ClimAID — Climate change impact using AI on Diseases
Version 0.4.1 (beta, under active testing) · Documentation · PyPI · Changelog · Paper
ClimAID is an integrated toolkit for modelling, forecasting and projecting climate-sensitive diseases such as dengue and malaria, using machine learning, a climate-informed transmission model and CMIP6 climate-model ensembles.
- Built-in climate data for South Asia: India, Nepal, Bhutan, Sri Lanka, Myanmar, Afghanistan, Pakistan and Bangladesh.
- Data from other countries are supported through the global mode of the browser interface.
Under active testing. ClimAID 0.4.1 has been checked on synthetic data with a known answer, but not yet validated on real surveillance data or real CMIP6 projections. Treat outputs as research estimates and do not use them as the sole basis for public-health decisions. See Status & validation.
What's new in 0.4.1
climaid ai: built-in assistant. Describe what you want in plain language ("forecast dengue in Pune for the next 6 months"); it asks for anything missing, runs the forecast or climate outlook, and explains the results and the methods. Fully offline and with no AI model: every number comes from ClimAID's own results. In the terminal, or as a chat page (climaid ai --browser, or Assistant in the dashboard).- SARIMAX as an optional v2 model (seasonal ARIMA with climate inputs), tuned, backtested and calibrated like every other model. Best of all models on the non-seasonal synthetic benchmark; weaker on the realistic one, so not a default.
0.4.0 was the first public release of ClimAID v2 (the previous public release was 0.1.2, v1 only):
probabilistic forecasts with calibrated likely ranges, a climate-informed renewal (transmission) model, a
seasonal-naive benchmark, models with compulsory leakage-safe tuning, rolling backtests ("hindcasts"), a hybrid
near-term + CMIP6 scenario outlook (DiseaseModel.project_v2()), plain-language reports with a trust rating,
and leakage fixes in v1. v1 metrics from 0.1.x were optimistic; rerun before citing them.
Full list, including every change that alters results: CHANGELOG.md.
Installation
pip install climaid # core
pip install "climaid[ml]" # + XGBoost, LightGBM, CatBoost
Requires Python 3.10 or newer. For development: pip install -e ".[test]" from a clone of this repository.
Optional local-LLM reports: install Ollama and run ollama serve. No extra Python package is
needed; without it, ClimAID uses its built-in deterministic C-DSI reports.
Two pipelines
| ClimAID v2 (recommended) | ClimAID v1 (legacy) | |
|---|---|---|
| Purpose | Forecasts for the coming months, and climate-scenario outlooks to 2050–2100 | Lag-optimised point predictions and CMIP6 projections |
| Output | Expected cases with likely ranges | Expected cases |
| Models | 21, including a climate transmission model and v1's stacked model | Stacked base → residual → correction models |
| Checked by | Held-out months, rolling backtests, calibrated ranges | One test year |
| Report | Plain-language summary + technical details | C-DSI deterministic report |
Quick examples
from climaid.climaid_model import DiseaseModel
dm = DiseaseModel(district="IND_Pune_MAHARASHTRA", disease_file="dengue.xlsx", disease_name="Dengue")
# v2: probabilistic forecast for the next 12 months
result = dm.forecast_v2(forecast_origin="2023-12-31", horizon=12, tuning="balanced", save_report=True)
print(result["report_path"])
# v2: climate-scenario outlook
outlook = dm.project_v2(end_year=2050, ssps=["ssp245", "ssp585"])
print(outlook["decades"])
# v1 (legacy)
dm.optimize_lags()
dm.train_final_model()
Lower-level v2 engine and migration notes: MIGRATION_v2.md.
Interfaces
climaid ai # built-in assistant: describe what you want; it guides you and runs ClimAID (offline)
climaid ai -b # the same assistant as a chat page in your browser (also in the dashboard menu)
climaid browse # browser dashboard (South Asian and global data); v2 and v1 pages
climaid wizard # terminal wizard (South Asian data); choose v1, v2 or both
climaid docs # open the documentation bundled with this installation (offline)
climaid --version
Documentation and project files
| Where | What |
|---|---|
| sam-as.github.io/ClimAID | User guide, API reference, status & validation, changelog |
| CHANGELOG.md | All changes, by version |
| MIGRATION_v2.md | Moving from v1 to v2 |
| V2_FEEDBACK_RESPONSE.md | Reviewer concerns and how v2 addresses them |
| benchmarks/ | Synthetic datasets, benchmark runner and saved results |
| README_DOCS.md | Building and releasing the documentation |
Designed for
Epidemiologists, climate scientists, public-health analysts and data scientists.
Citation
Sam, A.K., Phuleria, H.C. (2026). ClimAID: An AI-integrated Global Hybrid Climate-Disease Modelling Framework. Preprint: https://doi.org/10.21203/rs.3.rs-9394047/v1
License
MIT. Designed by Avik Kumar Sam & Harish C. Phuleria as open-access software through the National Disease Modelling Consortium.
-
Full text: LICENSE.
-
For technical feedback: avik.sam@iitb.ac.in | avik.sam@nus.edu.sg.
Metadata
Release files for climaid 0.4.1
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 |
|---|---|---|---|---|
| climaid-0.4.1.tar.gz | 12.0 MB | With provenance | Details for climaid-0.4.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| climaid-0.4.1-py3-none-any.whl | Python 3 | none | any | Details for climaid-0.4.1-py3-none-any.whl |
Total release size: 24.4 MB
Release files / climaid-0.4.1.tar.gz
| Download URL | climaid-0.4.1.tar.gz |
|---|---|
| Size | 12.0 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
2e7d55db9ee9650abfa911ad133551163fba1b06f1f33dc35455ab1bc9aad7e4
|
|
BLAKE2b-256 checksum How to use checksums |
12345ea3f4dc93629e95ba9e596ccb7c27bec947d7a11a0bb6f3f6606004b379
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 2, 2026.
Transparency logRelease files / climaid-0.4.1-py3-none-any.whl
| Download URL | climaid-0.4.1-py3-none-any.whl |
|---|---|
| Size | 12.4 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
8e27f6f1f3ed9c40a67268b5e715d7c5c81b9d3fbed7ffd8d748bc8fa865ece5
|
|
BLAKE2b-256 checksum How to use checksums |
358a61db43cd8d72ae6c769ae798e4696f88c614845b34d8932f91e46089168a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
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.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 2, 2026.
Transparency logSecurity for climaid 0.4.1
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.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for climaid-0.4.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 2, 2026.
Transparency logA 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.