Quantitative sensory testing dataset with capsaicin and menthol sensitization assessing modality-specific sex differences and sensitization effect sizes in 125 healthy volunteers
Description
This dataset contains experimental pain threshold measurements and classical pain model data collected from 125 healthy adult volunteers under both baseline and chemically-sensitized conditions. The data were generated using established quantitative sensory testing (QST) protocols, including mechanical (punctate and blunt pressure), thermal (heat and cold), and electrical pain modalities. Sensitization was induced using topical capsaicin (for heat and punctate pressure) and menthol (for cold) to assess pain plasticity and sensitization mechanisms. Each file provides pain threshold profiles, including raw measurements, transformed values for machine learning applications, and comprehensive participant metadata. File overview: File 1: exp_pain_data_orig.csv This file contains the original pain threshold data matrix with 125 rows (subjects) and 9 columns (variables). The first column ("ID") lists arbitrary subject identifiers. Columns 2 through 9 include pain threshold measurements from classical experimental pain models: vonFrey (punctate pressure, log gram von Frey), Heat (thermal, °C), Cold (thermal, °C), Pressure (blunt mechanical, N/m²), Current (electrical, mA), and their sensitized counterparts vonFrey_Capsaicin, Heat_Capsaicin, and Cold_Menthol. Values represent the stimulus intensity at which participants first reported pain, with some measurements reaching technical device limits. File 2: exp_pain_data_transformed.csv This file contains the same 125 subjects with 12 columns of transformed pain threshold data prepared for machine learning analyses. It includes eight transformed pain threshold variables (Heat, Pressure, Current, Heat_Capsaicin, Cold, Cold_Menthol, vonFrey, vonFrey_Capsaicin) and three calculated sensitization effect variables (Capsaicin_Effect_Heat, Menthol_Effect_Cold, Capsaicin_Effect_vonFrey). Transformations include signed logarithmic transformation (sign(x) × log(|x| + 1)) applied to all measurements, conditional linear multiple imputation for censored data (left-censored cold thresholds at 0°C, right-censored von Frey thresholds at log(301) mN), and z-score standardization. Sensitization effects were computed as differences between baseline and sensitized log-transformed thresholds. File 3: exp_pain_metadata.csv This metadata file includes 125 subjects with six columns. The first column lists subject IDs consistent with files 1 and 2. The second column provides age in years. Column 3 indicates smoking status ("Smoker" or "Nonsmoker"). Column 4 provides biological sex ("Female" or "Male"). Column 5 indicates the experimenter who conducted the measurements ("F" for female tester, "M" for male tester). Column 6 ("TrainingValidation") indicates assignment to training or validation sets, determined through optimal distribution-preserving downsampling (80% training, 20% validation). All data are anonymized and formatted for reproducibility and secondary analysis.
Files
Steps to reproduce
The data were gathered using established and standardized protocols to ensure reproducibility and comparability. Pain threshold measurements were conducted by trained investigators using validated devices and standardized procedures. Punctate pressure thresholds were assessed using von Frey hairs with logistic regression analysis of yes/no pain responses. Heat and cold pain thresholds were determined as medians of repeated measurements (5-8 trials at 25-35 second intervals). Blunt pressure thresholds were measured at the mid-phalanx of the right middle finger. Electrical pain thresholds were determined using 5-Hz sine wave current at the left mid-phalanx. Sensitization procedures involved topical application of capsaicin cream (for heat and punctate pressure) or menthol cream (for cold), with pain thresholds reassessed in the sensitized areas. All data transformations, including signed logarithmic transformation, imputation of censored values, calculation of sensitization effects, and z-score standardization, were performed using documented Python and R scripts. All experimental details and workflows are documented in the publications listed below.
Institutions
- Goethe University FrankfurtHesse, Frankfurt am Main