FairHealth
Trustworthy Healthcare AI — built from peer-reviewed research.
FairHealth is an open-source Python library for building fair, explainable, and privacy-preserving machine learning models for healthcare.
Built by Farjana Yesmin from 5 accepted research papers.
Install
pip install fairhealth
Modules and Results
| Module | What | Paper | Key Result |
|---|---|---|---|
fairhealth.fairness |
Demographic parity, equalized odds, disparate impact | MobiHealth 2026 | DI: 0.23 → 0.71 |
fairhealth.explain |
SHAP wrappers + Fuzzy-XGBoost hybrid | ICAIHE 2026, Waseda | 88.67% acc, 71.4% clinician preference |
fairhealth.federated |
FedAvg + CKKS HE + differential privacy | MedHE, CIBB 2026 | macro-F1=0.950, 97.5% comm reduction |
fairhealth.lowresource |
Dengue triage, multilingual, low-bandwidth | DASGRI 2026, Springer | F1=0.802, 75% satisfaction |
fairhealth.equity |
Fairness-aware flood aid allocation | CCAI 2026, IEEE | SPD↓41.6%, R²=0.784 |
Quick Example
import fairhealth as fh
import numpy as np
# Fairness audit
from fairhealth.fairness.metrics import demographic_parity_diff
dpd = demographic_parity_diff(
y_pred = np.array([1, 0, 1, 0, 1, 0]),
sensitive = np.array([0, 0, 0, 1, 1, 1])
)
print(f"DPD: {dpd:.4f}") # → 0.3333
# Dengue triage — English + Bangla
from fairhealth.lowresource.triage import assess_dengue_risk
result = assess_dengue_risk(age=8, gender="male",
area_type="urban", district="Dhaka",
language="bangla")
print(result["recommendation"]) # বাংলা output
# Flood aid equity
from fairhealth.equity.flood_aid import generate_priority_ranking
rankings = generate_priority_ranking(verbose=False)
print(f"Top priority: {rankings[0]['district']}") # → Sunamganj
# Federated privacy
from fairhealth.federated.privacy import sparsify
_, rate = sparsify(np.random.randn(1000), sparsity=0.975)
print(f"Communication reduced: {rate:.1%}") # → 97.5%
Research Papers
All papers are accepted. Preprint links below; final proceedings forthcoming.
| Paper | Venue | Preprint | Status |
|---|---|---|---|
| ECG Fairness | MobiHealth 2026 (EAI) | ResearchGate | Accepted |
| Maternal Health XAI | ICAIHE 2026, Waseda | ResearchSquare | Accepted |
| MedHE Federated | CIBB 2026 | arXiv:2511.09043 | Under review |
| Dengue Triage | DASGRI 2026, Springer LNNS | ResearchGate | Accepted |
| Flood Aid Equity | CCAI 2026 (IEEE), oral | arXiv:2512.22210 | Accepted |
Datasets (All Public — No Hospital Access Required)
| Dataset | Domain | Source |
|---|---|---|
| PTB-XL (4,367 records) | ECG biosignals | PhysioNet |
| Maternal Health Risk (1,014) | Risk prediction | UCI ML Repository |
| UCI Drug Reviews (215K) | NLP / drug effectiveness | UCI ML Repository |
| Bangladesh Dengue (4,700) | Symptom triage | Kaggle + DGHS |
| Bangladesh PDNA 2022 (87 upazilas) | Flood equity | Government open data |
Cite
@article{yesmin2026fairhealth,
author = {Yesmin, Farjana},
title = {FairHealth: An Open-Source Python Library for
Trustworthy Healthcare AI in Low-Resource Settings},
journal = {arXiv preprint arXiv:2605.08198},
year = {2026},
url = {https://arxiv.org/abs/2605.08198}
}
@software{fairhealth2026,
author = {Yesmin, Farjana},
title = {FairHealth: Trustworthy Healthcare AI},
year = {2026},
url = {https://github.com/Farjana-Yesmin/fairhealth}
}
Author: Farjana Yesmin · farjana-yesmin.github.io · MIT License
Metadata
Release files for fairhealth 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| fairhealth-0.1.1.tar.gz | 19.8 kB | Without provenance | Details for fairhealth-0.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| fairhealth-0.1.1-py3-none-any.whl | Python 3 | none | any | Details for fairhealth-0.1.1-py3-none-any.whl |
Total release size: 41.9 kB
Release files / fairhealth-0.1.1.tar.gz
| Download URL | fairhealth-0.1.1.tar.gz |
|---|---|
| Size | 19.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
6663117095f28cdc2c4353035b9bfa9b77b01559bd6924a3e8e6bb6f285cb359
|
|
BLAKE2b-256 checksum How to use checksums |
420db04b1c1ec338228d07365afd037156adb436a59f96973192fceea40db45c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.13
|
Release files / fairhealth-0.1.1-py3-none-any.whl
| Download URL | fairhealth-0.1.1-py3-none-any.whl |
|---|---|
| Size | 22.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
c68eee326ec948eddffed51c1f30eea177e5834483f2d129d8af39da83900fa8
|
|
BLAKE2b-256 checksum How to use checksums |
1c6fd04bf8ac38b5188bc6161aa4f781aed6e9ea7dd6234bea7e5cfc608e27df
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.13
|
Security for fairhealth 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.