BlueCLActivity records
Sensor events and profile information are handled by the application backend. These records can describe a person’s daily routine.
Liflow / Sensor intelligence · Team project
Everyday signals, easier to understand.
Liflow is a collaborative life-safety monitoring project connecting sensors, activity records and a guardian-facing interface. Weekly summaries bring event counts, common activities and changes into a readable view.
Development complete
Three steps
From the first step to a useful result.
- 01
Set up a care profile
The app workflow starts with the person and sensor configuration.
- 02
Collect activity signals
Connected sensors provide time-stamped activity events.
- 03
Review the weekly picture
Read the activity records alongside the weekly summary.
PRODUCT / IN VIEW
See the product.
Explore the interface, then open the service, source code or demo using the links above. Select an image to see it in full.
Getting started
Start with the right setup.
Access & setup
- Watch the published sensor and NFC demos to understand the interaction.
- Use the repository’s setup documentation to configure the app, backend and sensor integration.
- Use a controlled development setup before connecting personal activity data.
Permissions & data
Understand what the workflow uses.
Weekly summary input
Prepared activity counts and timing statistics are sent to the model provider. The summary describes activity records and is not a clinical assessment.
This is product usage guidance. The BlueCL website privacy notice covers this company website; consult the service’s own documentation for its deployed data handling. Website privacy ↗
Behind the product
How the product works.
A long list of sensor events is difficult to review. Guardians need an understandable view of recorded activity, including how often events occurred, common activities and changes from the previous week.
01Sensor eventsRecorded activity in the backend
02Seven-day contextCounts · frequent activities · previous week
03Activity summaryFactual Korean activity summary
04Guardian viewStored summary · reuse until data changes
Model integration
The FastAPI weekly-summary endpoint prepares seven-day activity statistics and calls Claude Sonnet 4.6 to produce two or three Korean sentences. The model summarizes recorded activity; sensor collection and event processing remain separate backend functions.
Implementation details
The summary cache checks the period, event count and latest event time. Empty periods return a local message without a model request. Output instructions exclude names, IDs and phone numbers and keep the summary within observed activity records.
Further improvements
Improve source-to-summary traceability and evaluate whether each sentence stays within the observed activity records.
When you need a hand
Frequently asked questions.
Is there a public web app I can open?
This page provides the team repository and public demos. A current public app URL has not been provided.
When does the summary refresh?
The cache checks the period, event count and latest event time. A changed event set triggers a new summary.
What happens when there are no records?
An empty period returns an informational message without requesting a model summary.
Product and source information updated: October 8, 2026