The letter N holds almost sacred influence in most disciplines of research. For many researchers, a quick glance at the value of N can determine whether a scientific paper is worth reading.
That’s because N stands for the number of subjects studied in the research, or the sample size. Intuitively, a higher N is desirable, as it enhances the ability to generalize findings across a larger group.
For example, since it’s fall, say you and a couple of friends are enjoying Pumpkin Spice Lattes (PSL) at Starbucks. However, something tastes different this year. Is that a hint of chocolate? One friend agrees, but the other isn’t sure. Curious, you approach other customers having the same drink. It becomes a game of statistics. The more people who confirm your hunch, the stronger the likelihood that a dash of cocoa powder has indeed been added to the recipe.
Sample size, accordingly, can make or break an experiment— especially in Randomized Controlled Trials (RCTs) [1], the historical gold standard in medical research.
A Layman's Introduction to Randomized Controlled Trials
Most medications and health services you have ever been prescribed were tested in RCTs to gain regulatory approval from authorities like the FDA. RCTs are studies of outcomes, such as whether drug A improves symptoms in patients with disease X.
While there are various RCT designs, the overall methodology is consistent. You take your N and randomly split it so that one group receives the treatment being studied, and the other receives either the existing standard of care or a placebo (this is called the control group). The idea is to recruit individuals in both groups with similar characteristics so that any differences in outcomes can be directly attributed to whether or not they received the treatment [2].
As such, RCTs are valuable because they can establish causality — not just correlation.
To explain in simpler terms, let’s head back to Starbucks for a moment. Imagine that cocoa powder has indeed been added to the PSL recipe — on the advice of a new intern.
Unfortunately, initial customer reviews indicate dissatisfaction with the latte, and the intern fears that the cocoa powder might be the issue.
To figure this out, the intern invites a thousand Starbucks fans and randomly divides them into two groups. Group A gets PSL with cocoa powder, and Group B gets PSL without it — everything else is identical. The outcome being studied is simple: “Did you enjoy your PSL today?”. If statistical analysis shows lower satisfaction within Group A, the intern can conclude that the cocoa powder is the likely culprit. If not, the problem lies elsewhere, and the intern can breathe a sigh of relief.
While oversimplified above, the line of thinking behind such RCTs is solid and has been instrumental in establishing evidence for numerous treatments over the years.
The Shortcomings of Randomized Controlled Trials
No research method is perfect, and clinical researchers are well aware of that when conducting RCTs. RCTs provide solid estimates of average effects, but averages don’t apply to everyone.
When a medication passes an RCT, it means that, on average, the treatment group showed better outcomes than the control group. But this doesn’t mean that every participant benefitted from the treatment. It only means that more people in the treatment group responded positively compared to those in the control group. More doesn’t mean all. The reality is that some individuals may not respond, and a few may even be harmed by the treatment.
Take conditions like depression, for example. Even with medications that have successfully passed RCTs, a significant portion of patients — between 30–40% — still don’t respond to treatments [3]. This teaches us that variability, or heterogeneity, in response to treatments is the rule rather than the exception. Factors like genetics, metabolism, microbiome composition, and lifestyle choices all contribute to this variability, and currently, RCTs are unable to account for every possible factor driving such individual differences.
To contextualize this, let’s take one final trip to the Starbucks Verse. Say the intern’s RCT showed that everyone in the group who drank PSL with cocoa loved it. The intern feels on top of the world until they receive an email from the Chief Recipes Officer (CRO; we made it up). The CRO, who was not part of the study, wants to try the new recipe. If the CRO doesn’t like it, the intern’s job could be in jeopardy.
The problem is no matter how well the new latte performed in the intern’s trial, there’s a non-zero chance the CRO dislikes it. Why? Because the CRO wasn’t included in the intern’s N. N can’t include everyone.
This highlights the inherent limitations of RCTs. While they provide invaluable evidence for populations, they fall short of predicting individual responses.
So, beyond RCTs, could there be a way to safely generate evidence that’s both reliable and tailored to the individual?
N-of-1 Trials: Evidence-Based Medicine by You for You
If N can’t equal everybody, conveniently enough, an effective way to tailor clinical trials to individuals is to conduct trials with a sample size of just one: N-of-1 clinical trials [4].
This approach may initially seem paradoxical. After all, we have discussed how a larger sample size increases the statistical power of a study and promotes the generalizability of results to the population. However, that is not the primary objective here. N-of-1 trials prioritize the individual. While they can sometimes contribute to population data (we’ll learn how a bit later), their main goal is to generate evidence on how a specific individual responds to a particular therapy.
This leads to an important question: who is the control in these studies? Well, the individual acts as both treatment and control prongs of N-of-1 trials. This makes sense because, in RCTs, the treatment and control groups are matched as closely as possible to minimize differences in outcomes unrelated to the intervention. In an N-of-1 trial, you effectively eliminate these interindividual differences entirely — because the same person is being tested under both conditions.
The key design feature here is the crossover method. In an N-of-1 trial, the individual switches between receiving the treatment and a placebo multiple times, often with a washout period in between. For instance, in a hypothetical trial testing a new depression medication, a participant might receive the drug for four weeks, switch to a placebo for four weeks, and repeat this process several times. Comparing the results during each phase helps determine how effective the treatment is for that specific person.
An N-of-1 trial can also test multiple treatments or different doses of the same treatment, which can lead to markedly different outcomes. The evidence produced is of high quality. In fact, the Oxford Centre for Evidence-Based Medicine classifies N-of-1 trials as Level 1 evidence — comparable to or even higher than the standard of evidence generated by RCTs [5].
All in all, these trials show that you might just be the best subject for studying your own health.
When to Use N-of-1 Trials
You might be wondering, if N-of-1 trials are so reliable, why haven’t you heard more about them?
That’s because N-of-1 trials aren’t suitable for every clinical scenario. These trials require longer timeframes to collect sufficient data points, which makes them impractical for acute events like accidents or infectious diseases.
Instead, N-of-1 trials are ideal for chronic conditions, which encompass a wide range of diseases that significantly impact quality of life. These conditions are also highly variable in how they present, making N-of-1 trials particularly useful in tailoring treatment to individuals.
Another advantage is their application in patients with multimorbidities — those dealing with more than one chronic illness. Historically, such cases have been excluded from RCTs, with exclusion rates as high as 73% [6]. N-of-1 trials, however, are well-suited to handle this complexity, offering a way to generate reliable evidence for these patients.
So whoever is excluded from RCTs, N-of-1 trials can make sure research doesn’t leave them behind. As Professor Sunita Vohra of the Faculty of Medicine at the University of Alberta, a leading N-of-1 trials expert, says:
“We are who we are. Research needs to wrap itself around us. And that’s what N-of-1 allows.”
Inspiring N-of-1 Case Studies
Hundreds of N-of-1 studies have been published, covering a wide range of conditions — from diseases of the digestive system to mental health disorders [7]. Let’s explore some of these case studies to see just how impactful these tailored trials can be.
1- Asthma Patient Paves the Way for N-of-1 History
N-of-1 trials first came to prominence in 1986 at McMaster University in Canada [8].
There, researchers were investigating a case of poorly controlled asthma. The patient had been prescribed a combination therapy that included the drug theophylline, which the researchers had some concerns about.
To address this, they designed one of the first N-of-1 trials to evaluate theophylline’s impact on the patient. After just two paired phases of theophylline and placebo, the patient chose to end the trial early. The reason was clear: theophylline worsened his symptoms and withholding it led to better asthma control.
This trial proved transformative, leading to improved symptoms, reduced drug burden, and lower costs. Best of all, its success encouraged researchers at McMaster to pursue further N-of-1 trials, resulting in 57 completed within two years, with a definitive therapeutic answer found for 88% of the patients studied [7].
2- N-of-1 Trial Helps Doctor Avoid “Cruel” Diet
One of the great advantages of N-of-1 trials is that they can often be self-conducted, especially by individuals with the necessary technical knowledge. This was the case for Dr. Alexander K. Smith of UC San Francisco, whose self-ran N-of-1 trial proved invaluable [9].
Dr. Smith was diagnosed with eosinophilic esophagitis, a chronic condition causing inflammation in the esophagus and making it difficult to swallow. His gastroenterologist recommended the standard 6-food elimination diet, which required removing dairy, wheat, soy, eggs, seafood, and nuts from his diet for six to eight weeks, gradually reintroducing the foods to identify the trigger.
Dr. Smith found the diet to be overly restrictive, calling it “cruel.” However, through his own literature review, he learned that dairy was the most common food trigger for the condition. To test this, he conducted an N-of-1 trial by eliminating dairy for eight weeks. His symptoms resolved entirely. He then reintroduced dairy by eating ice cream, butter, and milk — his symptoms quickly returned.
Thanks to his N-of-1 trial, Dr. Smith identified dairy as his sole trigger. Today, he avoids just one food group rather than six, allowing him to maintain a manageable diet without the extreme restrictions of the full elimination protocol.
3- Dose Reduction Increases Drug Efficacy for Cancer Patient
N-of-1 trials hold significant promise when augmented by AI, as demonstrated in our next case.
A patient with metastatic castration-resistant prostate cancer underwent an AI-powered N-of-1 trial in collaboration between UCLA Medical Center and the National University of Singapore [10].
The patient had been on high-dose combination chemotherapy for six weeks. Using an AI platform, researchers calculated an optimal dosage tailored to the patient’s condition. Contrary to conventional methods, where dosage is often increased until drug resistance develops, the AI suggested a 50% reduction in the chemotherapy dosage.
Unexpectedly, the patient responded remarkably well to the reduced dose. His PSA levels, a key biomarker for prostate cancer, reached their lowest levels, and after 16 months, his tumor shrank in size.
This trial marked a pivotal step forward in using AI and N-of-1 trials to enhance drug efficacy, reduce side effects, and lower treatment costs.
How to Get Started with Your Own N-of-1 Trials
N-of-1 trials require a certain degree of rigor to generate quality evidence. For example, placebos are used in most cases to control for psychological factors that may influence outcomes. Blinding, a technique in clinical trials that prevents both the subject and investigator from knowing whether the treatment administered is active or a placebo, is also standard practice.
However, on a case-by-case basis, it is possible to trade some elements of rigor for practicality. Indeed, some trial designs have been published that work around the expensive and time-consuming process of arranging placebos [11]. These trials are especially valuable for individuals seeking to optimize their lifestyles — across the board and beyond just medications.
For example, if you are curious whether CrossFit or spinning class provides better exercise for your body, start an N-of-1 trial. Engage in CrossFit for say two weeks, then switch to spinning for another two weeks. Repeat the process as many times as you wish and review the data.
The key is to have a clear outcome in mind — whether it’s improved performance or stamina — and track changes in its corresponding biomarkers with each intervention. That way you could make real, data-backed lifestyle improvements.
In the same way, you can use N-of-1 trials to test how different carbohydrates impact your blood glucose levels or how various longevity supplements influence your VO2 max, a widely used functional longevity biomarker. As long as you know what to measure and how, N-of-1 trials are powerful tools for understanding your own body.
On a broader level, running your own trials means you’re not a lab rat. It means you have a full grasp of the scientific method and know how to minimize bias to produce meaningful results. The more individuals attain this level of science literacy, the more democracy thrives in society.
Rejuve.AI: The Future of N-of-1 Trials
At Rejuve.AI, we are believers, advocates, and enablers of N-of-1 trials. With our in-development Longevity App, health enthusiasts can expect unprecedented N-of-1 accessibility.
Firstly, we’ve compiled hundreds of biomarkers for you to track a variety of health outcomes.
Our Longevity App will sync with your wearable devices to monitor your digital biomarker levels, and you can seamlessly upload your lab biomarkers for analysis as well. The optimal levels for all these biomarkers will be provided, allowing you to track how each intervention brings you closer to your health goals.
Your results will be displayed as a timeline of biomarker values and the actions you’ve taken over time, allowing you to observe trends, trade-offs, and the effects of different intervention combinations. This robust infrastructure will give everyone the tools to run decentralized and reliable N-of-1 trials right from the App.
That’s just on an individual level. Remember when we said N-of-1 trials can provide value for the wider population?
That is where our cutting-edge AI models come into play. Our Bayesian Expert offers vast applicability to N-of-1 trials. The power of the Bayesian approach lies in its ability to maximize the use of information from each user, as well as to incorporate reliable prior information into the statistical model. This method allows each N-of-1 trial to inform the next one.
With these models, we can aggregate N-of-1 trials to yield estimates of population treatment effects. The more users conduct N-of-1 trials on our platform, the higher the generalizability of the generated data to the wider population. This approach supports our goal of simultaneously fostering healthier individuals and populations.
So stay tuned to all our channels for more updates on the Longevity App and buckle up for a whole new era in wellness — from N-of-1 to N-of-all!
References:
[1] Institute of Medicine (US) Committee on Strategies for Small-Number-Participant Clinical Research Trials. (2001). Small Clinical Trials: Issues and Challenges. In C. H. Evans & S. T. Ildstad (Eds.), PubMed. National Academies Press (US). https://pubmed.ncbi.nlm.nih.gov/25057552/
[2] Schultz, A., Saville, B. R., Marsh, J. A., & Snelling, T. L. (2019). An introduction to clinical trial design. Paediatric Respiratory Reviews, 32, 30–35. https://doi.org/10.1016/j.prrv.2019.06.002
[3] Gloster, A. T., Rinner, M. T. B., Ioannou, M., Villanueva, J., Block, V. J., Ferrari, G., Benoy, C., Bader, K., & Karekla, M. (2020). Treating treatment non-responders: A meta-analysis of randomized controlled psychotherapy trials. Clinical Psychology Review, 75, 101810. https://doi.org/10.1016/j.cpr.2019.101810
[4] Vohra, S. (2016). N-of-1 trials to enhance patient outcomes: identifying effective therapies and reducing harms, one patient at a time. Journal of Clinical Epidemiology, 76, 6–8. https://doi.org/10.1016/j.jclinepi.2016.03.028
[5] OCEBM Levels of Evidence Working Group*. “The Oxford Levels of Evidence 2”.Oxford Centre for Evidence-Based Medicine. https://www.cebm.ox.ac.uk/resources/levels-of-evidence/ocebm-levels-of-evidence
* OCEBM Levels of Evidence Working Group = Jeremy Howick, Iain Chalmers (James Lind Library), Paul Glasziou, Trish Greenhalgh, Carl Heneghan, Alessandro Liberati, Ivan Moschetti, Bob Phillips, Hazel Thornton, Olive Goddard and Mary Hodgkinson
[6] Rothwell, P. M. (2005). External validity of randomised controlled trials: “To whom do the results of this trial apply?” The Lancet, 365(9453), 82–93. https://doi.org/10.1016/s0140-6736(04)17670-8
[7] Mirza, R., Punja, S., Vohra, S., & Guyatt, G. (2017). The history and development of N-of-1 trials. Journal of the Royal Society of Medicine, 110(8), 330–340. https://doi.org/10.1177/0141076817721131
[8] Guyatt, G., Sackett, D., Taylor, D. W., Ghong, J., Roberts, R., & Pugsley, S. (1986). Determining Optimal Therapy — Randomized Trials in Individual Patients. New England Journal of Medicine, 314(14), 889–892. https://doi.org/10.1056/nejm198604033141406
[9] Smith, A. K. (2019). A Case for n-of-1 Trials. JAMA Internal Medicine, 179(3), 452. https://doi.org/10.1001/jamainternmed.2018.7183
[10] Pantuck, A. J., Lee, D.-K., Kee, T., Wang, P., Lakhotia, S., Silverman, M. H., Mathis, C., Drakaki, A., Belldegrun, A. S., Ho, C.-M., & Ho, D. (2018). Modulating BET Bromodomain Inhibitor ZEN-3694 and Enzalutamide Combination Dosing in a Metastatic Prostate Cancer Patient Using CURATE.AI, an Artificial Intelligence Platform. Advanced Therapeutics, 1(6), 1800104. https://doi.org/10.1002/adtp.201800104
[11] Smith, J., Yelland, M., & Del Mar, C. (2015). Single Patient Open Trials (SPOTs). The Essential Guide to N-of-1 Trials in Health, 195–209. https://doi.org/10.1007/978-94-017-7200-6_15