PyHealth: Healthcare AI Toolkit
Overview
PyHealth is a comprehensive Python library for healthcare AI that provides specialized tools, models, and datasets for clinical machine learning. Use this skill when developing healthcare prediction models, processing clinical data, working with medical coding systems, or deploying AI solutions in healthcare settings.
When to Use This Skill
Invoke this skill when:
- Working with healthcare datasets: MIMIC-III, MIMIC-IV, eICU, OMOP, sleep EEG data, medical images
- Clinical prediction tasks: Mortality prediction, hospital readmission, length of stay, drug recommendation
- Medical coding: Translating between ICD-9/10, NDC, RxNorm, ATC coding systems
- Processing clinical data: Sequential events, physiological signals, clinical text, medical images
- Implementing healthcare models: RETAIN, SafeDrug, GAMENet, StageNet, Transformer for EHR
- Evaluating clinical models: Fairness metrics, calibration, interpretability, uncertainty quantification
Core Capabilities
PyHealth operates through a modular 5-stage pipeline optimized for healthcare AI:
- Data Loading: Access 10+ healthcare datasets with standardized interfaces
- Task Definition: Apply 20+ predefined clinical prediction tasks or create custom tasks
- Model Selection: Choose from 33+ models (baselines, deep learning, healthcare-specific)
- Training: Train with automatic checkpointing, monitoring, and evaluation
- Deployment: Calibrate, interpret, and validate for clinical use
Performance: 3x faster than pandas for healthcare data processing
Quick Start Workflow
from pyhealth.datasets import MIMIC4Dataset
from pyhealth.tasks import mortality_prediction_mimic4_fn
from pyhealth.datasets import split_by_patient, get_dataloader
from pyhealth.models import Transformer
from pyhealth.trainer import Trainer
# 1. Load dataset and set task
dataset = MIMIC4Dataset(root="/path/to/data")
sample_dataset = dataset.set_task(mortality_prediction_mimic4_fn)
# 2. Split data
train, val, test = split_by_patient(sample_dataset, [0.7, 0.1, 0.2])
# 3. Create data loaders
train_loader = get_dataloader(train, batch_size=64, shuffle=True)
val_loader = get_dataloader(val, batch_size=64, shuffle=False)
test_loader = get_dataloader(test, batch_size=64, shuffle=False)
# 4. Initialize and train model
model = Transformer(
dataset=sample_dataset,
feature_keys=["diagnoses", "medications"],
mode="binary",
embedding_dim=128
)
trainer = Trainer(model=model, device="cuda")
trainer.train(
train_dataloader=train_loader,
val_dataloader=val_loader,
epochs=50,
monitor="pr_auc_score"
)
# 5. Evaluate
results = trainer.evaluate(test_loader)
Detailed Documentation
This skill includes comprehensive reference documentation organized by functionality. Read specific reference files as needed:
1. Datasets and Data Structures
File: references/datasets.md
Read when:
- Loading healthcare datasets (MIMIC, eICU, OMOP, sleep EEG, etc.)
- Understanding Event, Patient, Visit data structures
- Processing different data types (EHR, signals, images, text)
- Splitting data for training/validation/testing
- Working with SampleDataset for task-specific formatting
Key Topics:
- Core data structures (Event, Patient, Visit)
- 10+ available datasets (EHR, physiological signals, imaging, text)
- Data loading and iteration
- Train/val/test splitting strategies
- Performance optimization for large datasets
2. Medical Coding Translation
File: references/medical_coding.md
Read when:
- Translating between medical coding systems
- Working with diagnosis codes (ICD-9-CM, ICD-10-CM, CCS)
- Processing medication codes (NDC, RxNorm, ATC)
- Standardizing procedure codes (ICD-9-PROC, ICD-10-PROC)
- Grouping codes into clinical categories
- Handling hierarchical drug classifications
Key Topics:
- InnerMap for within-system lookups
- CrossMap for cross-system translation
- Supported coding systems (ICD, NDC, ATC, CCS, RxNorm)
- Code standardization and hierarchy traversal
- Medication classification by therapeutic class
- Integration with datasets
3. Clinical Prediction Tasks
File: references/tasks.md
Read when:
- Defining clinical prediction objectives
- Using predefined tasks (mortality, readmission, drug recommendation)
- Working with EHR, signal, imaging, or text-based tasks
- Creating custom prediction tasks
- Setting up input/output schemas for models
- Applying task-specific filtering logic
Key Topics:
- 20+ predefined clinical tasks
- EHR tasks (mortality, readmission, length of stay, drug recommendation)
- Signal tasks (sleep staging, EEG analysis, seizure detection)
- Imaging tasks (COVID-19 chest X-ray classification)
- Text tasks (medical coding, specialty classification)
- Custom task creation patterns
4. Models and Architectures
File: references/models.md
Read when:
- Selecting models for clinical prediction
- Understanding model architectures and capabilities
- Choosing between general-purpose and healthcare-specific models
- Implementing interpretable models (RETAIN, AdaCare)
- Working with medication recommendation (SafeDrug, GAMENet)
- Using graph neural networks for healthcare
- Configuring model hyperparameters
Key Topics:
- 33+ available models
- General-purpose: Logistic Regression, MLP, CNN, RNN, Transformer, GNN
- Healthcare-specific: RETAIN, SafeDrug, GAMENet, StageNet, AdaCare
- Model selection by task type and data type
- Interpretability considerations
- Computational requirements
- Hyperparameter tuning guidelines
5. Data Preprocessing
File: references/preprocessing.md
Read when:
- Preprocessing clinical data for models
- Handling sequential events and time-series data
- Processing physiological signals (EEG, ECG)
- Normalizing lab values and vital signs
- Preparing labels for different task types
- Building feature vocabularies
- Managing missing data and outliers
Key Topics:
- 15+ processor types
- Sequence processing (padding, truncation)
- Signal processing (filtering, segmentation)
- Feature extraction and encoding
- Label processors (binary, multi-class, multi-label, regression)
- Text and image preprocessing
- Common preprocessing workflows
6. Training and Evaluation
File: references/training_evaluation.md
Read when:
- Training models with the Trainer class
- Evaluating model performance
- Computing clinical metrics
- Assessing model fairness across demographics
- Calibrating predictions for reliability
- Quantifying prediction uncertainty
- Interpreting model predictions
- Preparing models for clinical deployment
Key Topics:
- Trainer class (train, evaluate, inference)
- Metrics for binary, multi-class, multi-label, regression tasks
- Fairness metrics for bias assessment
- Calibration methods (Platt scaling, temperature scaling)
- Uncertainty quantification (conformal prediction, MC dropout)
- Interpretability tools (attention visualization, SHAP, ChEFER)
- Complete training pipeline example
Installation
uv pip install pyhealth
Requirements:
- Python ≥ 3.7
- PyTorch ≥ 1.8
- NumPy, pandas, scikit-learn
Common Use Cases
Use Case 1: ICU Mortality Prediction
Objective: Predict patient mortality in intensive care unit
Approach:
- Load MIMIC-IV dataset → Read
references/datasets.md
- Apply mortality prediction task → Read
references/tasks.md
- Select interpretable model (RETAIN) → Read
references/models.md
- Train and evaluate → Read
references/training_evaluation.md
- Interpret predictions for clinical use → Read
references/training_evaluation.md
Use Case 2: Safe Medication Recommendation
Objective: Recommend medications while avoiding drug-drug interactions
Approach:
- Load EHR dataset (MIMIC-IV or OMOP) → Read
references/datasets.md
- Apply drug recommendation task → Read
references/tasks.md
- Use SafeDrug model with DDI constraints → Read
references/models.md
- Preprocess medication codes → Read
references/medical_coding.md
- Evaluate with multi-label metrics → Read
references/training_evaluation.md
Use Case 3: Hospital Readmission Prediction
Objective: Identify patients at risk of 30-day readmission
Approach:
- Load multi-site EHR data (eICU or OMOP) → Read
references/datasets.md
- Apply readmission prediction task → Read
references/tasks.md
- Handle class imbalance in preprocessing → Read
references/preprocessing.md
- Train Transformer model → Read
references/models.md
- Calibrate predictions and assess fairness → Read
references/training_evaluation.md
Use Case 4: Sleep Disorder Diagnosis
Objective: Classify sleep stages from EEG signals
Approach:
- Load sleep EEG dataset (SleepEDF, SHHS) → Read
references/datasets.md
- Apply sleep staging task → Read
references/tasks.md
- Preprocess EEG signals (filtering, segmentation) → Read
references/preprocessing.md
- Train CNN or RNN model → Read
references/models.md
- Evaluate per-stage performance → Read
references/training_evaluation.md
Use Case 5: Medical Code Translation
Objective: Standardize diagnoses across different coding systems
Approach:
- Read
references/medical_coding.md for comprehensive guidance
- Use CrossMap to translate between ICD-9, ICD-10, CCS
- Group codes into clinically meaningful categories
- Integrate with dataset processing
Use Case 6: Clinical Text to ICD Coding
Objective: Automatically assign ICD codes from clinical notes
Approach:
- Load MIMIC-III with clinical text → Read
references/datasets.md
- Apply ICD coding task → Read
references/tasks.md
- Preprocess clinical text → Read
references/preprocessing.md
- Use TransformersModel (ClinicalBERT) → Read
references/models.md
- Evaluate with multi-label metrics → Read
references/training_evaluation.md
Best Practices
Data Handling
Always split by patient: Prevent data leakage by ensuring no patient appears in multiple splits
from pyhealth.datasets import split_by_patient
train, val, test = split_by_patient(dataset, [0.7, 0.1, 0.2])
Check dataset statistics: Understand your data before modeling
print(dataset.stats()) # Patients, visits, events, code distributions
Use appropriate preprocessing: Match processors to data types (see references/preprocessing.md)
Model Development
Start with baselines: Establish baseline performance with simple models
- Logistic Regression for binary/multi-class tasks
- MLP for initial deep learning baseline
Choose task-appropriate models:
- Interpretability needed → RETAIN, AdaCare
- Drug recommendation → SafeDrug, GAMENet
- Long sequences → Transformer
- Graph relationships → GNN
Monitor validation metrics: Use appropriate metrics for task and handle class imbalance
- Binary classification: AUROC, AUPRC (especially for rare events)
- Multi-class: macro-F1 (for imbalanced), weighted-F1
- Multi-label: Jaccard, example-F1
- Regression: MAE, RMSE
Clinical Deployment
Calibrate predictions: Ensure probabilities are reliable (see references/training_evaluation.md)
Assess fairness: Evaluate across demographic groups to detect bias
Quantify uncertainty: Provide confidence estimates for predictions
Interpret predictions: Use attention weights, SHAP, or ChEFER for clinical trust
Validate thoroughly: Use held-out test sets from different time periods or sites
Limitations and Considerations
Data Requirements
- Large datasets: Deep learning models require sufficient data (thousands of patients)
- Data quality: Missing data and coding errors impact performance
- Temporal consistency: Ensure train/test split respects temporal ordering when needed
Clinical Validation
- External validation: Test on data from different hospitals/systems
- Prospective evaluation: Validate in real clinical settings before deployment
- Clinical review: Have clinicians review predictions and interpretations
- Ethical considerations: Address privacy (HIPAA/GDPR), fairness, and safety
Computational Resources
- GPU recommended: For training deep learning models efficiently
- Memory requirements: Large datasets may require 16GB+ RAM
- Storage: Healthcare datasets can be 10s-100s of GB
Troubleshooting
Common Issues
ImportError for dataset:
- Ensure dataset files are downloaded and path is correct
- Check PyHealth version compatibility
Out of memory:
- Reduce batch size
- Reduce sequence length (
max_seq_length)
- Use gradient accumulation
- Process data in chunks
Poor performance:
- Check class imbalance and use appropriate metrics (AUPRC vs AUROC)
- Verify preprocessing (normalization, missing data handling)
- Increase model capacity or training epochs
- Check for data leakage in train/test split
Slow training:
- Use GPU (
device="cuda")
- Increase batch size (if memory allows)
- Reduce sequence length
- Use more efficient model (CNN vs Transformer)
Getting Help
Example: Complete Workflow
# Complete mortality prediction pipeline
from pyhealth.datasets import MIMIC4Dataset
from pyhealth.tasks import mortality_prediction_mimic4_fn
from pyhealth.datasets import split_by_patient, get_dataloader
from pyhealth.models import RETAIN
from pyhealth.trainer import Trainer
# 1. Load dataset
print("Loading MIMIC-IV dataset...")
dataset = MIMIC4Dataset(root="/data/mimic4")
print(dataset.stats())
# 2. Define task
print("Setting mortality prediction task...")
sample_dataset = dataset.set_task(mortality_prediction_mimic4_fn)
print(f"Generated {len(sample_dataset)} samples")
# 3. Split data (by patient to prevent leakage)
print("Splitting data...")
train_ds, val_ds, test_ds = split_by_patient(
sample_dataset, ratios=[0.7, 0.1, 0.2], seed=42
)
# 4. Create data loaders
train_loader = get_dataloader(train_ds, batch_size=64, shuffle=True)
val_loader = get_dataloader(val_ds, batch_size=64)
test_loader = get_dataloader(test_ds, batch_size=64)
# 5. Initialize interpretable model
print("Initializing RETAIN model...")
model = RETAIN(
dataset=sample_dataset,
feature_keys=["diagnoses", "procedures", "medications"],
mode="binary",
embedding_dim=128,
hidden_dim=128
)
# 6. Train model
print("Training model...")
trainer = Trainer(model=model, device="cuda")
trainer.train(
train_dataloader=train_loader,
val_dataloader=val_loader,
epochs=50,
optimizer="Adam",
learning_rate=1e-3,
weight_decay=1e-5,
monitor="pr_auc_score", # Use AUPRC for imbalanced data
monitor_criterion="max",
save_path="./checkpoints/mortality_retain"
)
# 7. Evaluate on test set
print("Evaluating on test set...")
test_results = trainer.evaluate(
test_loader,
metrics=["accuracy", "precision", "recall", "f1_score",
"roc_auc_score", "pr_auc_score"]
)
print("\nTest Results:")
for metric, value in test_results.items():
print(f" {metric}: {value:.4f}")
# 8. Get predictions with attention for interpretation
predictions = trainer.inference(
test_loader,
additional_outputs=["visit_attention", "feature_attention"],
return_patient_ids=True
)
# 9. Analyze a high-risk patient
high_risk_idx = predictions["y_pred"].argmax()
patient_id = predictions["patient_ids"][high_risk_idx]
visit_attn = predictions["visit_attention"][high_risk_idx]
feature_attn = predictions["feature_attention"][high_risk_idx]
print(f"\nHigh-risk patient: {patient_id}")
print(f"Risk score: {predictions['y_pred'][high_risk_idx]:.3f}")
print(f"Most influential visit: {visit_attn.argmax()}")
print(f"Most important features: {feature_attn[visit_attn.argmax()].argsort()[-5:]}")
# 10. Save model for deployment
trainer.save("./models/mortality_retain_final.pt")
print("\nModel saved successfully!")
Resources
For detailed information on each component, refer to the comprehensive reference files in the references/ directory:
- datasets.md: Data structures, loading, and splitting (4,500 words)
- medical_coding.md: Code translation and standardization (3,800 words)
- tasks.md: Clinical prediction tasks and custom task creation (4,200 words)
- models.md: Model architectures and selection guidelines (5,100 words)
- preprocessing.md: Data processors and preprocessing workflows (4,600 words)
- training_evaluation.md: Training, metrics, calibration, interpretability (5,900 words)
Total comprehensive documentation: ~28,000 words across modular reference files.
1---2name: pyhealth3description: Comprehensive healthcare AI toolkit for developing, testing, and deploying machine learning models with clinical data. This skill should be used when working with electronic health records (EHR), clinical prediction tasks (mortality, readmission, drug recommendation), medical coding systems (ICD, NDC, ATC), physiological signals (EEG, ECG), healthcare datasets (MIMIC-III/IV, eICU, OMOP), or implementing deep learning models for healthcare applications (RETAIN, SafeDrug, Transformer, GNN).4category: biology5license: MIT license6metadata:7 source: https://skillmd.com/skills/synthetic-sciences/pyhealth8 upstream: K-Dense-AI/scientific-agent-skills9 upstream-url: https://github.com/K-Dense-AI/scientific-agent-skills10 upstream-path: scientific-skills/pyhealth11 upstream-license: MIT12 upstream-relationship: derived13 adapted-by: Synthetic Sciences14 skill-author: K-Dense Inc.15---1617# PyHealth: Healthcare AI Toolkit1819## Overview2021PyHealth is a comprehensive Python library for healthcare AI that provides specialized tools, models, and datasets for clinical machine learning. Use this skill when developing healthcare prediction models, processing clinical data, working with medical coding systems, or deploying AI solutions in healthcare settings.2223## When to Use This Skill2425Invoke this skill when:2627- **Working with healthcare datasets**: MIMIC-III, MIMIC-IV, eICU, OMOP, sleep EEG data, medical images28- **Clinical prediction tasks**: Mortality prediction, hospital readmission, length of stay, drug recommendation29- **Medical coding**: Translating between ICD-9/10, NDC, RxNorm, ATC coding systems30- **Processing clinical data**: Sequential events, physiological signals, clinical text, medical images31- **Implementing healthcare models**: RETAIN, SafeDrug, GAMENet, StageNet, Transformer for EHR32- **Evaluating clinical models**: Fairness metrics, calibration, interpretability, uncertainty quantification3334## Core Capabilities3536PyHealth operates through a modular 5-stage pipeline optimized for healthcare AI:37381. **Data Loading**: Access 10+ healthcare datasets with standardized interfaces392. **Task Definition**: Apply 20+ predefined clinical prediction tasks or create custom tasks403. **Model Selection**: Choose from 33+ models (baselines, deep learning, healthcare-specific)414. **Training**: Train with automatic checkpointing, monitoring, and evaluation425. **Deployment**: Calibrate, interpret, and validate for clinical use4344**Performance**: 3x faster than pandas for healthcare data processing4546## Quick Start Workflow4748```python49from pyhealth.datasets import MIMIC4Dataset50from pyhealth.tasks import mortality_prediction_mimic4_fn51from pyhealth.datasets import split_by_patient, get_dataloader52from pyhealth.models import Transformer53from pyhealth.trainer import Trainer5455# 1. Load dataset and set task56dataset = MIMIC4Dataset(root="/path/to/data")57sample_dataset = dataset.set_task(mortality_prediction_mimic4_fn)5859# 2. Split data60train, val, test = split_by_patient(sample_dataset, [0.7, 0.1, 0.2])6162# 3. Create data loaders63train_loader = get_dataloader(train, batch_size=64, shuffle=True)64val_loader = get_dataloader(val, batch_size=64, shuffle=False)65test_loader = get_dataloader(test, batch_size=64, shuffle=False)6667# 4. Initialize and train model68model = Transformer(69 dataset=sample_dataset,70 feature_keys=["diagnoses", "medications"],71 mode="binary",72 embedding_dim=12873)7475trainer = Trainer(model=model, device="cuda")76trainer.train(77 train_dataloader=train_loader,78 val_dataloader=val_loader,79 epochs=50,80 monitor="pr_auc_score"81)8283# 5. Evaluate84results = trainer.evaluate(test_loader)85```8687## Detailed Documentation8889This skill includes comprehensive reference documentation organized by functionality. Read specific reference files as needed:9091### 1. Datasets and Data Structures9293**File**: `references/datasets.md`9495**Read when:**96- Loading healthcare datasets (MIMIC, eICU, OMOP, sleep EEG, etc.)97- Understanding Event, Patient, Visit data structures98- Processing different data types (EHR, signals, images, text)99- Splitting data for training/validation/testing100- Working with SampleDataset for task-specific formatting101102**Key Topics:**103- Core data structures (Event, Patient, Visit)104- 10+ available datasets (EHR, physiological signals, imaging, text)105- Data loading and iteration106- Train/val/test splitting strategies107- Performance optimization for large datasets108109### 2. Medical Coding Translation110111**File**: `references/medical_coding.md`112113**Read when:**114- Translating between medical coding systems115- Working with diagnosis codes (ICD-9-CM, ICD-10-CM, CCS)116- Processing medication codes (NDC, RxNorm, ATC)117- Standardizing procedure codes (ICD-9-PROC, ICD-10-PROC)118- Grouping codes into clinical categories119- Handling hierarchical drug classifications120121**Key Topics:**122- InnerMap for within-system lookups123- CrossMap for cross-system translation124- Supported coding systems (ICD, NDC, ATC, CCS, RxNorm)125- Code standardization and hierarchy traversal126- Medication classification by therapeutic class127- Integration with datasets128129### 3. Clinical Prediction Tasks130131**File**: `references/tasks.md`132133**Read when:**134- Defining clinical prediction objectives135- Using predefined tasks (mortality, readmission, drug recommendation)136- Working with EHR, signal, imaging, or text-based tasks137- Creating custom prediction tasks138- Setting up input/output schemas for models139- Applying task-specific filtering logic140141**Key Topics:**142- 20+ predefined clinical tasks143- EHR tasks (mortality, readmission, length of stay, drug recommendation)144- Signal tasks (sleep staging, EEG analysis, seizure detection)145- Imaging tasks (COVID-19 chest X-ray classification)146- Text tasks (medical coding, specialty classification)147- Custom task creation patterns148149### 4. Models and Architectures150151**File**: `references/models.md`152153**Read when:**154- Selecting models for clinical prediction155- Understanding model architectures and capabilities156- Choosing between general-purpose and healthcare-specific models157- Implementing interpretable models (RETAIN, AdaCare)158- Working with medication recommendation (SafeDrug, GAMENet)159- Using graph neural networks for healthcare160- Configuring model hyperparameters161162**Key Topics:**163- 33+ available models164- General-purpose: Logistic Regression, MLP, CNN, RNN, Transformer, GNN165- Healthcare-specific: RETAIN, SafeDrug, GAMENet, StageNet, AdaCare166- Model selection by task type and data type167- Interpretability considerations168- Computational requirements169- Hyperparameter tuning guidelines170171### 5. Data Preprocessing172173**File**: `references/preprocessing.md`174175**Read when:**176- Preprocessing clinical data for models177- Handling sequential events and time-series data178- Processing physiological signals (EEG, ECG)179- Normalizing lab values and vital signs180- Preparing labels for different task types181- Building feature vocabularies182- Managing missing data and outliers183184**Key Topics:**185- 15+ processor types186- Sequence processing (padding, truncation)187- Signal processing (filtering, segmentation)188- Feature extraction and encoding189- Label processors (binary, multi-class, multi-label, regression)190- Text and image preprocessing191- Common preprocessing workflows192193### 6. Training and Evaluation194195**File**: `references/training_evaluation.md`196197**Read when:**198- Training models with the Trainer class199- Evaluating model performance200- Computing clinical metrics201- Assessing model fairness across demographics202- Calibrating predictions for reliability203- Quantifying prediction uncertainty204- Interpreting model predictions205- Preparing models for clinical deployment206207**Key Topics:**208- Trainer class (train, evaluate, inference)209- Metrics for binary, multi-class, multi-label, regression tasks210- Fairness metrics for bias assessment211- Calibration methods (Platt scaling, temperature scaling)212- Uncertainty quantification (conformal prediction, MC dropout)213- Interpretability tools (attention visualization, SHAP, ChEFER)214- Complete training pipeline example215216## Installation217218```bash219uv pip install pyhealth220```221222**Requirements:**223- Python ≥ 3.7224- PyTorch ≥ 1.8225- NumPy, pandas, scikit-learn226227## Common Use Cases228229### Use Case 1: ICU Mortality Prediction230231**Objective**: Predict patient mortality in intensive care unit232233**Approach:**2341. Load MIMIC-IV dataset → Read `references/datasets.md`2352. Apply mortality prediction task → Read `references/tasks.md`2363. Select interpretable model (RETAIN) → Read `references/models.md`2374. Train and evaluate → Read `references/training_evaluation.md`2385. Interpret predictions for clinical use → Read `references/training_evaluation.md`239240### Use Case 2: Safe Medication Recommendation241242**Objective**: Recommend medications while avoiding drug-drug interactions243244**Approach:**2451. Load EHR dataset (MIMIC-IV or OMOP) → Read `references/datasets.md`2462. Apply drug recommendation task → Read `references/tasks.md`2473. Use SafeDrug model with DDI constraints → Read `references/models.md`2484. Preprocess medication codes → Read `references/medical_coding.md`2495. Evaluate with multi-label metrics → Read `references/training_evaluation.md`250251### Use Case 3: Hospital Readmission Prediction252253**Objective**: Identify patients at risk of 30-day readmission254255**Approach:**2561. Load multi-site EHR data (eICU or OMOP) → Read `references/datasets.md`2572. Apply readmission prediction task → Read `references/tasks.md`2583. Handle class imbalance in preprocessing → Read `references/preprocessing.md`2594. Train Transformer model → Read `references/models.md`2605. Calibrate predictions and assess fairness → Read `references/training_evaluation.md`261262### Use Case 4: Sleep Disorder Diagnosis263264**Objective**: Classify sleep stages from EEG signals265266**Approach:**2671. Load sleep EEG dataset (SleepEDF, SHHS) → Read `references/datasets.md`2682. Apply sleep staging task → Read `references/tasks.md`2693. Preprocess EEG signals (filtering, segmentation) → Read `references/preprocessing.md`2704. Train CNN or RNN model → Read `references/models.md`2715. Evaluate per-stage performance → Read `references/training_evaluation.md`272273### Use Case 5: Medical Code Translation274275**Objective**: Standardize diagnoses across different coding systems276277**Approach:**2781. Read `references/medical_coding.md` for comprehensive guidance2792. Use CrossMap to translate between ICD-9, ICD-10, CCS2803. Group codes into clinically meaningful categories2814. Integrate with dataset processing282283### Use Case 6: Clinical Text to ICD Coding284285**Objective**: Automatically assign ICD codes from clinical notes286287**Approach:**2881. Load MIMIC-III with clinical text → Read `references/datasets.md`2892. Apply ICD coding task → Read `references/tasks.md`2903. Preprocess clinical text → Read `references/preprocessing.md`2914. Use TransformersModel (ClinicalBERT) → Read `references/models.md`2925. Evaluate with multi-label metrics → Read `references/training_evaluation.md`293294## Best Practices295296### Data Handling2972981. **Always split by patient**: Prevent data leakage by ensuring no patient appears in multiple splits299 ```python300 from pyhealth.datasets import split_by_patient301 train, val, test = split_by_patient(dataset, [0.7, 0.1, 0.2])302 ```3033042. **Check dataset statistics**: Understand your data before modeling305 ```python306 print(dataset.stats()) # Patients, visits, events, code distributions307 ```3083093. **Use appropriate preprocessing**: Match processors to data types (see `references/preprocessing.md`)310311### Model Development3123131. **Start with baselines**: Establish baseline performance with simple models314 - Logistic Regression for binary/multi-class tasks315 - MLP for initial deep learning baseline3163172. **Choose task-appropriate models**:318 - Interpretability needed → RETAIN, AdaCare319 - Drug recommendation → SafeDrug, GAMENet320 - Long sequences → Transformer321 - Graph relationships → GNN3223233. **Monitor validation metrics**: Use appropriate metrics for task and handle class imbalance324 - Binary classification: AUROC, AUPRC (especially for rare events)325 - Multi-class: macro-F1 (for imbalanced), weighted-F1326 - Multi-label: Jaccard, example-F1327 - Regression: MAE, RMSE328329### Clinical Deployment3303311. **Calibrate predictions**: Ensure probabilities are reliable (see `references/training_evaluation.md`)3323332. **Assess fairness**: Evaluate across demographic groups to detect bias3343353. **Quantify uncertainty**: Provide confidence estimates for predictions3363374. **Interpret predictions**: Use attention weights, SHAP, or ChEFER for clinical trust3383395. **Validate thoroughly**: Use held-out test sets from different time periods or sites340341## Limitations and Considerations342343### Data Requirements344345- **Large datasets**: Deep learning models require sufficient data (thousands of patients)346- **Data quality**: Missing data and coding errors impact performance347- **Temporal consistency**: Ensure train/test split respects temporal ordering when needed348349### Clinical Validation350351- **External validation**: Test on data from different hospitals/systems352- **Prospective evaluation**: Validate in real clinical settings before deployment353- **Clinical review**: Have clinicians review predictions and interpretations354- **Ethical considerations**: Address privacy (HIPAA/GDPR), fairness, and safety355356### Computational Resources357358- **GPU recommended**: For training deep learning models efficiently359- **Memory requirements**: Large datasets may require 16GB+ RAM360- **Storage**: Healthcare datasets can be 10s-100s of GB361362## Troubleshooting363364### Common Issues365366**ImportError for dataset**:367- Ensure dataset files are downloaded and path is correct368- Check PyHealth version compatibility369370**Out of memory**:371- Reduce batch size372- Reduce sequence length (`max_seq_length`)373- Use gradient accumulation374- Process data in chunks375376**Poor performance**:377- Check class imbalance and use appropriate metrics (AUPRC vs AUROC)378- Verify preprocessing (normalization, missing data handling)379- Increase model capacity or training epochs380- Check for data leakage in train/test split381382**Slow training**:383- Use GPU (`device="cuda"`)384- Increase batch size (if memory allows)385- Reduce sequence length386- Use more efficient model (CNN vs Transformer)387388### Getting Help389390- **Documentation**: https://pyhealth.readthedocs.io/391- **GitHub Issues**: https://github.com/sunlabuiuc/PyHealth/issues392- **Tutorials**: 7 core tutorials + 5 practical pipelines available online393394## Example: Complete Workflow395396```python397# Complete mortality prediction pipeline398from pyhealth.datasets import MIMIC4Dataset399from pyhealth.tasks import mortality_prediction_mimic4_fn400from pyhealth.datasets import split_by_patient, get_dataloader401from pyhealth.models import RETAIN402from pyhealth.trainer import Trainer403404# 1. Load dataset405print("Loading MIMIC-IV dataset...")406dataset = MIMIC4Dataset(root="/data/mimic4")407print(dataset.stats())408409# 2. Define task410print("Setting mortality prediction task...")411sample_dataset = dataset.set_task(mortality_prediction_mimic4_fn)412print(f"Generated {len(sample_dataset)} samples")413414# 3. Split data (by patient to prevent leakage)415print("Splitting data...")416train_ds, val_ds, test_ds = split_by_patient(417 sample_dataset, ratios=[0.7, 0.1, 0.2], seed=42418)419420# 4. Create data loaders421train_loader = get_dataloader(train_ds, batch_size=64, shuffle=True)422val_loader = get_dataloader(val_ds, batch_size=64)423test_loader = get_dataloader(test_ds, batch_size=64)424425# 5. Initialize interpretable model426print("Initializing RETAIN model...")427model = RETAIN(428 dataset=sample_dataset,429 feature_keys=["diagnoses", "procedures", "medications"],430 mode="binary",431 embedding_dim=128,432 hidden_dim=128433)434435# 6. Train model436print("Training model...")437trainer = Trainer(model=model, device="cuda")438trainer.train(439 train_dataloader=train_loader,440 val_dataloader=val_loader,441 epochs=50,442 optimizer="Adam",443 learning_rate=1e-3,444 weight_decay=1e-5,445 monitor="pr_auc_score", # Use AUPRC for imbalanced data446 monitor_criterion="max",447 save_path="./checkpoints/mortality_retain"448)449450# 7. Evaluate on test set451print("Evaluating on test set...")452test_results = trainer.evaluate(453 test_loader,454 metrics=["accuracy", "precision", "recall", "f1_score",455 "roc_auc_score", "pr_auc_score"]456)457458print("\nTest Results:")459for metric, value in test_results.items():460 print(f" {metric}: {value:.4f}")461462# 8. Get predictions with attention for interpretation463predictions = trainer.inference(464 test_loader,465 additional_outputs=["visit_attention", "feature_attention"],466 return_patient_ids=True467)468469# 9. Analyze a high-risk patient470high_risk_idx = predictions["y_pred"].argmax()471patient_id = predictions["patient_ids"][high_risk_idx]472visit_attn = predictions["visit_attention"][high_risk_idx]473feature_attn = predictions["feature_attention"][high_risk_idx]474475print(f"\nHigh-risk patient: {patient_id}")476print(f"Risk score: {predictions['y_pred'][high_risk_idx]:.3f}")477print(f"Most influential visit: {visit_attn.argmax()}")478print(f"Most important features: {feature_attn[visit_attn.argmax()].argsort()[-5:]}")479480# 10. Save model for deployment481trainer.save("./models/mortality_retain_final.pt")482print("\nModel saved successfully!")483```484485## Resources486487For detailed information on each component, refer to the comprehensive reference files in the `references/` directory:488489- **datasets.md**: Data structures, loading, and splitting (4,500 words)490- **medical_coding.md**: Code translation and standardization (3,800 words)491- **tasks.md**: Clinical prediction tasks and custom task creation (4,200 words)492- **models.md**: Model architectures and selection guidelines (5,100 words)493- **preprocessing.md**: Data processors and preprocessing workflows (4,600 words)494- **training_evaluation.md**: Training, metrics, calibration, interpretability (5,900 words)495496**Total comprehensive documentation**: ~28,000 words across modular reference files.497