Baseline Characteristics for Early Pregnancy Outcome Prediction, 421 Patients

A dataset of 421 patient records from Lanzhou University Second Hospital between March 2023 and August 2024. It contains baseline characteristics and dynamic changes of β-hCG, progesterone, and estradiol for predicting early pregnancy loss. The data was used to build and compare six machine learning models, with results published by author Luming Chen.

Use Cases

• Training a Random Forest model for early pregnancy loss prediction based on hormone level dynamics.
• Comparing logistic regression and neural network performance on clinical outcome classification based on β-hCG, progesterone, and estradiol data.
• Conducting feature selection analysis using LASSO and RF-RFE on serial hormone monitoring data.
• Developing a clinical risk stratification tool based on the hormone features described.

Strengths

• Data is from a specific clinical study of 421 patients, providing a defined cohort.
• Model performance metrics (AUC, accuracy, precision) for six algorithms are reported, offering validation context.
• The dataset is focused on three specific hormones (β-hCG, P, E2) with both baseline and dynamic changes.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is from a single hospital in China, which may limit geographic generalizability.

Provenance

Source

Lanzhou University Second Hospital

Collection Method

Retrospective study screening patient records.

Time Range

March 2023 to August 2024

Freshness

Last updated 2026-04-27 17:25:46; freshness should be verified.

Geography

China (likely Lanzhou)