Automated wellbeing reports, trend analysis & NLG for ReMeLife
This project is licensed under the [Apache 2.0 License with ReMeLife custom clauses]
This module generates clear, human-readable reports from ELR® data, activity logs, and engagement metrics. It turns raw signals into actionable summaries for families, carers, and clinicians—automatically and on schedule.
- Natural Language Generation (NLG): Convert structured metrics into caregiver-friendly narratives.
- Wellbeing Trend Analysis: Detect changes in mood, engagement, cognition across time windows.
- Visual Summaries (optional): Produce charts/plots (PNG/SVG) for dashboards or PDFs.
- Template System: Reusable report templates per audience (family vs clinician).
- APIs & Agent Tools: Expose “generate_report”, “get_trends” for the LUKi agent or external services.
- NLG & Summarisation:
- Lightweight templates (Jinja2) + rule-based sentence assembly
- Optional LLM-assisted summarisation via internal agent (not in this repo)
- Analytics & Time Series: pandas, statsmodels/prophet (optional)
- Data Viz: matplotlib / plotly (static export)
- Schema & Validation: pydantic
- Orchestration: LangChain tools to let LUKi trigger reports
luki_modules_reporting/
├── __init__.py
├── config.py
├── data/
│ ├── schemas.py # pydantic models: ActivityLog, MoodEntry, etc.
│ └── loaders.py # adapters to pull metrics from stores/APIs
├── analytics/
│ ├── aggregate.py # rollups, stats
│ ├── trends.py # time-series analysis
│ └── viz.py # chart generators (png/svg)
├── nlg/
│ ├── templates/
│ │ ├── family.j2
│ │ └── clinician.j2
│ ├── builder.py # assemble narrative from stats + templates
│ └── summariser.py # optional LLM summarisation hook
├── interfaces/
│ ├── agent_tools.py # LangChain @tool wrappers
│ └── api.py # FastAPI endpoints (optional)
└── tests/
git clone [email protected]:REMELife/luki-modules-reporting.git
cd luki-modules-reporting
python -m venv venv && source venv/bin/activate
pip install -r requirements.txtfrom datetime import date, timedelta
from luki_modules_reporting.data.loaders import load_demo_metrics
from luki_modules_reporting.analytics.aggregate import aggregate_metrics
from luki_modules_reporting.nlg.builder import build_report
# 1. Load demo data (replace with real ELR/metrics adapters)
metrics = load_demo_metrics(user_id="user_123",
start=date.today()-timedelta(days=7),
end=date.today())
# 2. Aggregate & analyse
stats = aggregate_metrics(metrics)
# 3. Build narrative
report_text = build_report(stats, audience="family") # or "clinician"
print(report_text)from luki_modules_reporting.analytics.viz import activity_chart
fig_path = activity_chart(metrics, out_path="outputs/activity.png")
# pass fig_path to the template context inside build_report(...)# interfaces/agent_tools.py
from langchain.tools import tool
from .nlg.builder import build_report
from .data.loaders import fetch_metrics_window
from .analytics.aggregate import aggregate_metrics
@tool("generate_wellbeing_report", return_direct=True)
def generate_wellbeing_report(user_id: str, days: int = 7) -> str:
"""Return a text wellbeing report for the last N days."""
metrics = fetch_metrics_window(user_id=user_id, days=days)
stats = aggregate_metrics(metrics)
return build_report(stats, audience="family")- Do not log raw ELR text in this repo; only derived stats.
- Encrypt any temporary files (figures, PDFs) at rest.
- Respect consent flags—exclude hidden/sensitive categories from outputs.
- Keep PHI out of public issues; use synthetic examples.
- PDF export service (WeasyPrint / ReportLab)
- Multi-language report templates (i18n)
- Clinician-specific metrics (MMSE scores, med adherence)
- Alerting: threshold-based notifications (e.g., sudden drop in engagement)
- Differential privacy for aggregated reports across cohorts
Open to PRs. Follow CONTRIBUTING.md, keep tests green, and document new templates.
Apache-2.0 © 2025 Singularities Ltd / ReMeLife.
(Add via GitHub “Choose a license template” or paste the standard Apache-2.0 text in LICENSE.)
Turn data into insight. Help carers act, not guess.