Methodology

Every product Nime scans gets a single Harmfulness score, 0–100, where lower is better. Tapping the number opens the breakdown: four risk measures Nime computes from the product’s ingredient list, nutrition panel, packaging, and category context. Below that, a plain-English explanation of every ingredient on the label — what it is, what it does, what the research says about it. This page covers how each piece works, what’s measured directly versus estimated, and where the limits are.

The four risk measures

The four measures are surfaced independently so you can see which one is driving a given score, rather than collapsing everything into a single rating.

1. Additives

Nime classifies every E-number and named additive in the ingredient list. Each classification combines two things: where the substance currently sits in EU regulation (its EFSA position, its tolerable daily intake, any restrictions under Regulation 1333/2008 and its amendments), and what independent research has found since regulators last assessed it.

Notable research informing the classifications: the French NutriNet-Santé cohort (BMJ 2023 on cardiovascular disease; PLOS Medicine 2024 on cancer; Lancet Diabetes & Endocrinology 2024 on type 2 diabetes), and Rondinella et al. (Nutrients, February 2025) on the gut-microbiome effects of ultra-processed food. EFSA’s 2026 guidance on mixture effects — applicable to new applications from 20 July 2026 — is part of how Nime flags products with unusually high additive variety, since individually-approved substances have rarely been tested in combination. See the blog post on emulsifiers and the underlying research for one category worked through in depth.

Regulatory approval does not mean unlimited consumption. The classification is a research-aware lens on top of EFSA’s position, not a contradiction of it.

2. Ultra-processed level

Nime classifies how processed a product really is from signals in its ingredient list and nutrition panel. Long lists, dense additive use (emulsifiers, stabilisers, flavour enhancers, colourings), refined-flour and refined-oil bases, and engineered ingredients (protein isolates, fibre powders, sugar alcohols) all push the classification toward “more processed.” A short list with recognisable ingredients sits lower.

The classification is informed by the NOVA framework — the research-standard four-group classification of processing depth — without replicating it exactly. Where NOVA is binary about group 4 (“ultra-processed”), Nime grades along a continuous spectrum, so two products in NOVA group 4 can differ meaningfully in how additive-dense or ingredient-engineered they are. The blog post on high-protein products works through one category as a case study.

3. Pesticides

For raw and minimally-processed produce, Nime estimates pesticide exposure risk by category using EU pesticide residue monitoring data and the EFSA pesticide review programme. Categories with consistently higher residue findings (some non-organic stone fruits, soft fruits, certain leafy greens) score worse; categories with consistently lower residue findings score better. For processed products, Nime surfaces specific pesticide-linked ingredient signals when present.

This is a probabilistic risk estimate at the category and ingredient level, not a per-product residue measurement. A tested-low batch could exist within a higher-risk category, and the opposite. Where the estimate is uncertain, Nime says so.

4. Microplastics

Microplastic exposure is estimated from packaging type, processing steps, and category-level research on contamination patterns. Categories with documented contamination — teabags releasing particles in hot water, bottled water in certain plastics, tinned and microwaveable foods, salt — score higher. Categories with consistently lower contamination findings score lower.

Nime does not detect microplastics in individual products directly. The score is a probable-exposure estimate based on peer-reviewed research and the packaging category, not a lab-measured contamination value. The blog post on the 2026 BPA and PFAS packaging ban covers why packaging chemistry is hard to verify per-product even when regulators are moving.

The Harmfulness score: a single 0–100 number

The Harmfulness score is Nime’s headline number: a single 0–100 measure, weighted from the four risk measures above, where lower is better. The weighting is intentional rather than equal — additives and ultra-processed level move the score more than pesticides or microplastics for most product categories, because the underlying research signals are stronger and the per-product variation within a category is larger.

Tapping the score in the app reveals the per-measure breakdown, so a single 70 doesn’t hide whether the risk is concentrated in one dimension or spread across all four. The bars and the composite are designed to be read together — neither replaces the other.

Better alternatives

When you scan a product, Nime surfaces a small number of lower-Harmfulness options in the same category. Same shelf, same role in your basket, lower score. The point isn’t a “don’t buy this” framing — you decide what to buy; Nime makes the lower-Harmfulness option in the category visible so the choice is easier.

Category match matters. A better alternative for a pizza is another pizza, not a salad. A better alternative for a children’s cereal is another cereal that scores lower, not “feed them porridge instead.” Pragmatic substitution beats moralising every time.

Plain-English explanations for every ingredient

Below the Harmfulness score and the per-measure breakdown, Nime shows the full ingredient list with a plain-English explanation next to every item. What it is. What it does in the product (preservative, emulsifier, sweetener, flavour). What current EFSA and research say about it. Whether it has had recent regulatory or research attention.

You don’t need to recognise E-numbers, distinguish an emulsifier from a stabiliser, or know that “modified starch” is a class with a dozen specific variants. If it’s on the label, Nime tells you what it is in language a curious adult can read without a glossary. The blog covers the underlying research patterns; the app applies them per ingredient, per product, per scan.

What we measure vs what we estimate

Transparency about this distinction is part of the methodology.

• Measured directly: presence of specific additives and E-numbers; declared nutrition values; allergen flags; declared ingredient origin where available; category and packaging type.
• Classified from research: per-additive risk band (a combination of regulatory position + independent research evidence); ultra-processed classification (NOVA-informed inference from the ingredient list and additive profile); the Harmfulness composite (weighted from the four measures).
• Estimated from category and packaging research: microplastic exposure risk; pesticide residue risk for processed products.

Where a value is estimated rather than measured directly, the app surfaces it as such. The methodology is not perfect; the alternative is silence, which is worse for the reader.

Personalisation

Default scoring reflects general-population guidance. Personalisation re-weights the score for individual context — family allergen profiles, low-FODMAP, low-sodium, ultra-processed avoidance, pregnancy, GLP-1 medication routines, custom trigger ingredients. A product that scores fine for the general population may be wrong for a specific user, and the opposite. Set the profile once and the score adapts on every scan.

How often we update

The product database is updated continuously. The risk classifications are revised when EFSA issues new scientific opinions, when relevant peer-reviewed papers land (notably the rolling NutriNet-Santé output and the EFSA 2026 mixture-effect guidance applicable to new applications from 20 July 2026), and when EU regulation changes. This page is republished when anything material in the scoring framework changes; the date at the top is the last meaningful revision.

What Nime is — and isn’t

Nime is a shopping companion built on Nime’s own scoring framework, drawing on current EU regulation, EFSA positions, and peer-reviewed research. It is not a medical-grade diagnostic, it does not replace dietetic advice from a qualified professional, and it does not detect contamination directly. It reads labels faster than you can and surfaces the things the front of the pack tends not to.

We are deliberately conservative about what we claim and explicit about what we estimate. If a future revision of EFSA guidance, new peer-reviewed evidence, or EU regulation makes a current classification look wrong, we update it. The substance of food science moves; the methodology has to move with it.

Questions, corrections, or methodology challenges

We welcome them. The methodology lives in public on this page because reader scrutiny improves it. If something here looks wrong or out of date, please contact us— corrections and challenges with sources attached are read and actioned.