AI in Healthcare, Space Launches, Exponential Tech

AI in Healthcare

Sharing medical data through Federated Learning

We start off with something that has been gaining a lot of popularity in the past couple of years, namely federated learning.

Federated learning (also known as collaborative learning) is a machine learning technique that trains an algorithm across multiple devices or servers each holding local datasets without exchanging these datasets. Said in simple words, this allows sharing data to train an algorithm without compromising its privacy. This is not only important for privacy reasons but also because of the proprietary nature of some data.

I might go into further details as to what federated learning is and how it works in another installment. For now, let’s look at a very recent example of how this technology is used in an innovative way.

A team of researchers in New York implemented and assessed federated learning models using data from electronic health records at five separate hospitals to predict mortality in COVID-19 patients. After training their models and testing them on local patients data at each hospital, the researchers found that the federated models demonstrated enhanced predictive power and outperformed local models at the hospitals.

I personally love seeing such great examples of using AI to solve critical challenges like easier and earlier identification of serious disease cases. Furthermore, this is an advancement towards adopting technologies like federated learning which have the potential to solve one of the biggest challenges when it comes to innovation in the healthcare sector – namely access to data.

Facial Recognition Improvements 

Another example of AI innovation due to the covid pandemic comes from the US Department of Homeland Security. They tested a facial-recognition algorithm that can accurately identify masked airline passengers 96% of the time. This is a huge improvement compared to the pre-pandemic facial-recognition algorithms which were struggling a lot to recognize people’s faces and identities when there’s anything on their faces like masks, sunglasses, etc. 

More investments in AI for Healthcare

And finally, on the topic of AI for healthcare, there’s more good news worth sharing. The EUR Horizon 2020 Research and Innovation program recently granted 10 million euros to the INTERVENE project. 

INTERVENE stands for INTERnational consortium for integratiVE GeNomics PrEdiction. Its aim is to develop and test next-generation tools for disease prevention, diagnosis, and personalized treatment by utilizing the first US and European pool of genomic and health data.

If successful, the potential for prediction, diagnosis, and personalized treatments for complex and rare diseases will be unprecedented. All and all, this creates some promising conditions for the future of AI applications in healthcare.

Space Technology

Moving on to the topic of Space, where there has been plenty of interesting recent developments.

Space Launch Conundrums

SpaceX successfully launched another 60 small satellites as part of the Starlink mega-constellations on the 20th of January. This brings the total to 1015 satellites delivered to orbit from which 950 are still active. The launch was on a Flacon 9 and went through without any problems.

A few days prior to this, on the 14th of January, Blue Origin also had a successful launch of their New Shepard spacecraft consisting of a rocket and a capsule. It reached about 107 kilometers altitude before going back down and landing without a problem. 

It is interesting to point out the difference between SpaceX and Blue Origin’s launches. SpaceX brings together millions of live viewers on the various internet streams. For example, as of the time of writing this, the January 14th launch of Blue Origin’s New Shepard has about 400 000 views on YouTube. On the other hand, the more recent Starlink launch already has 1.5 million views on the official SpaceX YouTube channel.

This is yet another proof of the huge influence that Elon Musk has. As it’s widely known, neither Tesla nor SpaceX delegate big budgets on marketing and almost all the brand recognition comes from Musk’s personal efforts. At the same time, Jeff Bezos doesn’t have nearly as good reputation and following as Musk.

I also feel it’s important to mention here that in between the launches of SpaceX and Blue Origin, NASA also performed its own test.

On the 17th of January NASA attempted to test the core stage of the giant Space Launch System rocket known as SLS. It is supposed to launch the Artemis 1 mission, an uncrewed test flight of the agency’s Orion capsule around the moon in November 2021.

Instead of burning for the planned 8 minutes, which is the burn time during a potential Moon Launch, they shut down in a just after 67 seconds. According to the agency the shutdown “was triggered by test parameters that were intentionally conservative to ensure the safety of the core stage during the test.”

The contrast between the private space companies’ success and the inefficiency of NASA is staggering.

Telescopes and Moon samples

On another, more positive note, NASA and ESA are planning to finally launch the James Webb Space Telescope later this year. It is a direct successor of the Hubble space telescope and it’s roughly 100 times more powerful.

While Hubble is in a close orbit around the Earth and primarily studies the optical and ultraviolet wavelengths, Webb will focus on the infrared and it will be 1.5 million kilometers away from Earth at the Lagrange 2 point.

According to NASA, using infrared light, the Webb telescope will be able to look farther back in time than previous telescopes and will allow scientists to look through dust to see stars forming inside.

Finally, in a week of multiple attempted launches, we also witnessed the new president of the USA enter office under a space symbol.

NASA lent a Moon rock sample to be displayed at the Oval Office. The sample comes from the Apollo 17 mission from 1972 which was the last manned-mission to the Moon. The stone weighs 332 grams and is presumed to be 3.9 billion years old.

Exponential Technologies

Hydrogen startups stepping in

Ok, moving on to another topic – Hydrogen batteries, the hottest topic in the energy sector right now.

The Australian startup LAVO has built a residential hydrogen storage system allowing users to store solar power enough to run an average home for two days. This is three times better than Tesla’s Powerwall 2.

Such developments make me very excited and I believe that hydrogen-power will capture a big part of the energy market in the upcoming decades.

Earth Observation

As a final note, a I would like to draw your attention to the intersection between satellite earth observation and AI.

A team of researchers from the UK developed a deep learning algorithm able to detect elephants on satellite images. Traditionally it is a very difficult task to counting elephants in the wild. The new technology method offers accuracy comparable to human detection capabilities but more efficient as it’s cheaper and faster.