model_name: hospital-readmission-predictor
license: mit
language: en
tags:
- tabular
- classification
- healthcare
- patient-risk
- synthetic-data
metrics:
- accuracy
- f1
- roc_auc
Hospital Readmission Predictor
Model Overview
Hospital Readmission Predictor is a machine learning model trained to predict whether a patient will be readmitted to the hospital within 30 days.
The dataset is synthetic, based on typical electronic health record features.
Intended for research, ML demos, and education.
Model Details
- Model Name: hospital-readmission-predictor
- Model Type: Binary Classification
- Framework: scikit-learn / XGBoost
- Input Format: CSV / Tabular
- Output: Readmission risk (Yes / No)
Training Data
Synthetic patient dataset with features:
| Feature | Type | Description |
|---|
| patient_id | String | Unique patient identifier |
| age | Integer | Patient age |
| gender | Categorical | Male / Female |
| diagnosis_code | Categorical | Primary diagnosis code |
| comorbidities | Integer | Number of comorbid conditions |
| length_of_stay | Integer | Hospital stay length in days |
| previous_admissions | Integer | Number of prior admissions |
| readmitted_30_days | Binary | Target label |
Intended Use
✅ Research and ML tutorials
✅ Healthcare model demos (synthetic)
❌ Actual clinical decision-making
Evaluation Results
| Metric | Score |
|---|
| Accuracy | 0.84 |
| F1-score | 0.81 |
| ROC-AUC | 0.87 |
Limitations
- Synthetic patient data
- Limited diversity and complexity
- Not representative of real hospital data
Ethical Considerations
No real patient data is used.
Misuse for clinical purposes could risk patient safety.
License
This model is released under the MIT License.
