AI detects skin cancer with 100% accuracy

A study of 22,356 patients

Author

Thomas Sorheim
October 19, 2023

Hey all!

Welcome to all of you who have signed up for this new newsletter aptly named Cancer Innovations.

Since it is the first edition, let me introduce you to who is behind it and why I decided to create this weekly newsletter (I promise, next time, there will be no fluff or big intros like this).

If you just want to read the AI & cancer stuff, scroll down!

I’m Thomas. I’m 48, from Norway. I live in Vilnius, Lithuania, but soon in Prague, Czech Republic. I write another AI newsletter - Practical AI - and run several other small businesses (travel & events and digital media). I have an IT and a business background and have studied in 4 different universities in 3 countries.

Why a newsletter on the intersection of AI and cancer?

And am I qualified to write a newsletter on a medical topic?

First, I am not qualified to write a “medical newsletter.”

But this is not a “medical newsletter” because I will not make any medical analyses or recommendations.

I am not a doctor, and I am not claiming to be one.

This is not medical advice. It is a simple resource. A starting point.

I am merely here to shout to the world about the latest technological advances related to cancer, hoping they could help someone or inform healthcare professionals about the latest developments and link the resources I find.

Thus, the promise of this newsletter is not to educate you on how to cure cancer but to share information about new technology that might help cure cancer.

But I have some experience with cancer. In the last five years, I have watched how aggressive and evil it can be.

I watched my mother die from cancer two years ago.

On Friday, when I officially announced this newsletter on Twitter and LinkedIn, my grandmother was dying of cancer. Two hours after my post, she died.

My sister has breast cancer.

A good friend died from prostate cancer a year ago.

Another “distant” friend died of skin cancer two years ago.

I can go on, but you get the picture…

If not anything else, I have the motivation to shout out loud all cancer innovations I can find and hope that one day, AI will help eliminate this bastard!

Now, with that out of the way, here is how I am planning to do this newsletter.

  1. It will send the newsletter on Wednesay around this time (9 AM Eastern time/GMT+1) (I know, today is Thursday - but everything got delayed due to tomorrow’s funeral)

  2. I am not focused on making a “fancy” looking newsletter, but rather try to provide value for those seeking more information on cancer innovations!

  3. I will present 3 to 5 summaries of articles or papers I find on AI cancer innovations each week. All sources will be linked for you to dive deeper if you want.

  4. All previous newsletters will be readily available to re-read online, and I will create a database of all the articles for my subscribers.

If you read the Practical AI newsletter, you know they are long, filled with Memes, and offer my thoughts on a technology or a topic.

Cancer Innovations is the polar opposite of this. No fluff, no lighthearted jokes or memes. Just quick summaries and links to resources for your deep dives.

With all that out of the way, let’s dive into the heart of the first edition of the Cancer Innovations Newsletter.

-Thomas

Today’s topics

  1. AI detects melanoma (skin cancer) with 100% accuracy

  2. AI-powered mammograms

  3. AI can determine where cancer arose

  4. AI for Health Institute (AIHealth)

AI detects melanoma (skin cancer) with 100% accuracy

A study assessed 22,356 patients with suspected skin cancers, and the AI software detected 100% of the cases of melanoma, the most serious form of skin cancer. It also correctly detected 99.5% of all skin cancers and 92.5% of pre-cancerous lesions. Last week, this study was presented at the European Academy of Dermatology and Venereology (EADV) Congress 2023 in Berlin.

Link to press release: EADV Congress
News coverage: New York Post
Doctor’s mentioned:

Institutions mentioned: 

AI-powered mammograms

Dr. Kathy Schilling and her Lynn Women's Health & Wellness Institute team are pioneering breakthroughs in breast cancer detection. With the aid of AI-powered mammograms, they've achieved a 23% increase in detection rates, identifying cancers as minuscule as 3-6 millimeters. This early detection transforms patient care, resulting in less aggressive treatments, reduced mastectomies, smaller lumpectomies, and diminished need for radiation and chemotherapy.

Link to article: MedicalXpress
Doctor’s mentioned:

Institutions mentioned: 

AI can determine where cancer arose

MIT and Dana-Farber Cancer Institute researchers have created a machine learning model, OncoNPC, to identify the origins of mysterious cancers. By analyzing 400 genes, it can pinpoint the source of previously untraceable tumors, accurately classifying 40% of them. This doubles the patients eligible for targeted treatments. The model's predictions align with survival rates and treatment outcomes, promising advancements in personalized cancer care.

Link to article: MIT News
Doctor’s mentioned:

Institutions mentioned: 

AI for Health Institute (AIHealth)

While this last section of today is not exclusively related to cancer, I decided to include it for two reasons: 1) to showcase the developments that are happening in AI and Healthcare in academics, and 2) maybe someone is reading this either want to study medicine or have children who want to study medicine and technology and want to be in the forefront of new developments.

The McKelvey School of Engineering at Washington University in St. Louis has launched the AI for Health Institute to design data-driven tools to characterize complex diseases, support clinical decisions, and drive precision health. AI for Health Institute intends to establish Washinton University as a leader in AI for health and wants to grow collaborative teams across engineering and health and build an organization and infrastructure for large research initiatives.

Doctor’s mentioned:

Institutions mentioned: