GripMind

GripMind 是一套結合 IoT、LINE Bot、Web Dashboard 與 AI 訓練回饋 的智慧握力復健原型系統。
系統可接收握力裝置上傳的量測資料，記錄使用者每日訓練狀況，並透過 LINE 推播、網頁圖表與 AI 回饋，協助使用者建立更穩定的手部肌力訓練習慣。

本專案為智慧照護與復健輔助系統原型，不提供醫療診斷，也不應取代醫師、物理治療師或職能治療師的專業建議。

Project Overview

GripMind 的核心目標是解決傳統握力訓練裝置缺乏「紀錄、提醒、回饋」的問題。

一般握力器通常只能提供單次訓練，使用者很難知道：

今天是否有完成訓練
握力是否有進步
是否達成個人目標
是否需要調整訓練強度
長期資料是否能被追蹤與分析

因此，GripMind 將握力訓練流程拆成四個階段：

量測：由握力裝置取得握力數值
上傳：裝置將資料送至 Flask API
紀錄：後端儲存使用者與握力歷史資料
回饋：透過 LINE、Web Dashboard 與 AI 產生訓練建議

Key Features

1. IoT Grip Data Upload

裝置端可透過 API 上傳握力資料：

裝置 ID
握力數值
驗證 Token

後端會將資料儲存至 JSON 檔案，並保留歷史紀錄。

2. LINE Login Device Binding

使用者可透過 LINE Login 綁定自己的握力裝置。
系統會將 LINE 使用者與裝置 ID 對應起來，後續即可針對特定使用者推播訓練提醒與握力回饋。

3. LINE Bot Push Notification

當裝置上傳握力資料後，系統會透過 LINE Bot 推播使用者當日握力紀錄，例如：

今日握力紀錄：3.8 kg，你已完成今日目標，繼續加油！

系統也具備每日檢查邏輯，可用於提醒尚未完成訓練的使用者。

4. Web Dashboard

使用者可透過網頁查看指定裝置的握力歷史紀錄。

目前包含：

裝置資料查詢
握力歷史紀錄
目標握力顯示
使用者點數顯示
AI 分析入口

5. Target Weight Management

每位使用者可設定或更新自己的目標握力值。系統會根據目標值判斷當日訓練是否達標。

6. AI Training Feedback

GripMind 整合本地端 Ollama 模型，根據使用者的：

年齡
性別
身體狀況
訓練方式
目標握力
最近握力紀錄

產生簡短、保守、繁體中文的訓練回饋。

AI 回饋只作為訓練輔助，不作為醫療診斷。

7. REST API v1

本專案提供 /api/v1 API，方便未來串接：

手機 App
管理後台
研究資料分析工具
其他 IoT 裝置

System Architecture

Grip Device
Arduino / ESP8266 / Sensor
        │
        │ POST /gripdata
        ▼
Flask Backend
app.py / send.py / api_v1.py
        │
        ├── JSON Data Storage
        │   ├── users.json
        │   ├── data.json
        │   └── log.json
        │
        ├── LINE Bot
        │   ├── Login Binding
        │   ├── Push Message
        │   └── Auto Reply
        │
        ├── Web Dashboard
        │   ├── History Page
        │   ├── Target Setting
        │   └── Admin Page
        │
        └── AI Feedback
            └── Ollama Local Model

Repository Structure

gripmind/
├── app.py                 # Main Flask application
├── api_v1.py              # REST API v1 routes
├── send.py                # LINE push, data storage, user binding logic
├── ai_client.py           # Ollama AI client
├── requirements.txt       # Python dependencies
├── render.yaml            # Render deployment configuration
├── Procfile               # Deployment process file
├── docs/
│   └── api.md             # API v1 documentation
├── json/
│   ├── users.json         # User and device binding data
│   ├── data.json          # Grip records
│   ├── keep.json          # Runtime data
│   └── log.json           # System logs
├── templates/             # HTML templates
├── static/                # Static assets
├── temp/                  # Development utilities
└── zip/                   # Exported data archive

Tech Stack

Layer	Technology
Backend	Python, Flask
API	Flask Blueprint
Frontend	HTML, Jinja Templates
Data Storage	JSON files
Messaging	LINE Messaging API
Authentication	LINE Login
AI Feedback	Ollama local model
Deployment	Render, Gunicorn
License	MIT License

Getting Started

1. Clone the Repository

git clone https://github.com/cloud-driver/gripmind.git
cd gripmind

2. Create a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate

For Windows:

python3 -m venv .venv
source .venv/bin/activate

For Windows:

python -m venv .venv
.venv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Create `.env`

Create a .env file in the project root.

SECRET_TOKEN=your_device_upload_token
SECRET_KEY=your_flask_secret_key
SECRET_PAGE_PASSWORD=your_admin_page_password

URL=http://localhost:10000

LINE_LOGIN_CHANNEL_ID=your_line_login_channel_id
LINE_LOGIN_CHANNEL_SECRET=your_line_login_channel_secret
MESSAGING_API_ACCESS_TOKEN=your_line_messaging_api_access_token

OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_MODEL=qwen2.5:7b-instruct
OLLAMA_TIMEOUT=120

5. Run the Server

python app.py

Default local URL:

http://localhost:10000

API Usage

Upload Grip Data

POST /gripdata
Content-Type: application/json

Request body:

{
  "device_id": "device_demo_001",
  "grip": 3.8,
  "token": "your_device_upload_token"
}

Example:

curl -X POST http://localhost:10000/gripdata \
  -H "Content-Type: application/json" \
  -d '{"device_id":"device_demo_001","grip":3.8,"token":"your_device_upload_token"}'

API v1

Base URL:

/api/v1

Health Check

GET /api/v1/health

Get Device Records

GET /api/v1/devices/{device_id}/records

Optional query:

?limit=20

Get Device Summary

GET /api/v1/devices/{device_id}/summary

Returns:

Latest grip record
Today’s record count
Today’s maximum grip
Today’s average grip
Target weight
Goal reached status

Get Device Profile

GET /api/v1/devices/{device_id}/profile

Update Target Weight

PATCH /api/v1/devices/{device_id}/target
Content-Type: application/json

Request body:

{
  "target_weight": 4.0
}

Generate AI Training Suggestion

POST /api/v1/devices/{device_id}/analysis

Response example:

{
  "device_id": "device_demo_001",
  "suggestion": "最近握力表現穩定，建議您維持目前訓練頻率...",
  "disclaimer": "This suggestion is for training feedback only and is not a medical diagnosis."
}

For more details, see:

docs/api.md

Deployment

This project includes a Render configuration file:

services:
  - type: web
    name: gripmind-api
    env: python
    buildCommand: ""
    startCommand: gunicorn app:app
    autoDeploy: true

When deploying, remember to set the required environment variables in the Render dashboard.

Current Status

GripMind is currently a working prototype.

Implemented:

Flask backend
Web dashboard
LINE Login binding flow
LINE Bot message push
Grip data upload API
JSON-based local data storage
Target weight management
AI training feedback through Ollama
API v1 documentation
Render deployment configuration

Planned improvements:

Replace JSON files with a real database
Add complete hardware setup documentation
Add Arduino / ESP8266 firmware examples
Add unit tests
Add Docker support
Improve admin dashboard security
Add mobile app interface
Add multi-device support for one user
Add data export for rehabilitation tracking

Important Notes

This project is a prototype for smart rehabilitation assistance.

Please do not use GripMind as:

A medical diagnosis system
A substitute for professional rehabilitation treatment
A certified medical device
A final clinical decision-making tool

The AI feedback module is designed to provide conservative training suggestions only.

License

This project is licensed under the MIT License.

See:

LICENSE

Author

GripMind Team Created by Justus Cheng

Name		Name	Last commit message	Last commit date
Latest commit History 71 Commits
docs		docs
json		json
static		static
temp		temp
templates		templates
zip		zip
.gitattributes		.gitattributes
.gitignore		.gitignore
.renderignore		.renderignore
LICENSE		LICENSE
Procfile		Procfile
README.md		README.md
ai_client.py		ai_client.py
api_v1.py		api_v1.py
app.py		app.py
render.yaml		render.yaml
requirements.txt		requirements.txt
send.py		send.py
temp.txt		temp.txt

Folders and files

Latest commit

History

Repository files navigation