AI and Precision Health in 2025
Joe Cohen — Precision Health Biohacker, CEO @ SelfDecode
Joe Cohen explores how AI and precision health are revolutionizing personal wellness, moving beyond
"One size doesn't fit all; most things that work for me won't work for you. Precision health means knowing your unique genetic risks, lab markers, and symptoms to find what truly suits your body and leads to optimal results."
Summary
- Personalized approaches outperform generic blueprints; individual genetics and biomarkers determine effective supplements and treatments. - Most people need supplements due to environmental toxins, but type and dosage must be personalized based on genetics and lab tests. - Using AI and polygenic risk scores improves prediction of disease risk better than single genetic variants or traditional tests like PSA. - Integrating genetic data, lab markers, symptoms, and AI (like ChatGPT) enables tailored, dynamic health recommendations. - Precision health focuses on preventing diseases early by analyzing genetics, biomarkers, and lifestyle, thus enhancing longevity and quality of life.
Article
Precision health is not about following blueprints, but understanding your unique biology
Speaker at Tokyo summit challenges conventional wellness advice, arguing individual genetic and biomarker data should guide supplement choices rather than celebrity health protocols
At the Hololife Summit 2025 in Tokyo, Joe Cohen, founder and chief executive of precision health company Self Decode, delivered a pointed critique of the wellness industry's obsession with one-size-fits-all solutions. Drawing on two decades of personal biohacking and his experience overcoming 75 documented health issues, Cohen made the case that artificial intelligence combined with comprehensive genetic testing represents the future of preventative medicine.
"I think the vast majority, almost everybody does need supplements. We're living in a very toxic world with many challenges, so people do require supplements to be optimal," Cohen told the audience on Saturday. "The question is, how much and which ones?"
The failure of health blueprints
Cohen's message challenged the current trend of following influencer protocols and celebrity wellness routines. His own journey, he explained, was marked by repeated failures when attempting to replicate others' approaches.
"When I started, I tried to follow different people's blueprints and failed every time," he said. The problem with adopting someone else's complete regimen, according to Cohen, is the impossibility of determining what actually works. "Some things may work for you, others may not."
This realisation led him to develop a more systematic approach, one grounded in continuous measurement and personalised analysis. Cohen now tracks over a thousand biomarkers monthly and takes 175 supplements daily, but each intervention is documented and justified by specific genetic variants or laboratory findings.
Why single genetic variants tell an incomplete story
A central theme of Cohen's presentation was the inadequacy of conventional genetic testing. Many direct-to-consumer genetic testing companies analyse only a handful of variants, which Cohen argued provides dangerously incomplete information.
"If you only look at a single genetic variant, your prediction is like a coin flip," he explained. "Using sophisticated polygenic risk scores and additional factors can increase accuracy up to almost 90% for predicting diseases."
Polygenic risk scores use artificial intelligence to analyse millions of genetic variants simultaneously, providing a far more nuanced picture of disease susceptibility. Cohen gave the example of Parkinson's disease, where his personal genetic risk is substantially influenced by exposure to pesticides rather than by any single gene variant.
"AI is needed for genetics because chronic conditions are determined by a million variants," Cohen said. "Using AI to analyse many variants lets us predict risk profiles accurately, which is essential for precision health."
The role of biomarker monitoring
Beyond genetics, Cohen emphasised the critical importance of regular laboratory testing. He shared how tracking biomarkers revealed that adding manganese to his regimen improved muscle pain related to oxidative stress, despite manganese being a supplement he would not recommend to most people due to potential neurological risks at high doses.
The key, he suggested, is understanding that "normal" clinical ranges for markers like alkaline phosphatase or ALT are not the same as optimal ranges. By tracking these measurements over time and correlating them with genetic predispositions and interventions, individuals can fine-tune their health strategies.
"You want to check related lab markers and genetics for every disease," Cohen advised. "When you know early enough, you can prevent chronic conditions like Alzheimer's long term. It's easier to prevent than to treat once you have the condition."
Integrating artificial intelligence with personal health data
The most significant development Cohen highlighted was the potential for large language models like ChatGPT to synthesise complex health data. While these AI tools can provide general health advice, their true power emerges when integrated with personal genetic and biomarker information.
"The secret is to integrate AI and large language models with your genetic and lab data," he said. "When you combine all your information, you get personalised, actionable health insights rather than just general advice."
Cohen demonstrated how ChatGPT's new app integration capabilities could theoretically access symptoms, conditions and lifestyle data, though he noted that genetic risk analysis still requires specialised platforms like Self Decode. The combination, he argued, allows for far more sophisticated recommendations than following generic protocols.
Pathway-based functional medicine
For individuals already dealing with specific conditions, Cohen advocated for a pathway-based approach that examines genetic variants involved in particular biological processes. He gave the example of methylation, a crucial biochemical pathway affecting everything from DNA repair to neurotransmitter production.
By analysing up to 150 genetic variants related to methylation and comparing them with laboratory markers like homocysteine levels, Cohen was able to identify his specific weaknesses and address them with targeted nutrients including zinc, riboflavin and methylfolate. His homocysteine levels, initially predisposed to be elevated due to genetic factors, are now well within the optimal range.
"Functional medicine focuses on pathways," he explained. "With personal genetic and lab data, you can see exactly where your problem lies and address it with personalised interventions."
Rethinking longevity and disease prevention
Cohen's vision for precision health fundamentally reframes how we should approach ageing and disease. Rather than accepting illness as an inevitable part of growing older, he argued that most age-related diseases can be delayed or prevented entirely through early detection and intervention.
"People don't die of 'old age'; they die from something," he said, pointing to research on centenarians who simply push off diseases like heart disease or cancer by two decades compared to the general population.
This preventative approach requires looking beyond conventional medical screening. For example, Cohen noted that polygenic risk scores can be more predictive for prostate cancer than even advanced imaging techniques or standard PSA tests.
The implications extend to which laboratory tests individuals should prioritise. "Genetics also helps determine which lab tests are relevant for you, leading to more effective monitoring and targeted interventions," Cohen said.
Cutting through supplement confusion
In an industry flooded with trendy peptides, exotic compounds and celebrity-endorsed formulations, Cohen's advice was refreshingly practical. He cautioned against chasing novel interventions when basic nutrient deficiencies remain unaddressed.
"Ignore guru advice and trendy new supplements or peptides unless they address your specific deficiencies or needs," he said. "Basic nutrients often offer greater benefits than fancy interventions when you are deficient."
This approach requires considering supplements from multiple angles simultaneously. When evaluating something like curcumin, for example, one should assess its impact on symptoms, DNA risks, laboratory findings, lifestyle factors and personal health goals, all synthesised through AI-enhanced analysis.
The limits of conventional genomic medicine
While acknowledging that conventional medicine uses genetics for pharmacogenomics and identifying rare disease mutations, Cohen argued these applications represent only a fraction of genetics' potential utility.
"Conventional medicine uses genetics for pharmacogenomics or rare disease mutations, but this is limited," he said. "Precision health requires a broader, integrated approach."
This broader approach, he suggested, would incorporate nutrigenomics (how genetic variants influence nutrient metabolism and requirements) alongside traditional genetic screening, providing actionable insights that extend far beyond identifying monogenic diseases.
Practical steps for implementation
For those looking to adopt a precision health approach, Cohen outlined a straightforward process. First, obtain comprehensive genetic testing rather than panels examining only a few variants. Second, begin tracking relevant biomarkers through blood, urine or other laboratory tests. Third, document symptoms, existing conditions and health goals. Finally, integrate this information using AI tools or platforms designed for health data synthesis.
The tracking element, he stressed, is particularly crucial. "I often forgot why I started taking certain things and would just stop," Cohen admitted about his earlier biohacking attempts. "Now, you can use programs like Self Decode, which keep everything tracked for you."
The future of personalised medicine
Looking ahead, Cohen expressed optimism about increasing integration between different health data platforms and large language models. He noted that his company is working on such integrations, as are other organisations in the precision health space.
However, he emphasised that the quality of underlying data remains paramount. "Core underlying data is vital. Quality inputs yield quality outputs," he said.
The vision Cohen presented is one where generic health advice becomes obsolete, replaced by recommendations tailored to each person's unique genetic makeup, current biomarker status, symptoms, lifestyle and goals. In this future, preventing disease would be as simple as identifying genetic weaknesses early and addressing them before they manifest as illness.
As the wellness industry continues to grow, Cohen's message offers both a critique of its current celebrity-driven model and a roadmap for its evolution. Whether the future of health optimisation lies in AI-enhanced precision medicine or some other approach remains to be seen, but Cohen's two decades of systematic self-experimentation suggest that understanding individual biology beats following anyone else's blueprint.
Part of Hololife Summit 2025 Tokyo