Betting on a healthier humanity

Key points:

1. A healthier humanity makes it more innovating and risk-tolerant
2. Velocity in healthcare innovation by serving the under-served regions in the world
3. Aggressive on AI to lower unit costs to achieve break-even for healthcare in under-served regions
4. Medical AI for interstellar and beyond

A healthier humanity

I think the best way to structure this, is to start from my vision, my happy place and work backwards. The goal to not forget the dreams that captured me as a kid, the fire and the passion that captured my mind. To not waver in the face of "adulting", but to keep that naive optimising for a bright future, and the belief that I can change the world.

I inspired by the Govindappa Venkataswamy, this vision to bring the healthcare cost down for the poor, the veritcal integration, the innovation. I first discovered him in 2018 from Google 100th birthday tribute Doodle. I think that is an inspiring vision. I saw how people who are poor cannot afford healthcare, in the wards, back in 2016, a patient wanted to self-discharge because they knew they could not afford the treatment. The patient was not a citizen of Singapore, there are not support systems for immigrants.

I have written extensively about healthcare. Healthcare is a parasite of our economy, it's inefficiency drains the resources that can otherwise be used to pushed the boundaries of the human race. Healthcare, the like housework in a domestic home, is taken for granted, until it is not provided anymore. It is the bedrock of the society, it is the core of the human race to enable humans to do great things. Healthcare, is also the core that gives the comfort in people having children, the raise the next generation of humanity to future the human race. Just like the criticality of logistics support that allowed the US to win the Gulf war, healthcare is the critical logistical support for the human race to advance into the next generation as a multiplanetary species.

We cannot dare to dream, to be ambitious, if we are afraid of our livelihood, if we are afraid that we cannot afford the basic neccessities of life. We must be confident and self-assured of our basic needs, via maslow's hierarchy of needs, to then be able to dream larger. In this sense, healthcare as large positive externalities, in that it can create a more ambitious population, to dream bigger, reach futher, the make the big and risky bets that will bring humanity forward. If I am afraid of my ability to make a living, why would I bother thinking of starting a company, if I cannot be assured that I have a comfortable life, why would I take the risk to invest all my savings to start a company? This can be true in developed and developing countries. The aim of a comfortable life, is common throughout. The reason why finance/counsulting/medicine and professional jobs are so popular is because it guanrantees a comfortable life. It provides structures and guanrantees a good level of standard of living. And it is also one of the few roles that you don't have to be the best in the field to make good money. In many jobs, it disproportionately rewards the very best (football/singers/actors), partially due to the scalability of these jobs (infinite audience scaling).

The thesis is that less of humanity's consciousness should be invested in trying to survive and more of it should be focused on trying to push the boundaries of the next generation. The only way this is possible is that we have some guarantees take make the basic necccessities completely affordable. There is inefficiency in human consciousness, the time spent within 'games' of organisations of arbituarily created rules do not actually advance humanity of make anyone else's life better. It solely reinforces the system that has been constructed, the strengthen the rules of the game that benefits no one.

Healthcare, I believe, is the keystone towards a brighter future for humanity. To make healthcare, too cheap to charge, to empower the next-generation to dream big, to take risks and focus on geniuinity on what matters the most to them.

The second is the concept of mental loading. Preventive healthcare and lifestyle medicine have came up in popularity as a way for governments to find a way to spend less on healthcare while achieving better outcomes. It is very sound in theory, except for the fact that it does not take into account the mental load it takes to have a good lifestyle for optimal healthcare. If you are poor, and finding it difficult to make ends meet, would you care whether you ate something unhealthy, you will find the cheapest thing that keeps you alive. A good lifestyle requires sufficient mental load and capacity, to think about each lifestyle decision and consciously make an effort to change it for the better. If you come back from a long and tiring day of work, would you have energy to go to the gym? Compared to someone who works at the office in air-conditioning. The ideal healthcare experience is one that reduces mental load. Even traditional clinics and appointments carry a significant mental load. Do you have to travel 50km to get to the hospital, you have to take leave off work, you have to wait for hours in the waiting area, you have to call to change and appointment, you have to call to receive instructions about treatments/surgeries? All that is mental loading.

Just like how SpaceX has taken over the key effort for space exploration, I believe that in healthcare, an innovation should reduce the price of healthcare and not require extensive government funding to keep the country healthy, freeing up tax income, or even enable reduction of taxes to encourage more private innovation. a 1000x reduction in healthcare cost is the aim.

This is the bet on a healthier humanity. The goal is a complete transition of society's mentality, from a self-protective to a risk-taking innovative state. And this will drive the next great breakthroughs for the next generation. In the human body, an invasion by viruses activates a self-defensive mechanism by the body, releasing interferons. Interferons alter the metabolic state of a cell. This state makes it harder for the virus to replicate - almost 1000 times harder for the virus to replicate and spread, hence protecting the other cells from viral invasion. This self-protective state is effective at preventing total body collapse. However, in this interferon-filled self-protective state, it limits many of the cell's normal function, limiting the cells normal activities. In this state, it's growth, development and the ability to support the bodily functions is degraded. By securing basic neccessities for normal and healthy function in a human society, it enables each individual to fulfill their true potential to encapsulate the brilliance of the human mind, the creativity, the steadfastness and the recklessness to embark on the next chapter of human civilization.

How can this be achieved?

In a field where it is accepted to be slow moving, with heavy regulatory oversights (for good reason), the ultimate advantage is from speed. And this speed is well-justified too. There are lots of suffering in this world, many who cannot access critical and life-changing medical treatment. A technology that is a day earlier can bring hope and life to many and empower their potential. We must break the traditional medical paradigm. The key is to go into underserved regions in the world, where there is close to zero healthcare being provided. With 'do no harm' as a primary principle, it is easy to provide medical benefit, while remaining ethically sound. Safety is of upmost paramount, and is the number 1 priority in all cases. As a benchmark, clinical outcomes should be comparable to developed countries, and the bottom line should be a standard deviation above the previous state in that area. Of course, in these regions you cannot expect to be paid much. The key here is to have a small team of doctors augmented by AI and ML tools, to increase the efficiency and productivity to achieve unit economics that healthcare becomes an affordable and viable option for these groups. The goal is to build a profitable healthcare company in the world's most under-served regions. In such an environment, it allows for rapid development and iteration, every improvement in the system and ML models can save more lives, treat otherwise deadly diseases. The velocity will enable the leapfrog of affordable technologies beyond the rest of the world. Rather than a J-curve, this is ReLU financial model. Start with a non-profit model, establish break-even at low-prices, improving the tech to establish strong unit economics. Then move into developed markets at a few magnitudes above the cost price, but yet at a few magnitudes below the current market price. Fight aggressively on the packaging and service experience. The developed markets, it is service-first, think first-class on an airliner, invest in lounges etc. The primary differentiating factor. The catalytic effect of this have lots of positive externalities. Existing market players will be forced to lower their prices, innovate and drive the technology forward, to ensure lower cost healthcare for everyone.

I believe that it is only in diving in the lower end of health inequality that we can get an urgency to execute fast, to iterate fast and to save the lives and would otherwise be lost. This spirit and purpose is critical to the success of this.

Next is the specialty. Specialisation and vertical integration is important. For sure, all health monitoring and lifestyle medicine is completely out of the question. This does not mean conservative advice and treatment would not be given. But rather, we would stray away from influencer-type health obsession with excessive tracking, scans and tests. A lot of lifestyle medicine relies on have loads of mental capacity to have a choice, resources for a healthy lifestyle, which is often not an option for mentally loaded individuals. The goal is to find areas of minimal cost fixes with the maximal benefit. The two that stands out at the moments is general practice / internal medicine, and trauma and orthopaedics. T&O gives often great outcomes compared to other surgical specialties, and it's demand is ubiquitous. GP and internal medicine is heavy on the diagnostic aspect and with a pharmacaeutical management approach via generic drugs can be low cost. This part is heavy on intelligence and cheap on actual raw materials (drugs), and with the price of intelligence and service (history taking) decreasing with AI, the remaining cost would just be the drugs. If they are generic it can be a few cents.

What are the current software efforts to support this goal?

1. Interstellar Surgical (interstellarsurgical.com)

Surgery is expensive. It is also notoriously unscalable. A single surgeon only have 2 pairs of hands, and 24 hours in a day. To achieve affordable surgeries in the most deprived areas in the world, it must be automated. No surgeon is willing and able to work in conditions where the pay is not enough to survive. FSS (supervised) - full self-surgery (supervised). Cataract surgery / hip replacement / prostectatomy, in each surgical specialties that are bog standard surgeries, with procedures well-established, bread and butter surgeries. With automated monitoring, a surgeon would only be required to take over surgeries when the AI flags and error or edge case it is unsure about. The surgeon can then takeover the robotic surgery remotely via the control room (or even teleoperated). In terms of unit economics, if there is 1 edge case surgery in 1000, this means that if we charge £3.85 monitoring fee per surgery (meal deal pricing), then each surgeon can get £2000 a day for surgery (rest is profit margin), making it a viable career for surgeons. Teleoperated surgery also makes working-from-home a viable option for work-life balance. Teleoperative nature of surgery opens up a large funnel demand for surgeries from under-served area of the world, to the large supply of international surgeons all over the world, essentially expanding the pie while empowering individuals with life-saving surgery.

But how can surgery be automated?

First, we must operate on the assumption that the underlying AI/ML layer, robotics layer with general intelligence would be solved in the near future. Building the medical/surgical layer takes time, and it must start now. An AI-native surgical system have lots of quirks to be worked out, and the goal is to have this layer polished and refined so that when the underlying AI/ML layer is ready, FSS (supervised) can start immediately.

Second, there must be a concerted effort to build a dedicated surgical AI/ML data and training pipeline. Foundational models will get better, but surgery is not something a generally intelligent human can perform. We need better than general intelligence. Surgical data is rare, hard to come by. Without this data, it is hard for AI to perform at a precision where even a small mistake can be devastating.

To work on the second point, I am working on an AI simulated platform to train surgeons. By creating an interactive world-model it allows surgeons to practice virtual surgeries in silicon. In some sense, it is a chicken and egg problem, where to create this interactive world-model we also need surgical data. But the bet is having a critical mass to create an interactive world model is much less that what is required to create general surgical AI, because the former is a subset of the latter. Any proficient AI surgeon would have a sophisticated internal world model. Once and interactive world-model is created, it can be used by surgeons to practice surgeries. Full and complete surgeries will enable proficiency before the first real patient surgery to be done. It would allow complex and experimental techniques to be trialed before being implemented in real-life harnessing surgical innovation in a risk-free environment. These will generate a large trove of surgical data, particular detailed action-coupled data which would be critical for training surgical AIs.

Another side-effect of this is to produce turn exceptional 'layman' individuals into surgeon, by widening the talent pool search for surgeons. Traditionally, the barrier to entry into medicine and surgery is extremely high. And most of the times, the selection pressures do not neccessarily select for the right traits to be a surgeon. With a risk-free virtual surgical simulation, anyone with a computer and have a shot at surgeries, then moving onto surgical controls, and the top players with the best surgical outcomes and skills can be given a conversion programme to become real surgeons, opening the talent pool, and reducing the cost of surgeries.

A key direction of this surgical simulation is robotic surgeries with robotic arms. This is because the degree of freedom is much less and the arms can be controlled via keyboard controls. In contrast, open surgeries have much more complicated, you can position your hands and body position a trillion ways etc. Robotic surgeries helps to reduce the degrees of freedom for better results in action-data collection and AI driven robotic actions. This is the first step.

Of course, the future bet will be on humanoid robotic surgeries, with a much larger degree of freedom, and it would be wise to work on this early too, but the priority firstly is getting traditional robotic surgeries working first. One of the great advantages of humanoid robotic surgeries is the generalisability which means the cost of production will be much lower, because the market might be catered for all use cases which drives the prices down.

Currently, my key focus is to build the interactive world model for surgery. Using open-source surgical videos, and training on top of open-source interactive world models such as matrix-game 2.0. I have onboarded 5 urological robotic surgeons who are excited to be involved.

It is called interstellar surgical because an FSS (full self-surgery) will be critical for human's journey to be a multiplanet species. For surgery to be ubiquitous, in deprived area or in other planets.

2. Zargard (zargard.com)

This is an AI production monitoring platform, where an AI serves as an on-call engineer to monitor anything that is going on with your production build, by monitoring metrics and errors. And notifying the relevant teams to come onboard when something bad happens. It can also introduces fixes automatically. This is like sentry but with AI superpowers. To be able to draft fixes for production issues, so that software systems can be self-healing.

THe idea of continuous monitoring does not only apply to software products. Just like on-call engineers, there are on-call doctors. However, the monitoring process must be automated. With continuous monitoring systems with automated vitals tracking, bloods, etc. With IV access and drugs automation, medical management can also be implemented. To take it a step further, this sort of monitoring would apply to anaesthetic monitoring during a surgery and adjustments during that process.

The name Zargard comes from Melzargard, a One-punch man monster that have self-healing properties. Software systems and humans should also have self-healing properites like zargard.

3. ECG Dojo (ecgdojo.com)

This is a platform for medical students and registras can practice interpreting ECGs, via a broad doctor-annotated dataset. Hopefully it can help to collate clinical reasoning for better AI assisted ECG interpretation.

4. Kings Cross Hack Lab (kxhacklab.com)

This is the platform in which I gather the brightest minds in London into a hacker lab. A lab of building, shipping and innovating. This is the primary platform in which I will use to work on these software efforts. It serves as a community I can talk to, discuss interesting problems and get inspiration and insight. Also a great place to find co-founders.