This week we will investigate some new ways at looking inside the body. We also discover how Astronomy is helping fight skin cancer and a new way to predict earthquakes 48 hours before they occur. We examine how an Australian Aboriginal language is helping develop AI robot to human communication and finally we catch up with the progress that AlphaFold is making on determining how proteins fold within the body.
Seeing inside the Body
Two new techniques have been developed that will give your doctor much clearer data on what is happening inside your body.
Stickers that provide Ultrasound Images
A team at MIT have developed a sticker the size of a stamp (remember them) that can provide continuous ultrasound images of internal organs for 48 hours. Ultrasound is a safe and noninvasive way of looking into the body. It gives live images of internal organs allowing much better diagnosis.
Current Ultrasound equipment is bulky and requires specialized skills to operate. Using the equipment requires a visit to a hospital or doctors office. In the future these patches may be as accessible as bandaids.
The device is made from two thin layers of elastomer that encapsulate a middle layer of solid hydrogel, a mostly water based material that easily transmits sound waves. The gel is elastic and stretchy. The elastomer prevents dehydration of the gell allowing a longer view inside the body.
The researchers applied the stickers to volunteers in a range of activities including sitting, standing, jogging and biking. The device produced live, high resolution images of major blood vessels and deeper organs such as the heart, lungs and stomach. Doctors were able to monitor changes to the organs as the volunteers moved.
The team is working on a wireless version which will allow remote continuous monitoring of patients either in hospital or at home. Patches on different parts of the body would provide data for an AI to analyze on demand and notify your doctor when required.
PillBots
Californian startup Endiatx has developed a pill sized robot that can navigate through the intestinal track of the human body whilst sending real time video.
Once swallowed the robotic pill can be remotely controlled whilst inside the stomach and other parts of the digestive tract. Using 4 tiny propellors, a video camera, battery and wireless link the pillbot provides feedback to doctors in real time. Currently the bot is controlled using an Xbox gaming controller however touchscreen mobile phone controls are being developed.
Endiatx believes the manufacturing cost will be about US$25 per bot (we will pay vastly more but that is a function of the medical system not manufacturing capability). It will allow faster screening for a range of ailments that would otherwise be practical.
The first clinical trials will commence later this year (2022) and Endiatx believes that the pillbot will be investigating patients around the world in 2024. The next step is to enable the device to take tissue samples and then perform other surgical tasks. Long term the goal is to shrink the device to the size of a grain of rice, opening up capabilities beyond the digestive tract.
Tech from Astronomy fighting Skin Cancer
Astronomers have automated the search for new galaxies and stars. The gigantic amount of data provided by the range of telescopes on earth and in space is far too large for a human to analyse themselves. AI will look at updated star maps and detect any changes or movements. This is a perfect example of where computers can do repetitive analytical work 24 hours per day far more efficiently than humans.
This technology is now being applied to the search for skin cancer. High risk individuals have a regular full body photograph taken and the skin is treated like an astronomical image. The software will find the moles that change over time and action can be taken early.
In the same way that we develop maps of the stars and how they change over time, the teams from Oxford University Hospital and the University of Southampton are developing maps of benign moles that turn into melanoma to enable earlier diagnosis of problematic moles and better outcomes.
Predicting Earthquakes
Researchers from Ariel University in Israel have developed a method to predict major earthquakes, 48 hours before they occur with an 80% accuracy rate.
The team developed their method by studying changes in the Earth’s Ionosphere (the thin atmosphere just before the vacuum of space). They looked at potential precursors of large earthquakes that have taken place over the past two decades. The focus was on earthquakes that surpassed magnitude of 6Mw (on the Moment Magnitude scale, a more accurate scale than the Richter Scale).
The team developed a machine learning algorithm that applied a support vector machine technique with GPS mapping data. The technique calculates the electron charge density in a region and predicts the earthquakes that will occur with an accuracy of 80%. It also predicts when earthquakes will not happen with an 85.7% accuracy. The prediction of an earthquake not happening helps to reduce fear in earthquakes prone areas.
Approximately 20,000 people die each year as a result of earthquakes. Mostly in poorer countries with cheaper building techniques. This technique will give early warning to many of the world’s most vulnerable regions and hopefully will reduce the death toll considerably. Building damage can only be overcome by building more earthquake proof buildings. That will take time and money.
Aboriginal Language helping AI
Jingulu is a language spoken by the Jingili people in the Northern Territory in Australia. Jingulu is unique even amongst Aboriginal languages as it only has 3 verbs “come, go and do”.
This simplicity of language allows it to translate directly into AI commands that humans can understand. By being efficient in its’ syntax it reduces computational cost.
Professor Hussein Abbass from the University of New South Wales develops AI that works with swarm systems where groups of robots (or drones) work together to solve very complex problems or perform tasks. His systems draw inspiration from sheepdogs where a few dogs can control a large flock of sheep.
The problem he needs to solve is about robotic movements in different information, knowledge and physical spaces. The movements are represented mathematically as elements that get attracted to each other or repulsed. When developing a human to AI communication system, Dr Abbass found that most languages were richer than needed, thus overcomplicating solutions. Reading a PHD thesis about Jingulu it struck him that the simplicity of the language would solve his problem.
Working with linguistic experts at the University of Canberra and the Australian Defense Science and Technology group the team created JSwam. An AI language inspired by Jingulu. The language can be applied to any situation where communication between humans and a large number of AI agents (or robots) is required.
Folding Proteins
In December 2020 we spoke about AlphaFold the Google AI that could determine the way that a protein folds. As we discussed at the time, how a protein folds is vital to the function of the protein in the body. This is important in drug development. AlphaFold could determine the way a protein folds in a few days, a process that had taken several months previously. There are approximately 2,000,000 proteins involved in human biology.
The Google team has continued to speed up and improve the process. Now, 20 months later, they have announced that AlphaFold has determined how almost every protein known to man folds. All will be freely available. When we can automate the discovery process, progress is very swift.
Paying it Forward
If you have a start-up or know of a start-up that has a product ready for market please let me know. I would be happy to have a look and feature the startup in this newsletter. Also if any startups need introductions please get in touch and I will help where I can.
If you have any questions or comments please comment below.
I would also appreciate it if you could forward this newsletter to anyone that you think might be interested.
Till next week.