SoulDiaryConnect

SoulDiaryConnect is an AI-powered system designed to support patients in their psychotherapeutic journey by enabling journaling with personalized AI feedback, while keeping the therapist connected and in control. The platform allows patients to log daily experiences, receive AI-generated motivational and clinical feedback, and stay in touch with their physician. The AI used is Mistral 7B OpenOrca.

---

Features

• AI-Assisted Journaling – Patients can document their daily experiences and receive motivational feedback from an LLM.
• Personalized AI – Doctors can configure AI responses to provide clinical insights and tailor support to each patient.
• Intuitive User Interface – A web application with dedicated patient and doctor dashboards.
• Secure Data Management – Uses PostgreSQL for structured data storage.
• Advanced NLP Processing – Powered by Mistral-7B, running locally with llama_cpp.
• Multi-User Access – Patients and doctors have separate roles and functionalities.

---

Tech Stack

• Backend: Django
• Frontend: HTML, CSS, JavaScript
• NLP: Mistral-7B (Quantized GGUF) via llama_cpp
• Database: PostgreSQL

---

Installation Guide

1️. Clone the repository

git clone https://github.com/FLaTNNBio/SoulDiaryConnect.git
cd SoulDiaryConnect

2. Set up a virtual environment

python3 -m venv venv
source venv/bin/activate  # on Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Configure the database

Install PostgreSQL following the official guideline.
To exectute the queries:

python manage.py dbshell

Then:

\i souldiaryconnect.sql

Now edit setting.py to configure PostgreSQL:

DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.postgresql',
        'NAME': 'souldiaryconnect',
        'USER': 'your_user',
        'PASSWORD': 'your_password',
        'HOST': 'localhost',
        'PORT': '5432',
    }
}

Make database migrations:

python manage.py makemigrations

Run database migrations:

python manage.py migrate

5. Load the NLP model (Mistral-7B)

Download the Mistral-7B (quantized GGUF) model and place it in the 'models/mistral/' directory.

Update the model path in views.py and edit the context size (n_ctx value):

model_path = "./models/mistral/your_mistral-7b-openorca.Q8_0.gguf"
llama_model = Llama(model_path=model_path, n_ctx=2048)

6. Start the server

python manage.py runserver

Roles & Functionality

Doctor

• Manage patients – Access and review patient journal entries.
• Customize AI responses – Configure the AI to tailor feedback generation.
• Monitor therapy progress – View clinical trends and intervene when necessary.

Patient

• Write personal journal entries – Document daily thoughts and emotions.
• Receive AI-generated feedback – Get motivational and therapeutic insights.
• View therapist's comments – See personalized feedback from the doctor.