App overview & operator

📱 App Overview

This app is an AI-assisted pharmaceutical consultation tool that recommends over-the-counter (OTC) medicines in a chat format based on users' symptoms, physical condition, and lifestyle.

By combining proprietary algorithms with large language models, we aim to safely and flexibly recommend appropriate over-the-counter medicines for symptoms, creating an environment where everyone can practice self-medication with confidence.

🎯 Development Background

With aging population, increasing foreign tourists, and the spread of e-commerce sites, the demand for self-medication is growing year by year. However, language barriers and staff shortages prevent users from selecting appropriate medicines, raising safety concerns. Having worked at a drugstore myself, I have faced challenges with elderly people's varying hearing and comprehension abilities and language barriers for foreigners. To solve these issues, I developed a unique chat-based consultation tool that combines large language models with pharmaceutical knowledge.

🎯 Purpose

This app aims to help users correctly understand their symptoms and safely select appropriate over-the-counter medicines.

Through chat-based dialogue, we present over-the-counter medicine candidates and consultation guidelines for symptoms to promote self-medication.

It is also designed to be used as reference information before visiting pharmacies or making online purchases, and plays a role in helping with early medical consultation decisions.

This app does not replace diagnosis or guidance by doctors or pharmacists, but is positioned as a tool to assist users in making safe decisions.

👥 Target Users

• General consumers who don't know which medicine to choose
• People who are too busy to visit pharmacies or residents of remote areas
• Foreign visitors who have difficulty consulting due to language barriers
• Elderly people and those with varying hearing and comprehension abilities
• Users considering purchases on e-commerce sites or online pharmacies

🎯 Key Features

1. Natural Chat-based Consultation
   Even without specialized knowledge, simply input symptoms in conversation format to get medicine recommendations.
2. Safety Assurance through AI × Pharmaceutical Knowledge
   Combines pharmaceutical databases, pharmaceutical knowledge, and AI models for safe design that suppresses misinformation.
3. Medical Consultation Recommendation System
   When dangerous symptoms or serious diseases are suspected, AI automatically recommends medical consultation.
4. Multi-language and Multi-environment Support
   Planned support for Japanese, English, Chinese, and other languages.
   Available on smartphones, tablets, PCs, and all devices and environments (iOS/Android/Windows/macOS/Chrome/Safari, etc.).
5. Secure Data Management
   Input information is anonymized and not used for purposes other than medicine recommendations.
   User privacy is the top priority in design.

💪 App Strengths & Differentiators

• Balance of safety and flexibility through combined use of AI and proprietary algorithms
• Fusion of evidence-based recommendations and natural language understanding (LLM) dialogue capabilities
• Direct solution to on-site challenges such as staff shortages, language barriers, and information gaps
• Operability that anyone can use without confusion through simple UI design and easy introduction

⚙️ Proprietary Algorithm

The "Medicine Selection Algorithm," which is the heart of this app, consists of a proprietary algorithm that flexibly understands language through large language models and comprehensively evaluates elements such as drug efficacy, contraindications, user attribute information, and symptoms.

This enables evidence-based medicine selection rather than simple AI responses. AI responses always include "source information" and "warnings" to allow users to make their own judgments.

🛠️ Development Environment & Tools

• Backend: Python 3.9+, FastAPI (production web/API, ASGI), Jinja2 (templates), MeCab (Japanese morphological analysis)
• AI/NLP: OpenAI API (GPT-4o, GPT-4o-mini, etc.), rule-based NLU (hybrid recommendation system)
• Translation & Speech AI: Amazon Translate (Japanese, English, Chinese, Korean), Amazon Polly (Neural TTS)
• Database: PostgreSQL (Feedback persistence, session management, multi-instance support)
• Data Processing: Pandas, NumPy
• Frontend: HTML5, CSS3, JavaScript (ES6+), Vanilla JavaScript (no framework), Responsive Design
• Deployment: AWS ECS Fargate (ap-northeast-1), Cloudflare Tunnel & R2; Gunicorn + uvicorn.workers.UvicornWorker (ASGI workers)
• Monitoring & Logging: psutil, JSONL format (structured logs), access analysis, performance monitoring
• Version Control: Git (GitHub)

🚀 Future Outlook

We aim to strengthen collaboration with pharmacies, medical institutions, and local governments to utilize this as a digital support platform for regional healthcare.

We also plan expansions such as e-commerce site integration and medication guidance support functions to provide value to both users and sellers.

Ultimately, the goal of this app is to realize a society where "everyone can safely choose medicines anywhere."

📚 Pharmaceutical Database Sources

The pharmaceutical information used in this app references databases from the following public institutions:

• Japan Pharmaceutical Manufacturers Association (http://www.fpmaj.gr.jp)
• Japan Pharmaceutical Information Center (JAPIC) (https://www.japic.or.jp)
• Pharmaceuticals and Medical Devices Agency (PMDA) (https://www.pmda.go.jp)

🎨 Image Copyright

The copyright of illustrations used in this app belongs to Sofukore (Individual Operator), Tegakiz (てがきっず), and shigureni.

👤 Operator Information