Five shifts reshaping medicine: from reactive care to computable health

For most of human history, medicine was reactive, episodic, and confined to the textbook, the clinician’s mind, and the hospital’s walls. One fell ill, consulted a clinician, and received treatment based on population averages and professional intuition. That model served us well. It also left vast potential untapped. 

What we’re witnessing now is medicine becoming something fundamentally different: proactive rather than reactive, continuous rather than episodic, personalised rather than averaged, and distributed rather than confined.

The boundaries between “healthcare” and “life” are dissolving. The question is no longer just “How do we treat disease?” but “How do we optimise health across the entire human lifespan?"

Here are five shifts I believe are defining this transformation, and why they matter for anyone thinking seriously about the future of health, technology and human flourishing.

1. Medicine as an ecosystem

What it is

Medicine has evolved from a reactive, clinician-centric discipline treating illness into a proactive, distributed ecosystem managing health. The boundary between “healthcare” and “life” has dissolved. This enables longevity science to scale as legitimate medicine, as your watch, your toilet, your mirror, your phone, all become nodes in a health network that monitors, predicts, and intervenes continuously.

The patient is no longer a passive recipient but an active participant generating data and making informed decisions. Medicine now encompasses prevention, optimisation and wellness, not just the treatment of pathology. But not just this, the ecosystem encompasses thinking machines, rapid data sharing, and startups focused on niche aspects of health.

Telemedicine, hospital-at-home, and retail health (pharmacies becoming clinics) are reshaping where medicine happens, and devices are reshaping how it happens.

Tangible examples

Apple Watch detecting atrial fibrillation and prompting cardiology referrals before stroke occurs. Continuous glucose monitors (CGMs) like Dexcom/Libre being used by non-diabetics for metabolic optimisation. Or Whoop recovery scores guiding athlete training loads.

These aren’t “medical devices” in the traditional sense; they’re lifestyle products with medical utility, blurring the line entirely.

Future projection

Within 10 years, the concept of an “annual checkup” will seem absurd, like checking your bank balance once a year. Continuous multi-biomarker monitoring (glucose, cortisol, inflammatory markers, hormones) will feed into a personal health AI that manages your wellbeing longitudinally.

The doctor becomes a specialist consultant for edge cases, not the primary interface. Health insurance may be priced based on real-time data streams rather than annual questionnaires. The health “ecosystem” becomes as ambient and invisible as electricity.

2. The protein folding revolution

What it is

For 50 years, predicting how a protein’s amino acid sequence folds into its 3D structure was biology’s grand challenge: the “protein folding problem.” In 2020, DeepMind’s AlphaFold essentially solved it, predicting structures with experimental accuracy. This matters because a protein’s shape determines its function, and understanding function unlocks drug design, disease mechanisms, and biological engineering.

We went from knowing ~170,000 protein structures (accumulated over 50 years) to having predictions for over 200 million, virtually every known protein.

Tangible example

Drug discovery timelines are collapsing. Traditionally, understanding a disease protein’s structure could take years of lab work. Now, researchers can pull predicted structure in minutes and immediately begin designing molecules to interact with it.

Companies like Insilico Medicine used AI-driven approaches to identify a novel drug target and design a molecule for idiopathic pulmonary fibrosis in under 18 months, a process that typically takes four to five years.

Future projection

This is the foundation for “programmable biology”. If you can predict structure, you can design proteins that don’t exist in nature: custom enzymes, novel therapeutics, synthetic biology components.

Within a decade, we’ll see designer proteins for carbon capture, plastic degradation, and therapeutic applications we haven’t imagined. Combined with AI drug design, the pharmaceutical industry’s economics transform completely. Fewer failures, faster timelines, and an explosion of treatments for rare diseases previously too unprofitable to pursue.

3. Designer drugs, CRISPR & nanotechnology

What it is

We’ve entered the era of molecular precision in medicine. Revolutionary gene-editing tool CRISPR enables editing the genome with find-and-replace simplicity. Designer drugs (including mRNA therapeutics, are now being trialled for personalised cancer vaccines) allow us to programme the body’s own cells as drug factories. Nanotechnology enables targeted delivery, getting the right molecule to the right cell without systemic side effects. Together, these represent medicine operating at life’s fundamental unit: the molecule.

Tangible example now

CRISPR-based therapies have been approved for sickle cell disease and beta-thalassemia (Casgevy, approved 2023): the first genome-editing treatments where patients’ stem cells are extracted, edited to fix the defective haemoglobin gene, and returned. Effectively a cure for conditions previously requiring lifetime transfusions. Then there's Onpattro, an RNAi drug delivered via lipid nanoparticles that silences disease-causing genes in hereditary transthyretin amyloidosis.

Future projection

Gene editing will move from last-resort treatment to front-line prevention. Embryonic screening and editing (controversial but inevitable in some jurisdictions) eliminates hereditary diseases.

Nanoparticles become programmable delivery vehicles. Imagine “smart” drug carriers that circulate harmlessly until they detect cancer biomarkers, then release chemotherapy only at tumour sites.

The convergence with AI means drugs will become dynamically personalised: your genetic profile plus real-time biomarkers equals a therapeutic regime designed specifically for you, adjusted continuously.

4. Data democratisation and abundance

What it is

Healthcare has historically been data-poor and siloed: records fragmented across providers, privacy concerns blocking research, and entire populations invisible to algorithms trained on narrow demographics. Three shifts are dismantling this: 

• Synthetic data (AI-generated datasets that preserve statistical patterns without privacy risks)
• Federated learning (algorithms that train across institutions without data leaving)
• Wearable explosion (billions of people generating continuous health signals)

We’re moving from data scarcity to data abundance.

Tangible examples

Synthetic data companies (Syntegra, MDClone, Gretel) are allowing pharma companies to access “virtual patient populations” for drug trials without HIPAA concerns.

The UK Biobank – 500,000 participants with genetic, imaging and lifestyle data – becoming the foundation for thousands of studies that would never have happened with traditional siloed approaches.

Then there's Apple’s Research app that enables studies with millions of participants (Stanford’s heart study had 419,000 participants in months). In our own work, we generated chest X-rays with clinically significant pathology until the generated images confused expert practitioners as to whether or not they were real. In another study, we trained an algorithm from just seven base images to detect BMI reliably from a photograph.

Future projection

The data bottleneck that constrained medical AI will break completely. Algorithms trained on synthetic-but-statistically-real data from every demographic, every rare disease, every edge case.

Cross-border medical research accelerates: a researcher in Johannesburg can train on European data without it ever crossing borders.

This creates a “long tail” of medical knowledge, where conditions too rare for traditional study become tractable. The flip side: data ownership becomes a contested terrain. Who owns your continuous health data stream? Health data may become a regulated utility, like water or electricity.

5. Intelligent augmented care

What it is

Artificial intelligence is becoming the physician’s cognitive co-pilot, handling pattern recognition across scales no human can match, surfacing relevant research, predicting deterioration before clinicians spot it, and increasingly, interacting directly with patients.

This isn’t a replacement but an augmentation: amplifying clinical capability while enabling patients to access medical reasoning previously locked behind professional gatekeeping.

Tangible examples

AI is reading radiology images with superhuman accuracy for specific tasks: detecting diabetic retinopathy, identifying lung nodules, spotting breast cancer in mammograms. Or clinical decision support systems that flag patients at elevated sepsis risk, prompting earlier evaluation.

On the patient side: symptom checkers (Ada, Babylon) and AI health coaches providing 24/7 access to medical reasoning that would cost hundreds per hour with a physician.

We previously used AI tools to predict errors in laboratory medicine, a field which has been elusive for a long time. We also used it to diagnose childhood pneumonia, and predict missing data in HIV support research. It works now, both to solve problems directly, and to create tools that solve problems. This is the paradigm shift.

Future projection

The physician’s role will evolve from “knower of medical facts” to “orchestrator of care and human connection.”

AI handles the cognitive heavy lifting (diagnosis, treatment optimisation, literature synthesis) while doctors focus on communication, judgment calls, and the art of healing that transcends protocol.

For patients, medical knowledge becomes democratised: your AI health companion knows your complete history, has read every relevant paper, and is available at 3 am when you’re worried. This is simultaneously empowering and dangerous.

The quality of AI guardrails becomes a matter of life and death. Regulation struggles to keep pace.

Conclusion: medicine becomes computable

These five shifts share a common thread worth making explicit: medicine is becoming computable.

From the atomic structure of proteins to the population-level patterns in synthetic datasets, from the molecular precision of CRISPR to the algorithmic pattern matching of diagnostic AI, everything is increasingly reducible to data that machines can process, optimise, and act upon. This is not a threat to the humanity of medicine. If anything, it’s a liberation.

When AI handles the cognitive heavy lifting (the pattern recognition, the literature synthesis, the risk calculations), physicians are freed to focus on what machines cannot do: the human connection, the nuanced judgment, the art of healing that transcends protocol.

But this transformation also demands new thinking. The winners in healthcare’s next chapter won’t be those with the most beds or the longest-tenured specialists. There will be those who understand that biology has become an information science, that health is a continuous stream rather than a periodic event, and that the patient, armed with data and AI, is no longer a passive recipient but an active participant.

For investors, policymakers, and healthcare leaders alike, the question is no longer whether these shifts will reshape medicine. They already are. The question is whether we’ll shape them thoughtfully or be shaped by them.