you want to eliminate caffeine. This tool could also help you see how to do that with maybe a more regular sleep schedule or with maybe being able to incorporate naps into your schedule. Or what was really interesting as we saw that over time, the more days that you kind of go with sleep deprivation, it shows gradually how your response time degrades. And it also shows that if you’re able to increase your sleep over time, how your response time can improve. So it could also help you determine like, do I have enough sleep banked where I can kind of skimp on hours for this night and be okay? Or, you know, do I need to get into a more regular pattern? So we’ve only started playing with this, we already have multiple more ideas of things that we want to like put in and test and see what it comes back with in terms of recommendations. So we encourage all of you to jump onto the tool and test out the algorithms yourself and see what you find.

Lexy: From an ethics standpoint – since we have to tie it in somehow apart from the fact that she’s really fun – this does kind of bring to mind how are people likely to use this? Like if I were to think about this from an almost anticipate adversaries perspective, here I am and I’m basically doing this saying, hmm, this justifies my second cup of coffee for the day. At what point do you have to say, “Umm please seek medical attention for your addiction to caffeine, Lexy because you have a problem.”

Marie: Or you have a company that maybe says, oh, this shows us that… and again, there’s other science that goes to back up what this tool is showing, but you could have somebody like 5 Hour Energy say, Hey, this is when you should take your 5 Hour Energy shot and incorporate it into an app.

Lexy: Mmhmm. I personally would love to see this be a justification for the adoption of Siesta across the world.

Marie: I agree

Lexy: Because it seems to show that the early afternoon is when we’re more impaired, so clearly a nap is warranted right after lunch.

Marie: We should all adopt this great Spanish tradition.

Lexy: Way to go Spain, you figured it out.

Lexy: Having access to these tools and without it being a supervised kind of medically-warranted assessment is a little bit dangerous. Maybe not as much so as other things we’ve talked about it. It’s much more fun, but certainly something that should be taken with some caution. Don’t necessarily assume that this is a prescription for caffeine.

Marie: Oh, absolutely not. And I think if anything, this is a prescription… This shows that how important sleep is.

Lexy: Yes.

Marie: And it shows that you can work around that with caffeine, but that isn’t necessarily how you’d want to just do things by default all the time.

Lexy: Yeah.

Marie: There’s no replacement for sleep in the long run.

Lexy: True.

Lexy: The other thing that kind of was questionable to me here is it showed, as we said, we tinkered around with this a little bit and it’s very common that during the week is much busier. You don’t get as much sleep and then on the weekend you catch up.

Marie: Yeah.

Lexy: And I’ve heard multiple sleep experts say that there’s no such thing as catching up, but this kind of shows differently. Like it shows that as you were saying over the course of multiple days of not getting sufficient sleep, you will get progressively worse. But then if you get a good long nights rest, your impairment drops considerably. Admittedly, that’s not the only factor as to what sleep is doing for you, so your body isn’t recovering potentially in different ways, but you’re able to stay a little bit more alert by having that longer sleep. So that idea of catching up may just be our getting more alert even though the rest of our physiology is not being as impacted, like the way that you would get a regular full night’s sleep.

Marie: Good point. And the other thing that I think about because I have multiple people in my life that I talk about caffeine with because there’s coffee, there’s chocolate, there’s tea, there’s soda, there’s caffeinated drinks, you know there, there’s all these different things out there and there is this idea that you can use caffeine to perform a little bit better, to work a little bit harder. The flip side to that is should you set yourself up so you don’t need to use caffeine? And you can have a more regular sleep cycle and be more well rested and be more alert on your own. So what I find interesting about this app is there are going to be people that look at this and say, “but I’m a student. I need to get through these four years of college. And so I’m going to use caffeine to get through it regardless. Like there’s things that I got do, there’s stuff that I have to accomplish, this is how I’m going to do it.” And there is an argument for saying that people that use caffeine can be more productive and in a way this tool shows that. But I think there’s a question of, you know, should we base what we expect people to be able to do on a consistent basis on them being highly caffeinated or not?

Lexy: Yeah. That’s a much broader topic as to what do we value as a society. Productivity is definitely up there, but it’s not the only thing that, that we need in our society. And so making sure that people are healthy and have that productivity for longer is something that we’ve talked about previously. And certainly has an impact here. Obviously you don’t want to try and caffeinate yourself to the nth degree and wind up with heart palpitations or something like that. It’s very true. There is a maximum and that part of why they put in what is your desired maximum amount of caffeine per day just to make sure that you’re not going above some of those levels. So some other studies that have come out have said up to it think it’s like four or five cups of coffee a day is good for you. And then beyond that it starts to get really problematic. Definitely look at all the research, definitely do your homework. This is not the only thing out there. It’s not the only study that’s been done, but they have a fun web tool to use. So it’s a fun one to play with.

Marie: They do. And in terms of the data science ethics, I think that’s a theme that we kind of keep coming back to where if you look at one or two parameters, there is a way to optimize performance to get to that one or two parameters situation. But when you start looking at the broader context around things, that’s where the balance of more consideration comes into play and that’s where the the questions that we keep coming across pop up.

Lexy: Agreed. Good reminder to consider the context that you’re working with them.

Marie: Very true.

Lexy: Thanks so much for joining us on this episode of the Data Science Ethics podcast. I hope you get into to be alert and play around with it for yourself. Let us know what you find.

Marie: And the next time you have a caffeinated beverage you can think of us. So cheers to that.

Lexy: This has been Lexy Kassan

Marie: and Marie Weber.

Lexy: Thanks so much.

Marie: Bye.

We hope you’ve enjoyed listening to this episode of the Data Science Ethics podcast. If you have, please like and subscribe via your favorite podcast App. Also, please consider supporting us for just $5 per month. You can help us deliver more and better content.

Join in the conversation at datascienceethics.com, or on Facebook and Twitter at @DSEthics where we’re discussing model behavior. See you next time.

This podcast is copyright Alexis Kassan. All rights reserved. Music for this podcast is by DJ Shahmoney. Find him on Soundcloud or YouTube as DJShahMoneyBeatz.