Can AI, Precision Health and Longevity Science Save Healthcare from Financial Collapse?

Image: Unsplash

In 2030 – The Future of Medicine: Avoiding a Medical Meltdown, published 15 years ago, I argued that ageing societies, rising expectations and expensive new technologies would push healthcare costs well beyond likely GDP growth — and that rationing, higher premiums and growing public dissatisfaction would follow. That is precisely what happened. What I did not anticipate then was the degree to which technology itself, more a factor I saw as likely to add to the challenge, would become the most important force for solving it.

Speaking recently at Humanitas University in Milan, I made the case that precision health, longevity science and artificial intelligence — applied through Leroy Hood’s 4P framework — could do exactly that. Let me explain why.

Prediction

In contrast to our model of disease ‘striking’ patients out of the blue, we now both understand many of the factors that predispose us to chronic conditions. Genetic (or other ‘omic) screening, either at birth or later, can help us identify high risk populations for specific testing and early intervention. Take Alzheimer's. Not only do we understand the gene variants that dramatically increase risk of future disease, we have blood markers that detect pre-symptomatic disease and give the chance for both lifestyle changes and therapies to change its course. And AI can analyse common blood markers to spot future disease before either the patient or the doctor realise it. One of my companies, Metadvice, is doing just this with heart failure.

Prevention

As well as markers that can alert us to conditions we might prevent, we are growing in our understanding of the processes of aging that bring with them so many of the chronic diseases we experience in the second half of life. This is where longevity science comes in, whether we call our goal healthspan, healthy life extension, or “peakspan”. Our preventive strategies run the gamut from well-accepted exercise and nutrition programmes, through the more exotic (and often poorly evidenced) hyperbaric chambers, cryotherapy and plasmapheresis procedures, to new geroprotective drugs, cell and gene therapies that are mostly still at the experimental stage. And let us remember, as we move to personalisation, that such programmes need as much personalisation as disease treatment. The good news is that AI can help sort through the many diagnostic markers to guide clinicians to such personalised prevention programmes.

Personalisation

Whether we call it personalised, precision or tailored healthcare, we all know that focusing expensive treatments on those most likely to benefit can save huge costs and improve outcomes. The good news is the progress in identifying biomarkers to guide the process. For example we now know what gene variants determine how quickly certain drugs are broken down, usually by the liver. We call this pharmacogenomics. It’s still insufficiently used and the medical culture of ‘trial and error’ dies hard. AI is again very relevant: we have a growing pool of knowledge of treatments and their outcomes across populations and so a steadily increasing ability to identify patient subsegments that will benefit from different regimens. Treatment of autoimmune conditions, for example, will be transformed once we can transition from naming the disease a patient’s symptoms indocate, like lupus or psoriatic arthritis, to focusing on the specific immune system disorders that underlie them.

Participation

It’s long been recognised that active patient engagement with their own healthcare is vital for a cost-effective health system. The question on the table now is whether the new tools from the AI revolution can aid or hinder that process. We already see the use of ChatGPT-type tools by patients to answer questions (usually but not always accurately) about their disease and their care. But we will see the latest wave of ‘agentic AI’ make possible personal health coaches and guides that patients can use to access well-curated medical information and also connect with specialists both before and after episodes of care. The key will be to develop these tools in a way that increases personal ‘agency’ rather than passivity.

So, if we add all these possibilities together, can they save us from healthcare insolvency? Rough calculations suggest that about a third of all costs could be saved if all these were able to be implemented. Of course, those costs that are saved by better prediction and prevention will take a few years to feed through, but the ultimate total is very encouraging. A 4P approach to medicine, empowered by a combination of AI, precision health and the fruits of longevity science, could just save us from the meltdown I wrote of 15 years ago. If you are working on any part of this challenge — whether in policy, investment, clinical delivery or technology — I would very much like to hear from you.