N-of-1 Research Design and Safety Monitoring¶
This page defines a generic framework for prospectively measuring a single-subject intervention. It is a method for generating feasibility signals and identifying confounders; it does not establish efficacy, prescribe treatment, or justify exposing a person to an unvalidated compound, engineered organism, off-label drug, or combination.
The framework's evidence level is Mechanistic Extrapolation from standard clinical-monitoring and n-of-1 methods. Clinical decisions, prescription changes, acute-flare care, and abnormal results require qualified medical oversight.
1. Appropriate scope¶
An n-of-1 design is useful when:
- the intervention and exposure are already sufficiently characterized for the proposed context;
- the primary outcome can change within a practical observation window;
- baseline care can remain stable;
- safety monitoring and halt criteria are defined before exposure; and
- a single-subject result is explicitly treated as non-generalizable.
It is not an authorization path for experimental biologics, engineered strains, unapproved delivery systems, or medication changes. Those require their own preclinical, regulatory, and clinical gates.
2. Prespecify the experiment¶
Define these fields before collecting outcome data:
| Element | Required question |
|---|---|
| Hypothesis | What mechanism predicts the outcome, and what result would falsify it? |
| Intervention | What source, form, route, exposure, and timing are being studied? |
| Comparator | Baseline, withdrawal, crossover, matched control, or another appropriate condition? |
| Primary outcome | Which single measure decides the result? |
| Secondary outcomes | Which measures explain mechanism, adherence, or safety without changing the primary decision? |
| Cadence | When should exposure, target engagement, outcome, and safety be measured? |
| Confounders | Which diet, medication, sleep, training, infection, or behavior changes must remain stable or be logged? |
| Halt criteria | What result stops exposure and triggers clinical review? |
| Decision rule | What result means pass, revise, null, or stop? |
A retrospective hypothesis is not prospective evidence. If multiple variables change together, interpret the result as a confounded observation rather than attributing it to one arm.
3. Match cadence to mechanism¶
| Mechanism class | Typical observation window |
|---|---|
| Drug at steady state | 4–6 weeks after a change |
| Stored vitamin or mineral correction | 8–12 weeks |
| Antibody-mediated effect | 3–6 months |
| Microbiota-mediated intervention | 6–8 weeks |
| Acute pharmacodynamic effect | Hours to days |
| Tissue-level adaptation | Weeks to months |
| Long-latency risk modification | Annual or longer; surrogate markers may be more frequent |
Use the longer of the intervention's equilibration time and the outcome's response time. Add earlier safety measurements when the risk profile requires them.
4. Measurement hierarchy¶
Every experiment should distinguish four layers:
- Input identity and potency — what was actually delivered?
- Exposure — did the relevant compartment receive it?
- Target engagement — did the proposed mechanism change?
- Outcome — did the prespecified clinical or biological measure change?
A negative outcome with unverified input or exposure does not falsify the mechanism. A biomarker change without a matched outcome does not establish benefit.
Common safety measurements include CBC with differential, CMP, and hs-CRP where relevant. Additional tests should follow the intervention's known risk surface rather than a universal stack panel.
5. Gout and NLRP3 measurement map¶
These markers can help localize a signal, but none independently selects a compound or treatment.
| Marker | Research axis | Interpretation limits |
|---|---|---|
| Serum C5a (+ desArg) | CP0 complement priming | Strict pre-analytics: cold-chain EDTA, processing within 30 minutes, and frozen storage are required because warm transit can generate C5a in vitro. Onset and resolution measurements can estimate a decline slope, but the slope remains a mechanistic hypothesis. |
| Urinary LTE4 | CP6a leukotriene flux | A pharmacodynamic readout for 5-LOX engagement; it does not by itself establish flare benefit or identify why exposure failed. |
| Plasma SPMs (RvD1, MaR1) | CP5b active resolution | Research-grade LC-MS/MS; low values can reflect substrate, conversion, timing, or analytical limitations. |
| hs-CRP | Integrated systemic inflammation | Non-specific; cannot identify a chokepoint without mechanism-specific measurements. |
| Serum urate | Urate balance | Does not distinguish production, renal handling, intestinal export, or luminal degradation without additional measurements. |
Optional genotype stratification such as CFH Y402H or ABCG2 Q141K must be prespecified. A genotype association does not establish a carrier-specific intervention response.
6. Home and laboratory formats¶
| Marker | Lower-friction format | Reference format | Method note |
|---|---|---|---|
| Serum urate | Capillary UA meter | Venous serum UA | Pair the meter with a venous draw to estimate device-specific offset; emphasize within-device change. |
| Omega-3 index | Mail-in dried-blood spot | Venous RBC fatty-acid profile | Event-linked sampling is feasible; use the same format across comparisons. |
| Genotype | Existing array data or single-SNP assay | Clinical PCR | One-time measurement; trial-grade stratification requires an appropriate-quality assay. |
| C5a | None | Specialty venous assay | Pre-analytical handling is load-bearing. |
| Urinary LTE4 | Specialty urine collection | Specialty laboratory | Not a routine home test. |
| Plasma SPMs | None | Research LC-MS/MS | Limited availability and timing sensitivity. |
| hs-CRP | Some home immunoassays | Venous hs-CRP | Venous measurement is preferred for low-range interpretation. |
Use lower-friction formats for trajectory sampling only when analytically fit. Use reference methods for calibration anchors and decisions that require clinical-grade accuracy.
7. Daily outcome and confounder log¶
Keep entries brief enough to complete every day:
- timestamp and adherence;
- primary outcome on a fixed scale;
- relevant symptoms using the same scale throughout;
- medication, diet, alcohol, sleep, training, infection, and travel deviations;
- adverse events; and
- a free-text field for unexpected signals.
Examples include 0–10 severity scales, the Bristol Stool Scale, joint circumference, or a binary event count. Choose the measure before the experiment and do not redefine success after seeing the data.
8. Safety and halt criteria¶
The following existing project thresholds are conservative research triggers, not a substitute for clinical judgment:
- New GI bleeding — halt and seek same-day care.
- ALT or AST above 2× the upper limit of normal — halt and obtain clinical review.
- eGFR decline above 15% from baseline or a creatinine rise — halt and evaluate.
- New rash, urticaria, angioedema, or anaphylaxis — halt immediately; airway symptoms require urgent care.
- Unexplained weight loss above 5 lb over four weeks — halt and evaluate.
- New unexplained fever — halt and seek care.
- hs-CRP doubling from baseline — halt and confirm with clinical review.
- Persistent diarrhea over 72 hours — halt and evaluate for infection or dysbiosis.
- Any new severe symptom absent at baseline — halt and evaluate.
Define additional intervention-specific criteria before exposure. “Halt” applies to the investigational intervention; prescribed therapy should not be stopped without the responsible clinician unless emergency guidance requires it.
9. Privacy and consent¶
Raw laboratory results, daily logs, intervention plans, and identifiable health information do not belong in the public repository. Store them in a private repository, encrypted volume, or other controlled location with appropriate backup.
Data about another person requires their documented agreement. Public summaries must be de-identified and should report study design, deviations, qualitative outcome direction, and limitations without exposing raw personal health data.
10. Interpretation¶
Classify the result using the prespecified chain:
- Input failure — identity or potency was not as specified.
- Exposure failure — the intended compartment did not receive the intervention.
- Target-engagement failure — exposure occurred but the proposed mechanism did not move.
- Outcome null — target engagement occurred without the primary outcome.
- Safety stop — adverse findings prevent interpretation or further exposure.
- Feasibility signal — exposure, target engagement, and outcome moved in the predicted direction, subject to n=1 limitations.
An n=1 feasibility signal can motivate a controlled study. It cannot establish population efficacy, comparative effectiveness, a personalized treatment rule, or a dose recommendation.
11. Worked design patterns¶
Timing or formulation comparison¶
Use a characterized intervention, keep total exposure constant, randomize or alternate timing/formulation conditions where practical, and measure a short-latency outcome. A prior fungal-enzyme timing observation across approximately 30 meals illustrates how meal composition and lying flat after meals can emerge as confounders. The observation was single-subject, unblinded, and uncontrolled; it informs formal study design but does not establish a dosing framework.
Biomarker-linked mechanism study¶
For a candidate transporter or inflammasome mechanism, pair the clinical outcome with input verification, exposure, and a direct functional readout. For example, stool SCFAs alone are an exposure proxy and cannot establish epithelial ABCG2 trafficking or urate flux.
Ex vivo challenge¶
An ex vivo MSU challenge can measure within-subject IL-1β response under prespecified biological strata. It remains a subject-specific feasibility signal and cannot replace the tiered cell, organoid, or animal experiments in validation-experiments.md.