AI is revolutionising stroke care [300%]

Study finds tripling of patients recovering

Hi there

Here is the latest edition of Practical AI, the newsletter that introduces you to your future today.

Last week I said that this week was going to be about voice AI. But some amazing medical AI research was released on Tuesday, and I just had to share!

As someone who has had friends and family members suffer from a brain stroke, this hits home with me…

So let’s start today’s newsletter with amazing AI news directly from Oxford, UK.

Brainomix’s AI helps triple recovery from stroke

This week (Tuesday), new research was released from the UK showing that Brainomix’s e-Stroke suite of tools can triple patients recovering from a stroke.

The study found that the number of people able to perform daily activities after a stroke increased from 16% to 48%, thanks to the AI from Brainomix.

So what is e-Stroke?

It is basically a collection of tools that uses state-of-the-art AI to support doctors with the real-time interpretation of brain scans.

As someone who has watched his mother getting a stroke in real-time, I welcome any development that can help doctors better understand the stroke and shorten the time to treatment.

A brain stroke is a time-sensitive injury, and anything that can shorten the diagnosis of the damage can literally be the difference between life or death or the quality of life after.

As much as I love playing around with ChatGPT and MidJourney, the medical field is one of the biggest areas where AI can make a difference - in my humble opinion.

Yes, AI will disrupt many industries; people will lose their jobs, some jobs will become obsolete, new ones will be created, etc.

But AI can and will provide us with many amazing opportunities and aid humanity.

This medical use case is, in many ways, on the opposite side of last week's newsletter on the darker sides of AI (Deep Fakes).

Applications and implications of Brainomix’s AI

I think the application of Brainomix’s AI is pretty obvious.

And it is easy to imagine other areas where AI can be used similarly. Cancer, heart issues, lung issues, brain issues, thyroids, joints, ligaments, etc. Anything that involves a scan, basically.

Brainomix is working on solutions for cancer and Interstitial Lung Disease (ILD) scans.

Although I do not know of any other similar companies when writing this, I am pretty certain that Brainomix is not the only one working on this.

And this makes me hopeful for the future and for the future of AI.

I imagine AI can revolutionize the early detection of a range of diseases humanity suffers from (note: I found a story about early detection using AI while finishing up the newsletter, so I included it in the Quick Bites section further down).

Let's not forget the time savings.

The world has too few doctors and medical specialists.

AI may become a valuable assistant that increases productivity without sacrificing care or increasing workload stress for medical professionals.

But could this AI have a dark side? 

I guess one of the “misuses,” off the top of my head, could be a cynical sports club using AI to analyze new soccer/football/baseball/basketball players' health to get statistical probabilities of the athlete having "XYZ" issues during their contract with the club, thus misusing this information in negotiations, trades, or similar…

Any others?


In 2023 I will dive deeper into AI in the medical field and report my findings to you. Maybe it should even be a regular section in the newsletter.

What do you think?


If you are interested in reading more about how NHS in the UK use e-Stroke and their experience with Brainomix, visit their site here.

AI experiment of the week

This week, due to the Christmas holiday, I have not played around as much with AI tools as I usually would. Thus, I have not published any new articles on any of the blogs.

However, a few weeks ago, a new AI tool called Ellie AI was launched, and it is intriguing.

Ellie AI is an email assistant that can help you reply to emails faster.

As I understand it, right now, it is an extension you can use with webmail/online email and your web browser. It does not work inside your desktop email reader (like Apple Mail or MS Outlook).

However, you can use the pop-up function I used (picture below) to generate replies. But I assume it would be hard to train the AI with such a use case (?).

I somehow did not get Ellie to work inside my Gmail, but that might have just been me not following the instructions on their help page (here).

I just installed it and tried it out :) And it worked...

The extension works with Google Chrome and Firefox.

This is how it looks in my Chrome browser:

The tool uses the AI from OpenAI (GPT-3), the company behind the now very “famous” and popular ChatGPT.

To try it out, I simply opened last week’s Practical AI newsletter and wrote a reply. I used the “Casual + Interested” and the “Annoyed + Not interested” combo.

Here is my first output from EllieAI (Casual + Interested):

I love this reply!

In many ways, it is an even better reply than I would have written myself. And I love how it asks me a question about any other applications of AI I have been reading up on.

One word: impressed!

Here is my second reply (Annoyed + Not interested):

This is a good answer, but I do not love it the same way as the first one.

It implies I did not sign up for the email in the first place.

And what if I want to stay a subscriber?

I could, of course, adjust the answer before replying. And I assume after some training, the AI would not unsubscribe me everywhere :)

If you want to have a quick look at how I used Ellie AI, here is a screen recording of me trying it out:

I only have the free version and can only use it twice before I need to upgrade to a paid plan.

The developers, Danielle and James, have come up with a very cool AI use case, and I find the tool useful.

I can easily see how you can save time replying to less important emails with this AI tool.

If you want to take it for a spin, you can try it out for free on the Ellie AI webpage.

Right now, there is a 40% discount for the first 100 paying customers.

However, according to co-founder Danielle’s Twitter feed, the discount will be good until the end of the year. So I guess you have three more days if you want to grab the discount.

As of right now, Ellie AI has been downloaded 5893 times.

A new AI-powered search engine is taking on Google

Last year the search engine launched.

Honestly, I have never heard about them before.

I guess it is hard to compete with the search engine that holds the number one spot with over 93% market share worldwide…

However, on Friday, the 23rd of December, became the first (to my knowledge) search engine that incorporates AI and chat-style functions, making it able to answer questions and hold a conversation, much like ChatGPT from OpenAI.

It is currently in Beta, so it is also suffering from wrong answers like ChatGPT. But at least it is internet-connected and can serve up recent happenings, such as the soccer world cup results, etc.

Its search data comes from the largest search engine you have never heard of... Bing.

It is fun to watch the race to take on Google.

Do you think, OpenAI or any AI or tech will be able to de-throne Google?

Hit reply, and let me know your thoughts!

AI image of the week: “Harry Potter style library”

This image was generated using the AI tool MidJourney. This tool is a bit more complicated, as you need to do your prompts using Discord.

I have barely started experimenting with MidJourney, but you can expect more of this style of AI images in the future :)

Is there anything you want me to try to create for the next newsletter? Just reply to this email.

Quick bites - interesting articles on AI

AI definition of the week: Expert Systems

An expert system is a type of artificial intelligence (AI) that is designed to mimic the decision-making ability of a human expert in a specific domain. It uses a combination of machine learning algorithms and a set of rules or heuristics to make decisions and provide recommendations based on input data. Expert systems are often used in industries where there is a need for decision-making based on complex and specialized knowledge, and they can help to reduce the time and effort required for decision-making.

(This was written by ChatGPT - I asked for a simplified definition)

That’s it for the last edition of the Practical AI newsletter in 2022.

Thank you all for reading my ramblings on AI.

Happy New Year!

If you have any questions about AI or any feedback, just hit reply or tweet me @thomassorheim 


PS! Knock knock.


Who’s there?




AI who?


AI’ll be back!

This is not the end, it is where the fun begins!


or to participate.