Download App
Halflife Labs

Methodology

How Halflife Labs collects real-world protocol data, and how it sources and validates pharmacokinetic values for 45 compounds.

Last updated: May 2026  ·  Halflife Labs v1.0 Dataset
What this dataset is: Observational real-world data from opted-in Halflife app users. It captures what people actually do — not what they were instructed to do in a controlled trial. This distinction is both the dataset's greatest value and its most important limitation.

1. Data Collection

Data is collected exclusively from Halflife app users who have explicitly enabled the research contribution toggle in their Profile screen. This toggle is off by default. No data is collected from users who have not actively opted in.

When a user logs a dose and has research contribution enabled, a snapshot event is transmitted to the Halflife Labs database. This snapshot contains only the fields listed in Section 2 — no additional data is captured, stored, or inferred beyond what is explicitly listed.

All data transmissions occur over HTTPS (TLS 1.3). The database uses row-level security with no cross-user queries possible at the application layer.

2. What We Collect

Each research snapshot contains the following fields:

Field Value Stored Why
device_idRandom UUID (cryptographic)Longitudinal cohort tracking — links snapshots over time without identity
active_compound_idsArray of compound IDsUnderstand what compounds are run simultaneously (the stacking question)
dose_logs_7d / 30dInteger countAdherence calculation without storing individual dose timestamps
adherence_rate_7dFloat (0.0–1.0)Core protocol adherence metric
weeks_on_primaryInteger (week number)Duration context without calendar date
avg_energy_7dFloat (1.0–5.0) or nullSubjective energy score trend
avg_appetite_7dFloat (1.0–5.0) or nullGLP-1 appetite suppression tracking
weight_delta_kgFloat (change from baseline)Weight trajectory — delta not absolute, preventing anthropometric profiling
weight_bmi_categoryCategory string (underweight / normal / overweight / obese)Cohort stratification without exact weight
onboarding_bundleProtocol category (glp1, trt, recovery…)User type segmentation
country_code2-letter ISO (e.g. "US")Geographic distribution — not city or region
snapshot_numberInteger (1, 2, 3…)Longitudinal position index — enables cohort retention analysis without timestamps

3. What We Explicitly Do NOT Collect

✕ Free text notes from dose logs
✕ Exact calendar dates or timestamps
✕ Absolute body weight in kg
✕ IP address or precise location
✕ Device identifiers (IDFV, IDFA)
✕ Apple Health biometric data
✕ Name, email, or any PII
✕ Sexual health compound data (excluded by design)

4. Re-identification Prevention

Week numbers instead of dates. All temporal data is expressed as week number of protocol (e.g. "Week 14") rather than calendar dates. This removes the ability to cross-reference health data with external calendars or events.

Delta weight instead of absolute weight. We store weight change from baseline (e.g. "-4.2kg") rather than absolute body weight. Absolute weight combined with protocol data could contribute to re-identification. A delta cannot.

BMI category instead of BMI. Similarly, categorical BMI ranges provide the clinical stratification we need for research without preserving the granularity that could contribute to identification.

Sexual health exclusion. Users whose primary onboarding bundle is sexual health (PT-141, Oxytocin) are excluded from research data collection entirely, regardless of opt-in status. We made this design decision early because the combination of sexual health compound use with any demographic or behavioural data — even highly anonymised — carries elevated re-identification and stigma risk.

5. Data Quality Considerations

Self-reported data. All data originates from user self-reporting within the app. It reflects what users logged, not verified clinical measurements. Dose amounts, compound names, and subjective scores (energy, appetite) are unverified.

Selection bias. Opted-in users may differ systematically from non-opted-in users and from the general population of GLP-1 and peptide users. Users who choose to contribute research data are likely more engaged, more experienced, and more protocol-consistent than average.

Survivorship bias. Users who stopped using the app stopped contributing data. Any longitudinal analysis reflects users who persisted — not those who discontinued, which may represent a systematically different population (more adverse effects, less efficacy, life circumstances).

Not a clinical trial. No randomisation, no control group, no blinding, no clinical oversight. All correlations observed in this dataset are associative — not causal. We do not claim any compound produces any outcome based on this data alone.

6. Data Retention and Deletion

Research snapshots are retained for up to 3 years from collection date. An automated cron job runs monthly to delete snapshots older than 36 months.

Because snapshots contain no personally identifying information, individual deletion on request is not technically feasible — we have no way to identify which records belong to a given user. Users who opt out of research contribution stop generating new snapshots immediately. Past snapshots remain in the dataset but no new data is added.

7. Research Partnerships

We provide dataset access to academic researchers, telehealth platforms, and pharmaceutical companies under formal Data Use Agreements that prohibit re-identification attempts, sub-licensing, and use beyond the agreed research scope. All partner data sharing is governed by our B2B Privacy Policy and Partner Terms of Service.

For research collaboration enquiries: contact@halflife-labs.com

Disclaimer: This methodology document describes the Halflife Labs v1.0 dataset architecture as of May 2026. Methodology may evolve as the dataset grows and research applications expand. Material changes will be reflected in updates to this document.
What this database is: A pharmacokinetic reference index covering peptides, GLP-1 receptor agonists, growth hormone secretagogues, testosterone compounds, and related bioactives — continuously expanded as new compounds are added. Every half-life value is sourced from a traceable primary document: FDA prescribing labels for approved compounds, peer-reviewed human PK studies where available, and animal PK studies where no human data exists. Each source tier is clearly labelled.

1. The Evidence Tier System

Every compound in the database is assigned one of three evidence tiers. The tier reflects the quality and origin of pharmacokinetic data, and is displayed on every compound page and applied to all derived app calculations — concentration curves, dosing interval logic, and vial inventory projections.

Tier 1 — FDA Prescribing Label

The prescribing label (package insert) is the primary source for all FDA-approved compounds. Labels contain human PK data from the clinical trials required for regulatory approval, reviewed and validated by the FDA. Half-life values drawn from prescribing labels carry the highest confidence in the database.

Examples: semaglutide, tirzepatide, liraglutide, dulaglutide, exenatide, testosterone enanthate, testosterone cypionate, gonadorelin, tesamorelin, PT-141, oxytocin, insulin analogues, and more.
Browse all Tier 1 compounds in the database →
Tier 2 — Published Human Pharmacokinetics

For research compounds with peer-reviewed human PK studies indexed in PubMed, half-life values are drawn from those publications. These studies typically involve small cohorts and are not subject to FDA review, but they provide direct human data far superior to animal extrapolation.

Examples: sermorelin, CJC-1295 (DAC and no-DAC), ipamorelin, GHRP-6, GHK-Cu, MK-677, IGF-1, IGF-1 LR3, and more.
Browse all Tier 2 compounds in the database →
Tier 3 — Animal Study

Where no published human PK data exists, values are sourced from peer-reviewed animal studies (typically rat or rodent). These values cannot be assumed to apply to humans. All Tier 3 values are accompanied by an explicit data quality warning on the compound page and in the Halflife app.

Examples: BPC-157, TB-500, KPV, LL-37, DSIP, Selank, Semax, MOTS-c, SS-31, GHRP-2, Hexarelin, PEG-MGF, SLU-PP-332, Epitalon, Dihexa, and more.
Browse all Tier 3 compounds in the database →

2. Two Distinct Half-Life Definitions

The app and database distinguish between two half-life concepts, because they serve different purposes and often differ substantially for the same compound.

Plasma half-life (t½). The time for plasma concentration to fall by 50% following administration. This is a direct pharmacokinetic measurement from blood concentration time-course data — it describes how quickly the body clears the compound from circulation.

Effective half-life. The functional dosing interval — how long the compound produces meaningful biological activity. For many compounds this diverges substantially from plasma half-life via two mechanisms:

Downstream signalling persistence
The compound binds a receptor, initiates a signalling cascade, then clears from plasma — but the downstream biological effects continue for hours or days. BPC-157: ~15 min plasma half-life, VEGFR2/Akt-eNOS effects persist 24–72 h. GHRP-2: ~15 min plasma, GH pulse sustained 2–3 h.
Structural half-life extension
Chemical modifications deliberately extend plasma half-life beyond the native peptide. Semaglutide: C18 fatty acid albumin binding → t½ ~165 h (weekly dosing). CJC-1295 DAC: Drug Affinity Complex → t½ ~8 days. Tesamorelin: modified GHRH → t½ ~38 min but once-daily dosing.
Key principle: Plasma half-life = pharmacokinetics (what the body does to the drug). Effective half-life = pharmacodynamics (what the drug does to the body). The Halflife app uses effective half-life for dosing interval logic and concentration curve calculations displayed to users.

3. Data Fields Per Compound

Each compound profile contains the following standardised fields:

Field Definition Source
half_life_plasmaTime for plasma concentration to fall 50%Primary literature (tier-labelled)
half_life_effectiveFunctional biological activity durationPK/PD literature and protocol data
compound_classOne of 8 functional classification categoriesHalflife Labs classification schema
mechanism_of_actionPrimary receptor target and downstream pathwayFDA label / peer-reviewed literature
administration_routesValid routes with route-specific PK notesFDA label / PK study methodology
fda_statusApproved / Research Use Only / ProhibitedFDA drug database, current as of May 2026
evidence_tierTier 1 / 2 / 3 per the system aboveAssigned by Halflife Labs
primary_citationDOI or PMC ID of primary PK sourcePubMed / FDA DailyMed

4. Source Hierarchy

When multiple sources exist for a compound, we use the highest-tier source as the primary value and document discrepancies in the compound's reference section. We do not average across sources.

1
FDA prescribing label (DailyMed)
Authoritative for approved compounds. Contains Phase I–III PK data reviewed by the FDA.
2
PubMed-indexed human PK studies
Peer-reviewed human pharmacokinetic studies. DOI or PMC ID cited for every compound at this tier.
3
PubMed-indexed animal PK studies
Used only when no human data exists. Flagged explicitly as animal data on every compound page and in the app.
Not used: manufacturer claims, forums, grey literature
Marketing materials, community forums, conference abstracts without full data, and non-peer-reviewed content are excluded.

5. Compound Classification System

Compounds are organised into 8 functional categories based on primary mechanism of action and intended application — not chemical structure alone.

Metabolic
GLP-1 receptor agonists and metabolic regulators acting via GLP-1R / GIP-R. e.g. semaglutide, tirzepatide, liraglutide, dulaglutide, exenatide, and more.
Recovery
Tissue repair peptides targeting angiogenesis, wound healing, and gut protection via VEGFR2, actin polymerisation, NF-κB. e.g. BPC-157, TB-500, KPV, LL-37, PEG-MGF, and more.
Growth Hormone
GH secretagogues via GHRH receptor or ghrelin receptor (GHS-R). e.g. sermorelin, CJC-1295, ipamorelin, GHRP-2, GHRP-6, hexarelin, tesamorelin, MK-677, IGF-1, IGF-1 LR3, and more.
TRT
Testosterone and endogenous hormone support via androgen receptor. e.g. testosterone enanthate, testosterone cypionate, gonadorelin, and more.
Insulin
Insulin and analogues acting via insulin receptor. e.g. rapid-acting (lispro, aspart), intermediate (NPH), long-acting (glargine, detemir, degludec), and more.
Nootropic
Cognitive and neurological compounds with varied CNS mechanisms. e.g. Dihexa, Semax, Selank, DSIP, flmodafinil, SLU-PP-332, and more.
Longevity
Anti-ageing and cellular health compounds targeting telomerase, mitochondria, and extracellular matrix. e.g. Epitalon, GHK-Cu, SS-31, MOTS-c, and more.
Sexual Health
Sexual function compounds via melanocortin receptor (MC4R) and oxytocin receptor. e.g. PT-141 (bremelanotide), oxytocin, and more. Excluded from research data collection by design.

6. Handling Conflicting or Uncertain Data

Multiple values across studies. Where published studies report different half-life values for the same compound (due to different routes, populations, or analytical methods), we use the value from the highest-tier source and document the range in the compound's reference section. We do not average across studies.

Route-dependent differences. Half-life differs significantly by administration route for many peptides. Where route-specific values exist, we list them separately. The default value used for app calculations corresponds to the most commonly studied or clinically used route, stated explicitly on each compound page.

Animal-to-human extrapolation. For Tier 3 compounds, we do not apply species scaling factors or allometric conversion to animal half-life values. We report the published animal value directly and label it clearly as animal data — applying an unvalidated scaling factor would introduce a false precision that misrepresents actual uncertainty.

No data available. For compounds where no published PK study exists for any species, we display "Not established — no published PK data" rather than estimating from structural analogs. We do not interpolate.

7. Database Scope

The database covers both FDA-approved compounds (prescribing label as primary source) and research-stage compounds (Tier 2 or Tier 3 sourcing), with new entries added on a rolling basis. The full list is browsable at halflife-labs.com/database.

FDA-APPROVED
e.g. Semaglutide · Tirzepatide · Liraglutide · Dulaglutide · Exenatide · Testosterone enanthate · Testosterone cypionate · Gonadorelin · Tesamorelin · PT-141 · Oxytocin · MK-677 · Insulin analogues · and more
RESEARCH-STAGE
e.g. BPC-157 · TB-500 · CJC-1295 (DAC) · CJC-1295 (no-DAC) · Ipamorelin · GHRP-2 · GHRP-6 · Hexarelin · Sermorelin · Epitalon · GHK-Cu · SS-31 · MOTS-c · KPV · LL-37 · Selank · Semax · DSIP · Dihexa · IGF-1 · IGF-1 LR3 · and more

8. What We Do Not Publish

✕ Dosing guidance or recommended doses
✕ Clinical protocols or cycle recommendations
✕ Efficacy claims for research compounds
✕ Safety or tolerability assessments
✕ Interpolated half-life values (no guessing)
✕ Compound sourcing or supplier information

9. Update Cadence

The compound database is reviewed on a rolling basis. New FDA approvals are incorporated within 30 days of prescribing label publication on DailyMed. PubMed saved searches for each compound name surface newly published PK studies. When a higher-tier source becomes available for an existing compound — for example, a human PK study published for a previously Tier 3 compound — the database entry is updated and the change is documented with a revision date on the compound page.

For corrections, additions, or newly published PK data we may have missed: contact@halflife-labs.com

Disclaimer: Compound database entries are pharmacokinetic reference data only — not medical advice, dosing guidance, or endorsement of any compound's use. FDA approval status reflects current labelling as of May 2026. Research compounds are clearly identified and distinguished from approved therapeutics throughout the database and application.