Machine Learning Prediction Model for Histological Prostatitis in BPH Patients

A single-center retrospective study of 723 patients with benign prostatic hyperplasia (BPH) admitted to a tertiary medical center between June 2020 and June 2023. The dataset was used to develop and internally validate machine learning models for predicting the risk of histological prostatic inflammation. The best-performing artificial neural network model achieved an AUC of 0.852 in the validation set.

Use Cases

• Developing non-invasive risk assessment tools for histological prostatitis based on clinical variables like prostate volume and IPSS score.
• Comparing the performance of different machine learning algorithms (e.g., ANN, LASSO) for binary medical classification tasks.
• Applying SHAP analysis for model interpretability to identify key predictors such as neutrophil-to-lymphocyte ratio and acute urinary retention.
• Building online clinical decision support tools for pre-surgical risk stratification in BPH patients.

Strengths

• Dataset includes 723 patient records with a clear outcome label (53.5% had histological prostatitis).
• Model development followed a rigorous internal validation protocol with a 7:3 train-validation split.
• The best model's performance is quantified with specific metrics, including an AUC of 0.852 and a Brier score of 0.160.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown for the underlying data, which may limit suitability assessment.
• Data is from a single center, which may limit generalizability to other populations.

Provenance

Source

figshare, author Chunyang Meng.

Collection Method

Single-center retrospective study of patient records.

Time Range

June 2020 to June 2023.

Freshness

Last updated 2026-04-21 04:22:23; freshness should be verified.

Geography

Patients admitted to the urology department of a tertiary medical center (location unspecified).