Synthetic Dataset for Non-Invasive Female Hormone Monitoring Using Simulated Wearable Biosensor Signals

This dataset contains synthetically generated female hormone profiles and simulated wearable biosensor measurements developed for proof-of-concept evaluation of machine learning models for non-invasive hormone monitoring. The dataset was generated using literature-derived physiological hormone ranges and reported menstrual cycle dynamics and is intended exclusively for methodological validation. The hormone dataset includes Estradiol, Progesterone, Luteinizing Hormone (LH), Follicle Stimulating Hormone (FSH), and Cortisol concentrations generated for 300 simulated individuals across three consecutive 28-day menstrual cycles, resulting in 25,200 samples. A corresponding wearable biosensor dataset was generated by modeling biologically motivated relationships between hormone concentrations and six simulated biosensor features: Sweat impedance Sweat conductivity Sweat pH Electrochemical current Electrodermal activity (EDA) Skin temperature Gaussian noise was incorporated into the simulated biosensor signals to emulate measurement variability. Contents 1. synthetic_female_hormones.csv Synthetic hormone concentrations for 300 virtual patients across 3 menstrual cycles. 2. biosensor_simulated_data.csv Simulated wearable biosensor measurements derived from the hormone dataset. 3. generate_hormone_data.py Python script used to generate the synthetic hormone dataset. 4. generate_synthetic_sensor_data.py The dataset is intended solely for proof-of-concept algorithm development, benchmarking, educational research, and reproducible methodological evaluation. It does not contain clinical measurements and should not be used for clinical inference or validation.