The Greatest threat to human survival…is us.

Aditya Dewan
11 min readDec 20, 2020

We’re the most powerful creatures to exist on this planet.

Sounds blatant, but we’ve literally hacked nature itself to do impossible things.

Can’t fly? No problem! Airplane tickets are as low as a hundred dollars.

Tired of your planet? Don’t worry! We can launch ourselves into orbit with giant metal missiles!

We truly fail to understand just how far we’ve come.

In literally just the span of 100 years — we went from dying of colds, to mastering disease.

It’s incredible to see human evolution — we’ve invented tools like the wheel, factories, personal computing, the internet, and millions of others.

But the tool that’s most important of all isn’t electricity (though that’s amazing) — it’s the ability to create more things. It’s a meta-tool, the one-size-fits-all shoes at the back of the store. We aren’t really amazing at one thing, but pretty good at multiple.

And so far, this “general purpose” solution has worked pretty well for us. We don’t know everything (not even close) — but we know just enough to explore and gain better understanding about those things.

Here’s the best part: this general-purpose solution comes in one package — a lifetime, non-expiring Amazon Prime membership. Except, without a return policy. Kind of important. Once our packages are gone, they’re gone for good. No arguing with customer service for this one!

Luckily, our packages have their own defense system — reflexes deterring threats you can see, and a tiny army guarding against the ones you can’t.

Really — it’s the best package we could possibly own.

But there’s more. These systems have TONNES of limitations — and pretty big ones at that.

Let me explain.

Mother nature ISN’T peaceful. 🌳

Every second of your existence, someone’s trying to take it all away from you. And no, it’s not your friend Joe.

It’s nature itself.

Yep, you read that right. Trillions of tiny, little critters would LOVE to make you their new home — real estate agent or not. They’re shrewd too — multiplying by the hundreds of thousands to make sure it happens.

Don’t worry, it’s nothing personal — just in their nature!

Thankfully your body has its own set of bouncers to keep uninvited guests out.

Actually, that’s not true — your body has an entire military dedicated to wiping these critters out — like the world’s biggest extermination squad. And, it’s amazing at its job.

But guess what? It’s not perfect — sometimes, bacteria and viruses will find a way into your system and mess around with your stuff (please leave my lungs alone).

Now, some of these trespassers just want to play around, causing mild symptoms (like the annoying sniffles from the cold).

Others are like an entire demolition team — breaking everything down, causing serious and sometimes even fatal symptoms (like pneumonia). Our militaries need reinforcements, bigger arsenals, and better strategy.

So, we didn’t stop there. To help our tiny soldiers fight off their greatest battles, we’ve created new waves of weapons, each deadlier than the last.

But this time, the (massive) arsenal isn’t made of machine guns and tanks.

Nope — it’s equipped with something much, much deadlier.

Antibiotics and medicines.💉

What? How?

Don’t believe me? Just consider this.

There’re literally more bacteria in your tooth than humans in all of history. What then, would happen if you got a tooth infection?

You’d probably go to the doctor, who’d inevitably prescribe you with antibiotics.

Now imagine, what would happen if you took it.

The answer is termination — billions and billions of bacterial cells per use, wiping out entire infections a matter of days.

Fun fact — that’s more than every human ever killed.

Make no mistake — these are amazing inventions. They literally allow us to live longer, healthier, and happier than ever before — keeping our greatest foes at bay.

But they’re also a double-edged sword.

They’re too effective — antibiotics don’t know what they’re killing, terminating both infections + healthy flora and fauna. Antibiotics are rabid killing machines — incapable of awareness or understanding.

And that, is why they need humans — to apply them in the right situations, make the judgement calls, and take the shots. That’s also the problem — make the right decision, and you’ve just saved a life.

Miscalculate, and your patient could die.

You’re in for a surprise.🤯

It’s easy to think that those wrong decisions rarely happen. After all, don’t doctors, nurses, and countless medical staff go through decades of training, tests, and certifications? How then, could this be a widespread problem?

Surely, the numbers can’t be that high…right?

250 000.

That’s the number of deaths caused by clinical misdiagnosis, per year. Just in the U.S, might I add.

30 000.

The number of Americans who die from misdiagnosing infections.

3.

And, it’s third leading cause of death in the U.S — just behind cancer and stroke.

Think about that for a couple minutes.

We’ve accidentally killed almost as many people as the deadliest diseases in history.

In fact — by the time you finished thinking, someone else passed away because of a mistake.

This is just in the states. Imagine, the millions of people dying every year, in places less fortunate, the sheer number of lives lost. All due to human error.

The greatest threat to human survival isn’t some new super virus or disease — it’s us.

What are we going to do?🤔

It’s obvious that we need to solve this problem. But the question is…how?

The first instinct might be to blame the problem on medical staff. Well, that instinct is wrong.

Medical staff are, after all, humans — working in overcrowded hospitals, dealing with dozens of backlogs, medical bills, and sick patients. At that point, it becomes nearly impossible to give 100% attention to each and every patient that walks through your doors.

That’s the best-worst thing about humans — we’re imperfect beings. We make mistakes, adapt, and then learn, becoming better than before.

It’s easy to argue that these mistakes are horrible (and in this scenario, they definitely are), but it misses the full picture, the forest view.

Mistakes themselves aren’t the issue. You’re alive and reading this article because Fleming accidently left a plate a plate of bacteria out while on vacation, coming back to see that a fungus (penicillin) had grown on it. Your instant noodles are “instant” because Spencer made the mistake of taking a chocolate bar next to a microwave generator 😋.

I think you’re getting the point — our society is founded on mistakes, whether we like it or not. Mistakes are fine (or even beneficial) when they’re harmless, but not when every mistake leads to someone’s death.

Some mistakes are costlier than others.

We can’t allow a quarter of a million lives to be lost each year. But somehow, we do. It’s an unavoidable fate — mistakes will happen, and innocent people will have to pay the price.

There’s no way we can eliminate all mistakes from happening — but we can still minimize them, making them less frequent and less deadly.

Ideally, we would have a no-mistakes-diagnosis-system — diagnosing bacteria quickly, cheaply, and correctly. Basically, something that would ctrl-c the expertise of doctors, and ctrl-a-del their mistakes.

A system like that would be incredible. Just give an is a picture of a bacterium, get back its exact species, and boom! Bye bye misdiagnoses!

Yeah…no.

As you probably guessed, life isn’t that easy. Classifying bacteria is like trying to find a colour-changing polar bear in a snowstorm — some bacteria are like bright-red animals, easy to spot and identify. Others are completely white — way harder to put a finger on. For these ones, it takes weeks of tests and thousands of dollars to understand what in the world is going on.

What’s missing is consistency — a reliable way to understand the critters on sight, regardless of what it is.

And, preferably without breaking the bank.

Here comes the best part — we can actually make it. Right now, with the tech we have available. “But Aditya, wouldn’t we need some next level, advanced quantum physics generator to make something like that???”

Actually, no — the problem is pretty simple. And, so is the solution…kind of.

But it may not be what you expect.

The Robots are coming.🤖

And they’ll be here soon. But they’re not coming for you or your job…probably😅.

They’re coming for human error in hospitals.

Right now, chances are that you’re reading this article on your phone. If you are, you probably have a social media app on it too (which you’re probably not opening, because then you’ll have to respond to all your hundreds of unread messages).

These apps all have facial filters — and they’re good too. At the click of a button, you can turn yourself into a dog, cat, or whatever new animals they’ve come up with in the last 6 months. Really, the things they can do are nothing short of amazing — tracking your face, your expressions, applying filters, and then letting you share your pictures. All from the tiny camera perched on top of your phone!

The way they do this is via AI — an amazing technology that uses algorithms to mimic human intelligence. I won’t be explaining it in too much detail here (here’s an article if you want to better understand the tech behind the talk), but basically — it learns from past mistakes and gets better over time.

Cool, right? The point is to adapt to new, unseen situations — like being able to tell a random cat from a dog, a person from a fire hydrant, etc. — without having seen them before.

So, how does that help us? It’s not like recognizing cats and dogs is going to save lives (unless you’re a dog person like me, and love looking at dog photos — sorry cat people).

Here’s the kicker — if AI can understand differences between animals…could it find differences in bacteria? It sounds like a stretch — but it technically, it could be done. The system is the same — we just train and test it on bacteria, instead of adorable kittens and puppies!

Wait — we can’t just shove a bacterium into an AI algorithm. What’s the plan here?

While it struggles with running lab tests and experiments (the ways we usually classify bacteria), AI is AMAZING at understanding images. It’s achieved insane accuracies on small problems (like a specialist with 10 Ph.D.’s)

I think you can see where this is going. We can feed our algorithm microscopic images of bacteria — and then get it to classify them. It sounds so obvious in retrospect, but it makes perfect sense — get microscopic images of bacteria, make an algorithm, train algorithm on said images, evaluate it, and BOOM! No more misdiagnosis.

Unfortunately for us, it’s not that simple.

AI, like humans, isn’t perfect. It might not be as inaccurate, but it will still mistakes. And here, the mistakes aren’t Whoops, I accidentally thought that dog was a cat — the mistakes could be the difference between life and death. Kind of a big deal.

So, we’re going to have to make our algorithm more accurate. Which means more data, more images per bacterium, more time spent training, and before you know it this becomes the most expensive project on the planet. Just classifying cats and dogs takes thousands of images.

So then — how on Earth are we going to classify the (get ready for this) 5 million trillion trillion bacteria on this planet?

5 000 000 000 000 000 000 000 000 000 000

That’s going to require a little more power than your local supercomputer.

Getting this to work means we’re going to have to start…small. Really small (pun intended). Instead of trying to catalogue the entire planet’s worth of bacteria, we can start by just a couple of the most dangerous species. Just doing that could literally save thousands of lives.

Don’t believe me? Two of the biggest culprits of infection misdiagnosis are pneumonia and endocarditis –two species of bacteria, that lead to a huge part of bacterial misdiagnosis. Imagine the sheer number of people we could save — just by classifying these two species.

It’s not just a random person on the internet talking about this either. In fact, scientists are beginning to catch on this idea, developing new solutions to this problem as we speak.

And, these solutions are actually way better than expected.

Think back to before coronavirus, when you could drink coffee from your favourite café. Hopefully it’s not Starbucks, because it’s crazy expensive. In fact, it’s so expensive that a cup of Starbucks coffee is more expensive than classifying bacteria.

Your eyes aren’t playing tricks on you. Scientists can save a life, in 1/100th the price of your favourite latte. And that too, with 92% accuracy on three bacteria types.

That is the true power of AI.

This isn’t the only one either — dozens of papers on the topic have been published in the last 3 years, each reporting incredible accuracies and potential.

As you can see, using AI to prevent misdiagnosis isn’t some fantasy — it actually works.

Why then, is this field so new? AI has been around for more than a decade yet using it to classify microscopic images has really only been around for 2 years. Given that misdiagnosis rates are higher than ever, that’s a huge red flag. Despite AI still making some mistakes, it’s been shown to perform equally well (if not better) than healthcare professionals.

And even if it isn’t better, even if one life can be saved by using this available technology — isn’t it our moral responsibility to do so? Aren’t we obliged to save that one life? The one with goals, dreams, friends, family — people who they cared about?

What we need is not ignorance, but research — to iron out the wrinkles. If we truly want to save millions of lives, then we need to act — to push efficiency, pull down costs, and skyrocket efficiency.

It’s an investment alright — in better systems, access, more training data, and more smart people to take these solutions and 10x them. But it’s one that we’re going to have to take.

So that innocent people don’t die because of human error.

So that we can save 40 000 + people from an early death.

So that we can take accountability for our mistakes and do something about it.

Because the problem is growing — and it’s not slowing down anytime soon.

The solution is here. The technology is here. All we need to do, is to use it. To get this revolutionary tech out where we need it most.

Let’s rid the diseases that haunt us with the medicine of hope, and the cure of action.

Sources:

https://www.hopkinsmedicine.org/news/publications/hopkins_medicine_magazine/medical_rounds/fall-2019/the-big-3-in-diagnostic-errors

https://www.infectioncontroltoday.com/view/researchers-use-artificial-intelligence-identify-bacteria-quickly-and-accurately

--

--

Aditya Dewan

Building companies. Machine Learning Specialist @Actionable.co. Philosophy x Tech.