SDOH AND HEALTH OUTCOMES

Skin Cancer and Chronic Skin Conditions Study

Two complementary projects examine skin health: an aggregated survey of skin cancer and chronic skin conditions, and an All of Us v8 analysis of medication-associated skin-cancer patterns using machine learning and molecular modeling.

Survey study Drug-associated skin-cancer study Survey data downloads

Interpretation: These projects use different populations and methods. Survey results are self-reported and aggregated; drug associations are observational and do not establish that a medication causes skin cancer.

STUDY OVERVIEW

Available survey domains

422respondents in the broad family cancer-history question

45respondents in skin-cancer follow-up questions

60respondents in chronic skin-condition follow-up questions

6race and ethnicity reporting groups available

ALL OF US v8 · MACHINE LEARNING · MOLECULAR MODELING

Drug-Associated Skin Cancer Patterns

This observational project evaluated commonly prescribed medications and medication combinations in relation to malignant skin-cancer outcomes. The workflow combined All of Us v8 cohort analysis, relative and absolute risk estimates, photosensitivity characteristics, random-forest and gradient-boosting models, SHAP interpretation, and molecular modeling.

22commonly prescribed drugs evaluated

10drugs allowed in a modeled combination

0.89reported machine-learning AUC

4proteins considered for docking

Drug-level participant counts and reported relative risk

Drug	Therapeutic category	Participants	Reported RR
Levothyroxine	Thyroid hormone supplement	37,984	7.27
Simvastatin	HMG-CoA reductase inhibitor	31,718	7.42
Rosuvastatin	HMG-CoA reductase inhibitor	24,017	7.78
Omeprazole	Proton pump inhibitor	80,381	8.95
Atorvastatin	HMG-CoA reductase inhibitor	77,939	9.13
Metoprolol	Beta-blocker	61,366	9.26
Losartan	Angiotensin II receptor blocker	36,823	9.33
Fluticasone	Inhaled corticosteroid	91,760	9.55
Prednisone	Systemic corticosteroid	93,191	9.66
Duloxetine	SNRI	27,212	9.79
Hydrochlorothiazide	Thiazide diuretic	56,363	9.87
Lisinopril	ACE inhibitor	63,841	10.02
Escitalopram	SSRI	26,569	10.05
Montelukast	Leukotriene receptor antagonist	24,146	10.16
Fluoxetine	SSRI	23,512	10.24
Pantoprazole	Proton pump inhibitor	67,858	10.25
Sertraline	SSRI	36,700	10.66
Amlodipine	Calcium-channel blocker	56,007	10.99
Amphetamine	CNS stimulant	8,614	11.82
Carvedilol	Alpha/beta-blocker	17,298	12.05
Metformin	Biguanide	46,846	12.16
Acetaminophen	Analgesic/antipyretic	216,972	12.16

Values are transcribed from the final poster. They must be interpreted using the poster's cohort definitions and analysis methods, not as causal medication effects.

Reported predictive findings

Reported model AUC: 0.89.
Photosensitivity rating, drug class, and vasodilator status were reported as strong predictors.
SHAP interpretation highlighted prednisone, omeprazole, atorvastatin, and hydrochlorothiazide.
Predicted-versus-observed plots were used to assess four drug-specific cancer-risk metrics.

Photosensitivity interpretation

The analysis compared overall skin-cancer association with a photosensitivity-linked component. A smaller gap was interpreted as greater possible photosensitivity contribution, while a larger gap suggested that non-photosensitive mechanisms might contribute.

This is a modeled interpretation and does not demonstrate a biological mechanism.

Drug combinations highlighted in the poster

Drug combination	RR	AR
Lisinopril + Omeprazole + Pantoprazole	8.43	0.37
Atorvastatin + Prednisone + Hydrochlorothiazide + Omeprazole	7.20	0.37
Acetaminophen + Fluticasone + Hydrochlorothiazide	8.42	0.37
Duloxetine + Carvedilol + Omeprazole	8.18	0.37
Acetaminophen + Rosuvastatin	7.36	0.36
Rosuvastatin + Amphetamine	9.30	0.34
Acetaminophen + Prednisone + Omeprazole	7.33	0.33
Acetaminophen + Duloxetine + Metoprolol + Amphetamine	7.38	0.32
Fluoxetine + Pantoprazole	9.27	0.29
Acetaminophen + Fluoxetine	9.43	0.28
Metformin + Montelukast	10.18	0.26
Fluoxetine + Rosuvastatin + Amlodipine + Levothyroxine	5.92	0.25
Escitalopram + Rosuvastatin + Levothyroxine	3.87	0.24
Escitalopram + Simvastatin + Losartan + Levothyroxine	3.86	0.24
Escitalopram + Sertraline + Amphetamine + Fluticasone	10.04	0.23

Comparative molecular docking

Target	Metric	Atorvastatin	Prednisone	Omeprazole	Hydrochlorothiazide
COX2	Docking score	−11.40	−5.8	−5.6	−3.4
COX2	Glide e-model	−98.623	2.678	−61.911	−34.53
AHR	Docking score	−4.3	−4.2	−3.9	−2.4
AHR	Glide e-model	−44.800	−29.674	−35.809	−27.310
TP53	Detected interaction	No interaction reported
MC1R	Detected interaction	No interaction reported

The poster also presents COX2–atorvastatin and AHR–atorvastatin structural views, RMSD profiles, and interaction-contact distributions from molecular-dynamics analysis.

ORIGINAL SCIENTIFIC POSTER

Review the complete study poster

The poster contains the complete workflow, figures, captions, reported conclusions, limitations, and future directions.

Open full-resolution PDF

Clinical caution: These findings should not be used to stop, start, or change any medication. Confounding, exposure duration, indication, ultraviolet exposure, lifestyle, geography, ancestry, healthcare utilization, and other factors require further evaluation.

SKIN CANCER

Skin-cancer survey results

Age when first diagnosed

Adult (18–64)34 · 75.6%

Older adult (65–74)11 · 24.4%

Currently receiving treatment

Yes10 · 22.2%

No35 · 77.8%

Still seeing a healthcare provider

Yes29 · 64.4%

No15 · 33.3%

Skipped1 · 2.2%

Family members reported

Self	45
Mother	29
Father	22
Sibling	22
Grandparent	14
Daughter	3

Select-all question; responses are not mutually exclusive.

CHRONIC SKIN CONDITIONS

Eczema and psoriasis survey results

Age when first diagnosed

Child (0–11)	12 · 20.0%
Adolescent (12–17)	9 · 15.0%
Adult (18–64)	35 · 58.3%
Older adult (65–74)	3 · 5.0%
Elderly (75+)	1 · 1.7%

Currently receiving treatment

Yes29 · 48.3%

No31 · 51.7%

Still seeing a healthcare provider

Yes32 · 53.3%

No28 · 46.7%

Family members reported

Self	60
Daughter	21
Sibling	16
Mother	14
Son	11
Father	9
Grandparent	5

Select-all question; responses are not mutually exclusive.

STUDY DATA

Download filtered data

Each file contains only skin-cancer, eczema, psoriasis, or other skin-condition records extracted from the larger aggregated survey analysis.

Source survey rows Question summary Overall frequencies Race-stratified percentages Race coverage Race disparity screening Yes/no association tests Categorical association tests Thematic question index

LIMITATIONS

Responsible interpretation

The information is aggregated and self-reported, not verified clinical-record data.
The skin-cancer follow-up group is small and racially imbalanced; subgroup comparisons are exploratory.
Family-history questions allow multiple selections, so their counts cannot be summed as unique participants.
The treatment questions indicate only yes or no; treatment type, response, duration, and outcome are unavailable.
Adjusted regression and causal conclusions require participant-level data and appropriate approvals.