Welden tells you how much microplastic, plasticizers, forever-chemicals, and pesticides you're exposed to each year and where those exposures come from. Then, you get a plan to reduce your exposure with one-and-done product swaps in the home and easy behavior changes. Before you trust those numbers, you should know exactly where they come from, who tested what, and where we still have gaps.
One thing to keep in mind: this service is not medical advice. You should discuss any medically-related questions with your healthcare provider.
If you only have a minute, this is the whole thing. If you have more questions about the details, continue reading to the end.
Want the detailed version? Keep reading.
That's (roughly) the question most of our early users had first. We don't know that on our own. We know it because peer-reviewed researchers measured it, published it, and we use their metrics with the citation attached.
Every input in the Welden model, every "this product releases this much," every "your body absorbs this fraction," comes from a published peer-reviewed paper. We compiled over 100 studies spanning microplastic counts, phthalate migration, PFAS in cookware, indoor air sampling, dietary intake of plasticizers, and biomonitoring. Here is a short list of the studies that produced some of the most-visible numbers on your dashboard, each linked to the original paper:
Those are seven examples. If a number appears in your score, a paper appears in the citations.
Your Welden Score is a 0 → 10 composite. What it does is take your daily routines and translate them into a population-relative ranking across eight exposure classes.
Each dimension is placed on an absolute scale anchored at 0 and 10. We set two anchors per contaminant: a floor (score 0) — the cleanest intake a careful person actually reaches in the model — and a reference point (score 10) — a named high-exposure scenario that sits around the 80th percentile of US exposure. The scale soft-caps rather than clipping, so a genuinely high exposure can score above 10. Your intake sits log-linearly between them, so each point up is a fixed multiplicative step in exposure (like decibels or pH). Where a human biomarker exists, we locate that reference point using log-normal distributions fit to CDC NHANES 2017–2018 biomonitoring (serum PFAS, urinary phthalates and BPA, blood and urinary metals) and peer-reviewed exposure modeling. The eight sub-scores are then weighted and averaged. Because typical US exposure is itself elevated, the default American profile lands near 6, not 5 — typical is not the same as safe.
We use a lognormal distribution rather than a normal distribution because most exposures are right-skewed: the people with the highest exposures have disproportionately higher exposures than a bell curve would predict. The lognormal shape captures that long upper tail. It's an assumption on our part, but we feel it's the appropriate one for this class of data.
For each dimension we set two anchors. The floor (score 0) is the cleanest-achievable modeled intake — the lowest a careful person realistically reaches. The reference point (score 10) is a named high-exposure scenario for that contaminant, which lands near the 80th percentile of US exposure. Heavy metals is the exception: its anchors come directly from CDC NHANES biomonitoring rather than a named scenario. For the two dimensions without a clean human biomarker — microplastics and VOCs — we use the modeled population spread instead. Your reported intake is then placed log-linearly between floor and ceiling.
Your intake is placed log-linearly between the floor and the reference point to get the dimension score. The final score is the weighted average:
dimension_score = 10 × [ ln(intake) − ln(floor) ] / [ ln(reference) − ln(floor) ], floored at 0, soft-capped above 10
Welden Score = Σ (dimension_score × weight) / Σ (weights)
Each dimension is scored on three criteria — evidence of harm, severity, and biological persistence — each on a 1–4 scale. The sum is the weight. PFAS gets 12 (the maximum) — the only contaminant that maxes all three: IARC Group 1, EPA drinking-water limits in the parts-per-trillion, and a half-life in years that drives bioaccumulation. Microplastics (the plastic particles themselves, in isolation) get 5 because the human-health evidence is still emerging and the WHO currently rates drinking-water levels as low concern. Heavy metals (9) and UV (7) were added with the eight-dimension model and carry composite weights; their per-criterion breakdown is not yet published, which is why those three columns are left blank below rather than filled in after the fact.
Your exposure comes from four distinct source categories: drinking, eating, breathing, and personal care. Each pathway has its own intake equation, its own modifiers, and its own data sources. They are calculated independently and added together at the end — drinking water doesn't share math with indoor air, which doesn't share math with personal care. The cards below describe how each pathway is treated.
Drinking is the pathway with the strongest data backbone. For each beverage source you report — tap water, bottled water, canned drinks, to-go cups, coffee — the model multiplies your reported weekly volume by a per-liter contaminant load specific to that source. Tap water is the most personalized leg: we use EPA UCMR5 monitoring data keyed to your zip code so the PFAS value reflects your actual utility where data exists. When your zip is below the lowest reportable level or absent from UCMR5, the model falls back to the state median, then the national mean, in that order.
Eating is the most varied pathway because food contact happens through dozens of distinct micro-paths. We split it into three groups, each with its own intake equation. Cookware captures heat and abrasion-driven release of PFAS from nonstick coatings, scaling with your weekly cooking sessions and the condition of the pan. Food storage and preparation captures phthalate, bisphenol, and plastic migration, which is sensitive to your food heating habits. Your actual food has chemicals that travel through the food chain to you. Dairy fat concentrates phthalates, grains can carry glyphosate, and conventional produce carries USDA-PDP-measured pesticide residues weighted by your reported organic-shopping share.
Indoor air microplastic concentration is modulated by your reported HEPA-purifier use, the synthetic-clothing fraction of your wardrobe, carpet coverage, and proximity to busy roads. Indoor VOCs are broken into discrete sub-sources: mattress and furniture off-gassing, gas-stove combustion benzene and NO₂, and attached-garage gasoline vapors if you have those. Outdoor ambient VOCs use EPA AirToxScreen data resolved to the census tract, so your zip code determines real local values rather than a national average.
Personal care has the trickiest data problem of the four pathways: most of the consequential exposure comes from phthalates that legally do not have to be disclosed, because they hide inside the "fragrance" umbrella ingredient. Our model handles this by applying a fragrance premium — a higher per-use phthalate load — to any product you report as scented. Dermal absorption is then weighted by surface area (a body lotion covers more skin than a hand lotion) and by leave-on time (a deodorant stays on; a body wash rinses off). Thermal-receipt BPA, cosmetic PFAS in lipstick and foundation, and skin-contact phthalates from synthetic clothing are added as separate sub-equations.
Each pathway's intake feeds the relevant dimension; the dimension scores are placed on the 0 to 10 scale, then weighted and averaged. When new research changes a single pathway's intake equation, only that part of your score moves.
To walk you through how we'd calculate a typical user's PFAS score, here's a single survey answer traced step by step. The number that comes out is computed from your reported behavior … not from a national average that we slap your zip code onto.
5,600 ng/year from tap. A user who reported 4 glasses a day would have half this number. A user who reported 12 would have 50% more. Same zip, same filter, different intake.−0.1. Two people in the same zip code who both buy the same RO filter will see different deltas because their cookware, fast-food, and popcorn baselines are different. Nothing about your projected change is a stock number.Every product page in your reduction plan ran a calculation like this against your survey data. There is no static "this filter saves you 0.2 points" number. The number is built for you, from your answers, using the formula above.
Once we know your exposure pathways, the question is which products actually reduce them. Most "non-toxic" lists you've seen conflate three different things: a brand that says "BPA-free," a brand that names a third-party lab, and a brand that says nothing at all. We don't. Every product is evaluated against ten chemicals, and one of five states is assigned to each.
Brand states absence AND cites a named third-party authority — NSF, AHAM, OEKO-TEX, MADE SAFE, Consumer Reports lab, Mamavation acid-bath, or a peer-reviewed paper.
Manufacturer claims absence on its own page, no third-party test cited. Accepted conditionally; flagged in the row.
The chemical class is listed in the product's own materials. Honest disclosure, but the user should know.
The brand doesn't address this chemical anywhere on its product page or technical documentation. Silence is not absence. We mark it accordingly.
The chemical isn't plausibly relevant — flame retardants on glass, phthalates in cast iron.
The state we use most is NOT_DISCLOSED. Most consumer brands simply don't test for the majority of chemicals on this list.
Four gaps in the model to be aware of. We name them up front so you can decide for yourself whether the score on your dashboard is something you can act on.
Welden predicts where your exposure ranks in the US population based on your reported behaviors. It's calibrated at the population level against NHANES biomonitoring. It is not yet validated against your individual urinary or serum biomarkers.
Most microplastic studies measure particles down to about 5–20 microns. Anything smaller — including most nanoplastics — passes through the instrument's filter and doesn't show up in the count. Pletz 2022's metareview applied power-law extrapolation and estimated that the particles below the measurement floor outnumber the visible ones, sometimes by orders of magnitude. Your reported MP intake reflects what science can currently count. The real number is almost certainly higher.
Drinking water is the strongest pathway. EPA UCMR5 gives us zip-code-level PFAS data, NSF certifies the filters, and the peer-reviewed literature is mature. Personal care is the weakest, because the "fragrance" umbrella loophole legally protects the most consequential ingredient class from disclosure. We say so on every personal-care product card rather than pretending the data is symmetric.
The bisphenol family includes BPA, BPS, BPF, and a dozen analogs with similar endocrine-disrupting activity. Most "BPA-free" claims address only BPA. When a brand is silent on BPS and BPF, we mark the bisphenols column FREE_CLAIM rather than FREE_VERIFIED. You'll see this on a lot of glass-with-plastic-lid food storage. It's an industry gap, not a defect in our scoring.
The assessment is free, takes about three minutes, and doesn't require an account. The full citation library is also free.