What the data says, in numbers.
Bar = min–max concordance for that item across the compared panels. Weakest items in most panels: CRP and GGT (75–82%).
Repeated sessions build a personal baseline: averaging several sessions matched the blood panel better than a single same-day reading, and trends persisted across weeks.
Continuous research, core numbers only.
CARNA AI keeps comparing its wellness reference indicators with conventional reference measures across markets, and adds each new study here as it completes.
Blood panel vs CARNA — two-subject pilot
2 adults · 1 session each · 13 blood items
- Overall agreement 87.5% / 85.1%
- HDL-C 96.5–97.2%, glucose 93.1–95.8%, triglycerides 91.4–92.4%
Reference ranges normalized to 0–100; agreement = closeness of normalized scores.
Blood panel vs CARNA — male, 60s
1 adult · 1 session · 257 markers · 22 items compared (5 direct)
- Overall 82.3% · direct-match items 96.2%
- Bilirubin 96–98%, triglycerides 95.6%, glucose/HbA1c 93.1%, LDL-C 92.8%
Normal/abnormal interpretation match; complete = 100%, partial = 75–85%.
Blood panel vs CARNA — male, 60s
1 adult · 1 session · 257 markers · 19 items
- Overall 89.1% · direct-match 96.5–97.2%
- Liver enzyme & bilirubin items 98–99%, BUN 97%, creatinine 96%
Same method as above; mean discrepancy 10.9%.
Blood panel vs CARNA — female, 50s
1 adult · 1 session · 261 markers · 20 items
- Overall 81.6% · direct-match 94.7%
- Bilirubin 95–97%, triglycerides 93.8%, creatinine 92.1%
Mean discrepancy 18.4%.
Post-surgery recovery case — blood vs CARNA
1 adult (male, 60s) · ~3 months after surgery · 18 items
- Overall 85.9% · direct-match 96.4%
- 7 complete / 11 partial matches; HbA1c 95.9%, total cholesterol 94.2%
Complete ≥ 90%, partial ≥ 70% after normalization.
Blood panel vs 6 repeated sessions — female, 30s
1 adult · 6 sessions over 2 months · ~260 markers · 1 blood panel
- Direct items: same-day 87.6%, 6-session mean 84.3%
- Liver panel: 6-session mean 89.3% beat same-day 83.6%
- HbA1c 92.9%, creatinine 95.1%, total cholesterol 90.7%, B12 90.5%
Categorical (normal/low/high) and directional agreement, not numeric equality.
8-session longitudinal — male, 50s
1 adult · 8 sessions over 57 days · ~255 markers each
- Wellness index range 49.4–71.2; 5 same-window sessions mean 59.7
- Intra-individual CV: liver-fat index 6.9%, pH index 9.8%
Natural intraday variability of 10–20% expected; fixed measurement time and 2–3 h after meals recommended.
Group screening, users aged 50+
25 adults (50–82 y, mean 61.6) · 1 session each
- Wellness index mean 64.3 (50.8–75.4); 50s 67.8 vs 70+ 59.2
- 3.84 attention items per person; BMI mean 24.9
Descriptive population summary; no blood comparison.
4-session repeat — male, 50s
1 adult · 4 sessions over 2 months
- Wellness index 67.0 → 62.7 → 67.5 → 71.7; attention items 4/5/4/3
Persistent trends tracked across sessions (e.g. bone-turnover and vitamin D indices).
5-session repeat — male, 40s
1 adult (BMI 26.4) · 5 sessions over 4 weeks
- Wellness index 66.0 → 55.2 → 56.0 → 61.3 → 67.7 (+12.5 from low)
- Liver-fat index 7.2–8.9, pH index 31.3–37.0 — consistently out of range in all 5
Illustrates week-to-week trend sensitivity.
7-session repeat over 70 days — male, 30s
1 adult · 7 sessions · 258+ indicators
- Wellness score mean 62.0, SD 5.2, range 55.1–68.0
- Liver-fat index 7/7 sessions out of range (mean 7.88, SD 0.40)
- Morning vs afternoon score 63.8 vs 61.3; HR 88–92 bpm stable
Repeatability and time-of-day effect.
Consumer genetic test vs 14-session CARNA mean — female, 50s
1 adult · 14 sessions over 10 weeks · 16 items
- Mean agreement 76.9%; selenium 93.6%, iron 93.4%, vitamin D 92.9%
- Wellness index 58.9–77.8 across 14 sessions
Genetic risk tier mapped to a reference-range position; agreement = 100 − |difference|.
Most comparisons are single-person case studies (n = 1); the largest group study is n = 25. Concordance definitions differ between reports (normalized-score closeness vs categorical/directional agreement), so figures are indicative rather than pooled statistics. All summaries exclude names, contact details and institution names; participants consented to research use. Figures are published for transparency about how CARNA AI’s wellness references behave in practice — they are not diagnostic performance claims.