Convergence of Bio, Health-tech, and Digital.

​
This is the final of our new weekly strategic series of Top Ten Anticipations for 2025 – 2030. Topic number nine, "Robotics: from warfare to welfare", can be found here: https://lnkd.in/g9Ydu3Zx
​

Biological sciences – such as genetics and genomics – are increasingly interacting with digital tools such as analytics to revolutionise and personalise patient care, reducing risk and improving health outcomes.
​
Between 2025 and 2030 we will see an explosion of health data as digital health tools are used more often, resulting in new cures as well as a step-change in preventative medicine thanks to the pooling of enormous datasets.
​
The processing of vast amounts of data from clinical trials, insurance companies, and electronic medical records will improve patient care and healthcare delivery.
​
Scientific discovery will also be supercharged by models which can autonomously analyse huge amounts of data, deepening our understanding, creating new hypotheses, and uncovering health innovations.
​
Breakthroughs in deep learning will enable researchers to unlock findings that would have been near-impossible otherwise and accelerate the rate of scientific discovery more broadly.
​
Diagnoses and treatments will be well on the way to being specific to the individual by 2030, leveraging genetic, biomarker, and other patient-specific data to tailor treatments, significantly improving their efficacy and safety.
​
Telemedecine and telehealth will be the default come 2030 – both for non-urgent ailments in urban environments, and also for emergency situations in remote areas.
​
Depending on where you live and your means, by 2030 patients will have the option to opt-in for additional automated algorithmic screening as a safeguard; during either a regular medical or something routine like a mammogram.
​
Proteins like DNA are made up of chains of amino acids that fold into unique 3D shapes.
​
Already, Google’s Deepmind AI system AlphaFold can accurately predict 200 million known protein structures.
​
Genetic engineering – designing plants or flowers that have never existed before – will see new smells and unique forms created.
​
Let’s also not forget how your health-tech wearables will advance and contribute to wellness, becoming implants (as explored in week four of this Anticipation series).
​
Despite positive progress, environmental and ethical considerations will rightly be omnipresent when considering the impact of digital on Bio and Health-tech – there are risks to personal privacy, social considerations, and the energy required to run everything.
​
Nevertheless, the period 2025 to 2030 promises more accessible, personalised, and effective healthcare for many – though, sadly, not all.
​
In addition to improving healthcare, longer-term this convergence has the potential to accelerate advances in disease prevention and regenerative medicine - ultimately enhancing the wellbeing of society.
​
We hope these Strategic Anticipations into the future have given you much to ponder about the next half-decade which will be here before we know it; please do respond to this Email and let us know your feedback.

Ouroboros - we may see this in the New Year.

​
The Ouroboros is an ancient symbol depicting a serpent or dragon eating its own tail.
​
With Christmas being over, next up is Hogmanay (as Scots know it) on December 31st, if one follows the Gregorian solar calendar.
​
Following on shortly in late January 2025 will see a change in the Spring Festival (also commonly known as Chinese New Year - CNY).
​
Each CNY is associated with one of the 12 zodiac animals, each paired with one of the five traditional elements (Wood, Fire, Earth, Metal, and Water) in a 60-year cycle.
​
The combination of the two traditionally defines the year's characteristics.
​
Coming next is the Wood Snake.
​
The Snake is the sixth sign in the Chinese zodiac, and is associated with wisdom, intuition, and metamorphosis - Wood adds creativity, growth, and versatility to the Snake's traits.
​
These ancient observations can potentially tell us much about what the future may hold.
​
2025 could see us hitting the buffers when it comes to creation and consumption of data, especially in the realm of Artificial Intelligence (AI).
​
Previous post on Model Collapse: https://lnkd.in/gmiA5SdN​
​
In order to learn, any and all AI algorithm needs to be trained on huge amounts of representative data - and of that, there's only so much.
​
Synthetic data is thus becoming more common, valuable for use cases such as autonomous vehicles.
​
The timeless adage always holds, regarding of the tech: garbage in, garbage out.
​
This recursive pattern of rinse-and-repeat, when extrapolated out, results in the models iteratively shifting further away from the original "real" data.
​
The outputs they generate can thus be blurred, unreliable, distorted, or warped - feedback loops degrading as they go.
​
Not only that, but folding in on itself translates to outputs that become not only less representative but also more uniform, as there is minimal variability introduced - eventually resulting in potential irrelevance.
​
May 2025 and the Wood Snake deliver us creativity and wisdom, to avoid Ouroboros in action resulting in Model Collapse.

Riddle me this: "I am easy to lift, but hard to throw - what am I?".

​
A feather.
​
Humans know the meaning of words and can thus reason - Artificial Intelligence (AI) models, right now, do not and thus cannot.
​
AI wouldn't stand a chance against supervillain character The Riddler (pictured), the criminal mastermind who appears in DC Comics and is an enduring enemy of Batman.
​
Large Language Models (LLMs), which underpin Generative AI, use statistical analysis to find patterns in enormous bodies of text.
​
LLMs thus effectively "guess" what word best follows another word - based on their extensive training data and also user input.
​
When you ask a question, AI works through the relationships it's spotted between words, phrases and ideas, and uses that to predict the most likely answer to your prompt; rinse and repeat.
​
AI excels at routine tasks and pattern recognition - but not at abstract thinking; no algorithm possesses common sense...at least not yet.
​
All models have a general lack of grounding in the world, which makes basic, flexible reasoning and creative contextual thinking a struggle.
​
The current generation of AI uses neural networks which are modelled after the structure of the brain itself - as much as we understand it.
​
AI can do amazing things today - solving problems that humans can't dream of achieving when it comes to complexity at scale, especially when there is sufficient representative data that it can learn from.
​
To be clear, Gen AI models will answer some riddles correctly - but only if they have encountered them previously during learning via training data.
​
However, the manner in which many chatbots often definitively answer - much like an expensive young management consultant fresh out of business school who has little practical experience - can lead users astray.
​
AI and humans can work really well together - but don't rely on them too much, and always check the veracity of any output(s).
​
If you need lateral thinking and ingenuity when it comes to solving a puzzling situation - for the near future it's still best to ask a cunning or creative human who can use non-artificial intelligence.