Datasetpublic

SpotBite/skin-score-322-foods

SpotBite Skin Score Dataset — 322 foods scored for acne-related diet factors A small, open dataset of 322 common foods, each scored 0–100 for how likely one typical serving is to contribute to acne, based on four diet factors that published trials link to breakouts: glycemic load, dairy, inflammatory fats, and anti-inflammatory boosters (omega-3, polyphenols). The scores are the same ones shown in the SpotBite app and on its public food pages (spotbite.app/foods/<slug>). Every… See the full description on the dataset page: https://huggingface.co/datasets/SpotBite/skin-score-322-foods.

sourceHugging Facecc-by-4.0updated 11d agoView on Hugging Face

0likes47downloads

Dataset Card

SpotBite Skin Score Dataset — 322 foods scored for acne-related diet factors

A small, open dataset of 322 common foods, each scored 0–100 for how likely one typical serving is to contribute to acne, based on four diet factors that published trials link to breakouts: glycemic load, dairy, inflammatory fats, and anti-inflammatory boosters (omega-3, polyphenols).

The scores are the same ones shown in the SpotBite app and on its public food pages (spotbite.app/foods/<slug>). Every row is deterministic: the same inputs always give the same score, and the scoring code is described below so anyone can reproduce or audit it.

License: CC BY 4.0 — free to use, share, and adapt with attribution to SpotBite (spotbite.app).

Files

File	What it is
`spotbite-skin-scores.csv`	322 rows × 22 columns, one food per row
`spotbite-skin-scores.json`	Same data with metadata (version, license, citation)
`category-stats.json`	Per-category mean score and share of GLOW / AVOID foods

Quick facts (version 2026-09-16)

—322 foods across 12 categories, one typical serving each
—Score distribution: GLOW (≥90) 164 · GOOD (70–89) 91 · WARN (40–69) 64 · AVOID (<40) 3
—Lowest scores: milkshake 32, cheesecake 34, milk tea 35, frozen yogurt 40, scone 43
—Highest-scoring categories: fruit & vegetables (mean 97), nuts & seeds (95). Lowest: bakery & snacks (60), dairy (69), carbs & grains (71)
—Dairy-plus-sugar combinations (milkshake, milk tea, ice cream, cheesecake) dominate the bottom of the table because they are penalised on two axes at once

Columns

Column	Type	Meaning
`slug`	string	Stable id, matches `spotbite.app/foods/<slug>`
`name`	string	Display name
`aliases`	string	Alternative names, `\	`-separated (may be empty)
`category`	string	One of 12 categories (Dairy, Drinks, Protein, …)
`serving_g`	number	Typical single serving in grams. All values below are for this serving
`skin_score`	0–100	Final score, higher = more skin-friendly
`level`	enum	`GLOW` ≥90 · `GOOD` ≥70 · `WARN` ≥40 · `AVOID` <40
`glycemic_load`	number/null	GI × carbs ÷ 100. Null when the food has no meaningful carbs
`penalty_glycemic`	0–50	Points removed for glycemic load
`penalty_dairy`	0–25	Points removed for dairy
`penalty_saturated_fat`	0–12	Points removed for saturated fat
`penalty_omega6`	0–13	Points removed for omega-6 fat
`boost_omega3`	0–15	Points added back for omega-3
`boost_polyphenol`	0/5/10	Points added back for polyphenols (low/medium/high)
`carbs_g`	g	Carbohydrates per serving
`glycemic_index`	number/null	Glycemic index used (null if no carbs)
`dairy_g`	g	Grams of the serving that are dairy
`saturated_fat_g`	g	Saturated fat per serving
`omega6_g`	g	Omega-6 fat per serving
`omega3_g`	g	Omega-3 fat per serving
`polyphenol_level`	enum	`low` / `medium` / `high`
`url`	string	Public page with the full explanation and sources

How the score is computed

skin_score = clamp( 100 − penalties + boosts , 0, 100 )

penalties = glycemic (max 50) + dairy (max 25) + saturated fat (max 12) + omega-6 (max 13)
boosts    = omega-3 (max 15) + polyphenols (0 / 5 / 10)

Axis	Input	Rule
Glycemic (50)	glycemic load GL = GI × carbs_g / 100	GL 0–10 → 0–18 pts, 10–19 → 18–32 pts, 20–50 → 32–50 pts (linear within each band)
Dairy (25)	grams of dairy	25 × dairy_g / 30, capped at 25 (≈30 g of dairy = full penalty)
Saturated fat (12)	grams	12 × sat_g / 10, capped
Omega-6 (13)	grams	13 × o6_g / 5, capped
Omega-3 (+15)	grams	15 × o3_g / 2, capped
Polyphenols (+10)	low / medium / high	0 / 5 / 10

The weights reflect the relative strength of evidence: glycemic load has the most and strongest randomized-trial support, dairy the largest observational evidence base, and fat quality the least direct evidence, so it carries the fewest points. The score describes the food on its own; in a real meal, protein and fat eaten alongside carbs lower the glycemic response, which this per-food score does not model.

Where the numbers come from

—Scoring rules — SpotBite's Skin Score v2 (2026-04-27), the same code that runs in the app. Rules are documented above in full.
—Serving sizes and nutrition values (carbs, GI, dairy, fats, polyphenol level) — estimated per food by a large language model (Gemini via Perplexity Agent API) using a fixed prompt, then frozen on 2026-09-02 and spot-checked by hand. They are estimates for a typical serving, not laboratory measurements, and are the main source of uncertainty in this dataset. Values for branded or regional variants of a food will differ.
—Evidence for the axes — the trials and reviews below.

Evidence the scoring is based on

Glycemic load

—Smith et al. 2007, Am J Clin Nutr — 12-week RCT (n=43): low-glycemic-load diet reduced acne lesion count by 23.5 vs 12 in controls. PubMed 17616769
—Burris et al. 2018, J Acad Nutr Diet — 2-week low-GI diet (n=66) lowered IGF-1 by 9%. PubMed 29691143

Dairy

—Hoppe et al. 2004, Eur J Clin Nutr — 1.5 L skim milk/day (n=24) raised IGF-1 by 19% within 7 days. PubMed 15054433
—Juhl et al. 2018, Nutrients — meta-analysis of 14 studies (78,529 people): any dairy intake raised acne odds by 25%. PubMed 30096883
—Adebamowo et al. 2006, Dermatol Online J — milk intake associated with acne in teenage girls. PubMed 17083856

Inflammatory fats

—Caperton et al. 2014, J Clin Aesthet Dermatol — double-blind RCT (n=14): a single dose of 100% cocoa increased lesions, peaking at day 4–7. PMC4025515
—Vongraviopap & Asawanonda 2016, Int J Dermatol — 25 g dark chocolate daily worsened acne within 2 weeks. PubMed 26711092

Omega-3 and polyphenols

—Khayef et al. 2012, Lipids Health Dis — 12 weeks of fish oil cut inflammatory lesions from 10.1 to 5.8 on average. PMC3543297
—Saric et al. 2016, Antioxidants — review of tea polyphenol trials: green tea extract reduced sebum and inflammatory lesions. PMC5384166

Reviews

—Meixiong et al. 2022, JAAD Int — Diet and acne: a systematic review. PubMed 35373155
—Dall'Oglio et al. 2021, Int J Dermatol — Diet and acne: review of the evidence from 2009 to 2020. PubMed 33462816

Full source list with plain-language summaries: spotbite.app/science

Limitations

—Per-food, single-serving scores. Meal context (mixing protein and fat with carbs) is not modelled.
—Nutrition inputs are LLM estimates for a "typical" serving, not measured values.
—The evidence base is small: most trials have n < 70 and report averages. Individual responses to diet vary a lot.
—This is lifestyle information, not medical advice. Acne has many causes besides diet.

Citation

SpotBite (2026). SpotBite Skin Score dataset: 322 foods scored for acne-related diet factors (version 2026-09-16). https://spotbite.app/foods/

BibTeX:

bibtex

@misc{spotbite2026skinscore,
  title  = {SpotBite Skin Score dataset: 322 foods scored for acne-related diet factors},
  author = {{SpotBite}},
  year   = {2026},
  url    = {https://spotbite.app/foods/},
  note   = {Version 2026-09-16, CC BY 4.0}
}

About

SpotBite is a mobile app that scores a photo of a meal for its likely effect on acne-prone skin. spotbite.app · App Store / Google Play · questions: support@spotbite.app