Our Methodology

NutriGenAI uses two complementary, peer-reviewed frameworks — each applied where it was designed to be used.

Dual-Framework Approach

NutriGenAI employs two distinct, peer-reviewed tools at the appropriate level of analysis:

Individual Study Quality

Q-Genie v1.1 — Purpose-built for assessing the methodological quality of genetic association studies. Scores 11 dimensions (1–7 each), yielding a total out of 77.

Multi-Study Evidence Synthesis

Grimaldi et al. (2017) — Designed for evaluating the overall strength of evidence for a gene × diet interaction across multiple studies.

Q-Genie v1.1

The Quality of Genetic Association Studies (Q-Genie) tool was developed by an international panel of experts and validated for assessing the methodological quality of genetic association studies.

Unlike reporting checklists (e.g. STROBE, CONSORT) which measure transparency, Q-Genie evaluates actual methodological quality.

Sohani ZN et al., BMC Medicine (2015)

11 Quality Criteria (each scored 1–7)

Q1
Rationale for Study

Adequacy of hypothesis and scientific rationale for chosen genes.

Q2
Outcome Definition

Appropriateness of outcome classification, participant sampling, and assessor blinding.

Q3
Comparison Groups

Control group selection, comparability, and description of selection procedures.

Q4
Technical Genotyping

DNA source, genotyping platform, call rates, HWE testing, and imputation methods.

Q5
Non-Technical Genotyping

Blinding of genotyping assessors, batch consistency, and sample randomisation.

Q6
Other Sources of Bias

Disclosure and discussion of time-lag bias, attrition bias, and other potential biases.

Q7
Sample Size & Power

Adequacy of sample size and whether a priori power analysis was conducted.

Q8
A Priori Planning

Analysis plan description, reporting completeness, and pre-specification of subgroups.

Q9
Statistical Methods

Confounding control, missing data handling, and multiple testing corrections.

Q10
Genetic Assumptions

Testing of haplotype inference, relatedness, and verification of sex/ethnicity.

Q11
Inferences from Results

Whether conclusions are proportionate to the evidence and limitations acknowledged.

Scoring: Total score = sum of Q1–Q11 (range 11–77)

With control groups: ≤35 Poor · 36–45 Moderate · >45 Good

Without control groups: ≤32 Poor · 33–40 Moderate · >40 Good

Grimaldi Framework (Phase 2 Synthesis)

The Grimaldi et al. (2017) framework was developed within the EU FP7 Food4Me project to establish minimum standards of evidence for the scientific validity of nutrigenetic gene × diet interactions.

NutriGenAI applies this framework when synthesising multiple studies investigating the same SNP, assessing the cumulative evidence across five dimensions.

Grimaldi et al., Genes & Nutrition (2017)

5 Synthesis Dimensions

25%
Consistency of Mechanistic Direction

Do studies agree on which genotype group shows differential response?

20%
Combined Sample Size & Power

Pooled sample size across studies and statistical power.

25%
Biological Plausibility

Collective mechanistic evidence from different angles.

15%
Dose-Response Evidence

Evidence of genotype gradient across studies.

15%
Study Quality Distribution

Distribution of Q-Genie quality scores across the study set.

Evidence Classification:

A. Convincing>90%
B. Probable66–90%
C. Possible33–66%
D. Insufficient<33%

How They Work Together

Upload Paper(s)

Each paper assessed individually

Q-Genie Assessment

11 criteria, quality score

Grimaldi Synthesis

Multi-study evidence grade