Eldercare Exercise: Voice-Guided Bedridden Rehabilitation & Home Assistant Integration

0 0 Updated: 2026-07-30 17:06:29

A lightweight, TTS-guided exercise program for bedridden or seated elderly. Features 3 levels (super light, light, moderate), integrates with Home Assistant for cron scheduling, motion detection, SOS conflict prevention, multi-elder support, and daily reporting. Safe for 90+ with physician approval required.

Install
npx skills add https://github.com/nclamvn/bom-cham --skill eldercare-exercise
Skill Details readonly

Hi everyone!

Today I want to share a project that I find particularly meaningful — an intelligent rehabilitation exercise program designed specifically for bedridden elderly peopleTo be honestI have elderly family members tooNot long agomy grandmother fell and had to stay in bed for a whileWatching her lying still day after daynot only were her muscles wasting awaybut her mood also became lowAt that timeI thought: if only there were something that could regularly guide Grandma through simple movements.

Then I discovered this projectand I was hooked at first sight! It's called eldercare-exercisea Skill developed by Vietnamese developer nclamvn for the Astro platform (notbut it runs on theenvironment)This Skill uses Home Assistant's TTS (Text-to-Speech) voice synthesis to guide the elderly through exercises step by step in language they can understand (Vietnamesebut the principle is universal)The key is: it takes into account the bedridden elderly's physical conditionhearing problemsand even whether it might conflict with other smart devicesIt's really thoughtful.

What exactly can it do?

Simply put: it creates a personalized private rehabilitation coach for your elderly family memberone that won't be interruptedThis coach calls out through the speaker in the room at a fixed time every day (e.g9:00 AM): "Bà ơimình tập thể dục nhẹ nha!" (Grandmalet's do some light exercise!)Then it leads through the exercises section by sectionThe movements are super simplelike deep breathingwrist rotationsfinger flexing and extendingarm liftsankle pumpsAnd after each movementit pauses for 10 seconds to let the elderly restThe voice speed is also slowed downcounting the rhythm one by one.

This routine is divided into three levels:

  • Level 1 (Ultra Light): Suitable for elderly who cannot get out of bed at all7 movementsabout 10 minutes totalAll movements are done lying in bedsafety first.
  • Level 2 (Light): Suitable for elderly who can sit upBased on Level 1adds shoulder rotationsside-lyingleg raises in sitting positionAbout 15 minutes.
  • Level 3 (Moderate): Suitable for elderly who can stand with supportAdds holding a chair to standmarching in placeheel raisesAbout 20 minutes.

And you don't need to worry about whether the elderly person can followbecause every instruction is read out through TTS as a description of the movementFor example: "Bà nâng tay phải lên từ từhạ xuống" (Grandma slowly lift your right handlower it)Super clear!

It's truly smart when integrated with Home Assistant

This Skill doesn't run independently; it needs to work with Home AssistantI happen to have HA set up at homeso installation was very convenientThe workflow is roughly like this:

  • Detect if the elderly person is awake: Use a motion sensor in the room (e.gsensor.grandma_room_motion_minutes) to check if there has been activity in the last 30 minutesIf the elderly person is still sleepingskip and postpone to 10:00 AM for retryavoiding disturbance.
  • Avoid conflicts: If the elderly person is on an emergency call (SOS) or video callthe exercise won't start either.
  • Record logs: After each sessionrecord the dateleveldurationand completion statusand store them in memoryThis allows family members to ask via chat"Has Grandma exercised today?" and get an immediate answer.
  • Chat control: You can control everything by sending messages through your phoneFor example: "Start exercise""Stop exercise""Switch to level 2""Have Grandma exercise now"etcThe Skill responds to all these commands.

Ohand it can also automatically send the daily exercise report to the family Zalo group (similar to WeChat)like: "🏋️ Grandma completed Level 1 exercise this morning at 9:15 (10 minutes)"So even when you're at workyou can stay informed about the elderly person's condition.

Installation & Configuration Guide

If you want to try it toothe steps are simple (you need a Home Assistant environment firstand theSkill framework installed):

npx skills add https://github.com/nclamvn/bom-cham --skill eldercare-exercise

After installationyou need to configure a few entities:

  • A media player (e.gXiaomi speaker or Google Home) to play TTS
  • A motion sensor to detect activity minutes in the elderly person's room
  • Optional: configure Zalo group chat bot for sending reports

Then send a chat message "bật tập thể dục" (start exercise) to activate itBy defaultit reminds at 9:00 AM every day with Level 1You can also use "đổi level 2" (change to level 2) to upgrade the difficulty.

Special note: The developer emphasizes that all exercise plans should be evaluated by a doctor firstIf the elderly person has heart diseasefracturesetcdon't mess aroundThe settings are disabled by defaultand must be manually enabled by family membersThis design is very safe.

Why I think this is great

What touched me most after using it is the human-centered details: the voice volume is set to maximum (because elderly people often have poor hearing)the speech rate is slowed to 0.7 (much slower than normal)there is a 10-second rest between each movement (since elderly people move slowly)and after each movementit praises with "Giỏi quá!" (Great job!)Every time Grandma hears thatshe smiles and is willing to keep going.

There's also a "multi-elderly support" featureIf you have more than one elderly person living together (e.ggrandma and grandpa)you can assign different sensors and speakers to each personand the Skill will automatically select which elderly person is at home based on the profileHoweverthis requires additional configuration of eldercare_profilesdocumentation(the original text was cut off).

📢 Sponsored