Category: Artificial Intelligence in clinical research

Artificial Intelligence in clinical research. Weekly tips about how Artificial Intelligence support evidence generation in healthcare. The generation of RWE at multisite level using Natural Language Processing on Electronic Medical Records. Our team is now combining data scientists and clinicians with experience in clinical and translational research (lead by oncologists).

To really add impact, this is the way of Artificial Intelligence in clinical research:

  • Sometimes the simplest way can also be the best one.
  • And the best way is not a cut of a database, nor a group of preselected variables, nor a certain number of patients. It’s not that. And neither a very costly registry.
  • The best way we can imagine is to retrieve the actual complete information about what is happening at the points of care.
  • It’s having access to the complete medical records information.
  • In order to get this, you basically need:
    • A combined team of data scientists and clinicians with experience in research (in our case, lead by oncologists).
    • A system able to retrieve information from any healthcare provider (as long as they have electronic medical records -EMR-, paper doesn’t work).
    • Natural Language Processing, because 80% of the variables and outcomes are going to be in the clinical narratives’ free text.

Father of AI says that in health, almost everything is good.

Ignacio H. Medrano, founder and Chief Medical Officer:


LinkedIn is an amazing place. Oh, if I didn’t depend on it to sell projects, you can bet I would not be there…

Professional gossip, cheap advice. What a place.


Once, I was listening to the TV in the background, and they were in a debate with famous cinema directors and TV executives.

That was when Netflix was starting to emerge.


They were saying that platforms were a scam and that the good stuff was theirs. Traditional TV and cinema. That the movies and series on the platforms would never win anything.

In other words, they were saying, ‘My goodness, how scared I am.’


That’s the same thing I saw on LinkedIn the other day.

It was a head of a CRO saying, ‘Neither AI nor big data nor machine learning… what’s important are people and their passion’ (or something like that).

Poor guy.

I’m not laughing at him at all.

I would never laugh at someone who is having a hard time.

Unless it was a genocidal maniac on the scale of Putin.

We all have fears and miseries. I’m the first.


But saying that…

Knowing the absolutely brutal socio-economic revolution that is brewing at lightning speed, by the hands of AI.

Which is going to make the Industrial Revolution look like child’s play.


The good, the beautiful thing about AI in health is that it is one of the few areas where almost everything it is going to bring is positive.

It’s not like in other internet worlds.

I’m not the one saying it. Geoffrey Hinton, the father of Google’s Machine Learning, recently told Topol in an interview.

Almost everything, good.


At Savana, we use AI for good things.

That doesn’t mean it doesn’t have its effects.

When we manage to generate disease records without a single human having to fill out variables by hand, in general, it is good. Although I acknowledge that it messes up others.


If you want to create a database of a disease using AI, this is the place.

What healthcare AI course to take?

Ignacio H. Medrano, founder and Chief Medical Officer:


I have a friend named Nuria Gago.

She called me a while ago because she was writing a book, and there was a character with Alzheimer’s in it, and she wanted me to tell her about it.

I told her.

And she included some literal phrases, which is something curious to read later when you see it.

She won the Planeta Prize, the most prestigious literature one in the hispanic world.


So I asked her the typical stupid question, whose answer I already suspected:

– Nurigago, what writing course do you recommend to improve my writing?

– None. Just start writing.


Always that response…

Ask a good fisherman what fishing course he recommends. You’ll see what he says.

Or ask Michael Jordan what basketball course he recommends.

Or ask Dabiz Muñoz, the best cook in the world, what cooking course he recommends. You’ll see…

Or ask the best dad you know what course to be a father he recommends.


I’m often asked about a course on AI and healthcare.

Sometimes I recommend one out of obligation, but deep down, I think: none. Start a project.


And if not, ask Elon Musk which MBA he recommends.

Well, I think the idea is clear.


Courses are good for silencing consciences and procrastinating.

For learning, do.


And for an AI project, you can do it here.


P.S. I think very short courses, of a few hours, to provide context, to know if you like something and where to start, that’s what really makes sense.

Why AI is going to kill Google.

Ignacio H. Medrano, founder and Chief Medical Officer:


Google is going to die.
I think.
Most likely, you also thought about it.
I have been thinking about it for years.
How many more years will Google last?
I didn’t have an answer.
But I got it within 2.5 minutes of using ChatGPT.
So I bought Microsoft stocks.
They have appreciated by 60% to this day.
I bought a small amount, so it doesn’t change much for me.
But I don’t do these things to make money, but rather to make my accurate (and inaccurate) predictions tangible.
That’s why I bought Bitcoin in 2014.
But wait, I’m getting off track.
I was saying that Google is going to die, and not because OpenAI’s GPT-3 model is more powerful than Google’s BERT.
But because whoever imposes their AI, this new way of searching for information will end Google’s model based on clicks.
That’s why its end is approaching, either at the hands of another or by cannibalizing its own product.
How many smart people will be there?
People who know technological disruption perfectly well.
They have read all the books about Kodak, Motorola, and such.
And still, boom. Six feet under.
And do you know why I think it is?
Because the ability to withstand disruption does not correlate with intelligence. Not with knowledge.
It correlates with humility.
It’s an attitude.
That’s why when AI sweeps away the old way of generating clinical evidence, when normal data collection, normal statistical analyses no longer matter… the biggest won’t survive.
Not the smartest.
Not those with a higher H-index.
The humblest will survive.
Those who accept reality and quickly adapt to it.
In fact, those who have already started doing so.
And if you’re reading these emails, it’s very likely that you are one of them.


The error of patenting an AI algorithm.

Ignacio H. Medrano, founder and Chief Medical Officer:


When I was a resident, I had a very smart and funny colleague (who later coincidentally worked at Savana for a while) who talked about the “circle of death.” 
She referred to iatrogenesis. 
To patients who were relatively well, but due to following “the protocol,” we would admit them or send them to the ICU, triggering a catastrophe. 
I thought about it recently when a relative of mine was found to have asymptomatic metastases from a kidney primary tumor he had 20 years ago! 
So, I informed my brother: 
“He has an incidental finding. Since he’s 80 years old, the only sensible thing would be to leave him be. But as Medicine is what it is, and the healthcare system is what it is, they will undoubtedly give him immunotherapy. It is highly likely (I said ‘highly likely’) that he will experience serious adverse effects, complications will arise, and that may be what kills him. It’s called the ‘circle of death.’ That’s just the way it is, and we can’t do anything about it (this family member would never prioritize my opinion over that of his oncologist).” 
He had a heart attack after the first dose. 
It could have not happened, but it’s a matter of probability. 
And I have no clue about Oncology or Cardiology. 
But I believe in the data. 
And I think about the missing data. 
The countless number of cases like this that will happen every day. 
And which are not documented anywhere. 
Simply because certain administrations haven’t taken action. 
The problem is not technological. 
It’s not about the budget. 
It’s about taking action. 
I enjoy writing for people who like to take action. 
Those who respond to these emails saying, “I’ve been wanting to connect A and B for a while. Can I do it with Savana?” 

What to do with AI-intruders.

Listen to what Matilde Sánchez Conde, an Infectious Disease physician tells me in an email:

It bothers me that I think I’ve been a little ahead of all this, wanting to do things like defining the fragility model for more than 10 years now…it’s now the bandwagon everyone’s jumping on, and despite having a clear understanding of it from the beginning, I’ve never been able to get funding for anything in this line. At first, nobody understood it, and now everyone’s an expert.


That’s right, dear internet friend.

Matilde is suffering… (drumroll…) the innovator’s dilemma.

There’s a very famous book on the subject.


But this book can be summarized quickly.


The idea is that when you come with something very new, nobody will pay attention.

And, as technology is exponential, when it explodes, it will do so suddenly.

And then everyone is into it, talking about it, and acting like an expert.


And that poor innovator, ignored for a decade, feels frustrated and powerless because their ahead-of-their-time initiative is diluted in the excited mass.


If you’re reading these emails, it’s probably happened to you.

And it may have happened to you with AI.


Well, I’ll tell you something. And you know I’m not one to sugarcoat things.

It’s not a big deal.

Nothing at all. Just relax.


Because those who are jumping on the AI trend, but were in the Metaverse yesterday and will be in genomics-at-home tomorrow (which is probably the next big thing), are harmless.

Because their manifest lack of depth won’t take them anywhere.

To do an AI project that’s worthwhile, it’s not enough to copy and paste ideas.

You have to understand what’s going on behind the scenes.

Why it’s happening.


Why one algorithm is more relevant than another.

Or how often it needs to be retrained.

Or if the question is clustering or stratifying the risk.


You don’t acquire that knowledge by reading news.

You learn it over the years, by proposing studies and making mistakes.

That will protect you from any intruder.


Just ask three questions, three, and you’ll know if the person behind it really knows why machine learning is completely changing medical research.

And that, precisely that, is what we’re going to take away when we do a project together here.


What are the 3 questions? Here.

Complete the info, and a KAM will contact you ASAP:

Want to use it?:

Start with your proposed AI + RWE use case:

This is the first step for AI + RWE: