Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/malikwhitten67/apple-doctor

Apple Doctor is an app aimed to enhance self-assessments / doctor workflows.
https://github.com/malikwhitten67/apple-doctor

Last synced: 5 days ago
JSON representation

Apple Doctor is an app aimed to enhance self-assessments / doctor workflows.

Awesome Lists containing this project

README 

        
# Apple Doctor



Apple doctor is an ai revolved app made to simplify self analysis - and workflow management,  communicating between doctors and patients



- Made by a mark cuban ai camp student.



### Tech Stack



1. Pocketbase - used for persistent reflections between patient's and Doctor's

2. Daisyui x Tailwindcss - css library making css easier and efficient.

3. React x Astro - server side rendering for fast cms

4. Express x machine learning 4 kids api - for ai to frontend communication and model training



## Running/Installing



1. Install all frontend dependencies used for development



```bash

npm i

```



* Setting up backend

   - go to [machine_learning_4kids](machinelearningforkids.co.uk) and click try it - then create a new project and fill 2 labels with data from `data.json`

        - the data you fill in is the symptoms copy each symptom inside of " " then apply it to the label.

        -  go back and select test and train your model



   - Dowload the backend code from the backend branch then continue to do the following:

       



   - Next click make - python and then copy your api key -> your then going to place key inside of .env file

     

     ```env

     api_key=key

     ```

  - or if using [replit](https://replit.com/@MALIKWhitten/Apple-Doctor-backend) then refer to secrets -> click the three dots next to api_key and replace it with your own

    ![Replit](https://github.com/MalikWhitten67/Apple-Doctor/assets/65188863/9a75d59b-6884-40fe-bd23-9a002eb09512)

    Skip the next few if you are using replit

    

  - Download the dataset from branch backend and then create a config.json file in the main dir where server.js is located

  ```json

  {

    "min_confidence":"20",

   "rated_confidence":"60",

   "dataset":"./data.json"

  }

  ```

  - Now you can run server.js which sets up all endpoints

  - You can check the status of the model on the [status page](http://127.0.0.1:3000/status)

  - When you get to the frontend portion be sure to swap the api url for analyzing requests with yours! 

    

* Database

   - Pretty simple go to [pocketbase website](https://pocketbase.com) download the latest version run the following command || you can use [Pockethost](https://pockethost.io) which can host it for free on their servers

     

    - To start the server run the following in terminal either windows or linux:



    ```bash

     pocketbase serve

     ```

    

   - goto [admin panel](http://127.0.0.1:8090/_/) || `https://your_pockethost_api_route.pockethost.io/_/` - and finish installment

 

    

   - Now that you have an database swap to backend branch go into pocketbase folder and download the schema

   - Go back to your panel - then settings and under import collections either click to upload the file or copy and paste

     

* Run the frontend

    - Download the code and do the following

    - Swap the api route to your url

      - ```js

     const api = new Pocketbase(`https://replace_with_url/`) // do not include /_/ or /api just the base route `https://route`

     ```

    - Your done! that was easy now we need to just run the frontend

     

    ```bash

     npm run dev

    ```



# Dev Log



1. Oct 28th: 

   - Implemented Login / Register both as a patient and a doctor

   - Started working on the dashboard

   - Reworked Readme

2. Oct 29th: Design layout refractoring - made  the ai navigatable / consistent and easy to use

   - Doctors ui is seperate from user ui

   - Messages page done - you can create chats by searching your doctor or vice versa

     - realtime reflective messages work aswell

     - grabbing useful data for accessments - analyzing data from reputable sources and making useful prompts to better get a quality answer.

     - Optimized ai backend api to better give an answer based on median confidency rating

3. Oct 30th:

     - Finished working on analysis output / analyzing for user dashboard

     - Reorganized / fixed dataset inconsistencies by using concise inputs

     - small bug fixes on backend

     - Proper ui sizing for pc monitors

     - Adding common symptoms to datasets for users to view - to get a general understanding of the disease/illness

     - Adding suggestions - which can tell if a user can do it at home or if they should contact their doctor about the issue

     - Cleaned the ui for pc users

     - Made it so you cannot create duplicated chats

     - Added chat deletion option - and send to doctor button works

     - Added bulk label updates allowing you to append data to multiple labels from the dataset all at once and auto train the model



4. Oct 31st

    - Did alot of ui fixes and enhancements

    - Added more data to the database cleaned up the dataset to be more accurate for both doctors and regular users

    - Implemented Attachments in messages allowing users to send their assements as an attachment inwhich shows as a card in the chat

    - Bug fixes on the backend & frontend

      

5.  Nov 1st - 3rd

    - Implemented doctor and user profile editing avatars and more

    - Finished settings page

    - Added message deleting by hovering

    - Fixes to message ui

    - Published the website to be presented