The latest articles on DEV Community by Abinesh S (@abinesh-0). https://dev.to/abinesh-0 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4170785%2F4c56037a-c30d-4663-b52a-5342d3fd3bdc.jpg https://dev.to/abinesh-0 en Abinesh S Thu, 08 Oct 2026 09:17:43 +0000 https://dev.to/abinesh-0/medikiosk-system-1hhd https://dev.to/abinesh-0/medikiosk-system-1hhd <h1> 🏥 MediKiosk </h1> <h3> AI-Assisted Multilingual Patient Intake System </h3> <p>MediKiosk is a patient-facing digital clinical intake system designed to make the initial patient information-gathering process faster, more structured, multilingual, and accessible.</p> <p>Instead of asking every patient the same fixed set of questions, MediKiosk uses an <strong>adaptive clinical interview workflow</strong> to analyze the patient's input and generate relevant follow-up questions.</p> <p>The system supports <strong>Tamil and English</strong>, voice-assisted interaction, structured patient data collection, safety screening, and clinical summary generation.</p> <blockquote> <p>⚠️ <strong>Medical Disclaimer</strong></p> <p>MediKiosk is a clinical intake and decision-support prototype. It does not diagnose diseases, prescribe treatment, or replace a qualified healthcare professional.</p> </blockquote> <h1> 🎯 Problem </h1> <p>Traditional patient intake can be:</p> <ul> <li>Time-consuming for healthcare staff</li> <li>Repetitive</li> <li>Difficult to manage with large patient volumes</li> <li>Challenging for patients with language barriers</li> <li>Inconsistent when collecting important symptoms and history</li> <li>Difficult to convert into structured clinical information</li> </ul> <p>Patients may provide information in their own words, but important details such as duration, severity, associated symptoms, and relevant history may still need to be collected.</p> <h1> 💡 Our Solution </h1> <p>MediKiosk provides an AI-assisted conversational intake workflow.</p> <p>The patient can:</p> <ol> <li>Select <strong>Tamil or English</strong> </li> <li>Enter or speak their symptoms</li> <li>Receive an AI-generated relevant follow-up question</li> <li>Answer using text, predefined options, or voice</li> <li>Continue through multiple adaptive questions</li> <li>Receive safety/urgency screening</li> <li>Complete the intake</li> <li>Generate a structured clinical summary for healthcare staff</li> </ol> <p>The system focuses on <strong>information collection</strong>, not medical diagnosis.</p> <h1> 🧠 AI-Powered Adaptive Interview </h1> <p>The core feature of MediKiosk is its <strong>adaptive questioning system</strong>.</p> <p>Instead of following only a fixed questionnaire:</p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code> text Patient Input ↓ Analyze Patient Response ↓ Identify Missing Clinical Information ↓ Generate Relevant Follow-up Question ↓ Patient Answer ↓ Repeat if More Information Is Needed ↓ Clinical Summary Patient: "I have fever" ↓ MediKiosk: "How many days have you had the fever?" ↓ Patient: "3 days" ↓ MediKiosk: "Do you have cough, cold, or breathing difficulty?" ↓ Patient: "No" ↓ Continue collecting relevant information ↓ Structured Clinical Summary **🔄 System Architecture** ┌──────────────────────────────┐ │ Patient / Kiosk │ │ │ │ Tamil / English │ │ Text / Voice │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ Flask Web Server │ │ │ │ Session &amp; Intake Management │ │ Question Flow │ │ Language Management │ │ Safety Workflow │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ AI Service Layer │ │ │ │ Ollama │ │ Gemma 4 E4B │ │ │ │ Adaptive Questions │ │ Response Analysis │ │ Clinical Summary │ └──────────────┬───────────────┘ │ ▼ ┌──────────────────────────────┐ │ Data Layer │ │ │ │ SQLite / MySQL │ │ Patient Intake Data │ │ Session Information │ │ Structured Responses │ └──────────────────────────────┘ **🛠️ Technology Stack** | Layer | Technology | | -------------------- | ------------------------------------ | | Backend | Python | | Web Framework | Flask | | AI Runtime | Ollama | | AI Model | Gemma 4 E4B | | ORM | SQLAlchemy | | Database | SQLite / MySQL | | MySQL Driver | PyMySQL | | Frontend | HTML | | Styling | CSS | | Client Logic | JavaScript | | Voice | Browser-supported voice capabilities | | Development Platform | Windows | | Python | Python 3.x | **📁 Project Structure** MediKiosk/ │ ├── app.py ├── config.py ├── requirements.txt │ ├── models/ │ └── ... │ ├── routes/ │ └── ... │ ├── services/ │ └── ... │ ├── templates/ │ └── ... │ ├── static/ │ └── ... │ ├── demo_ui/ │ ├── app.js │ └── ... │ ├── VERIFY_PROJECT.py ├── JUDGE_PITCH.md ├── PROJECT_ARCHITECTURE.md │ ├── .env └── medikiosk.db </code></pre> </div> ai programming javascript python