Back to Guru's profile
16views

Q-MED TRIAGE INDIA : QUANTUM POWERED MULTIMODAL HEALTHCARE INTELLIGENCE

“Smarter Insights. Faster Triage. Better Quantum Hybrid AI.”

Guru Gourav Panda
  • Video of it working
  • Problem, approach and result explained
  • Expert reviewed· 6 reviews
Q-MED TRIAGE INDIA : QUANTUM POWERED MULTIMODAL HEALTHCARE INTELLIGENCE

92%

audio recordings

90%

diseases flagging and symptoms and medicine recommendation

90%

follow up question matrix

Walkthrough

Overview

The problem

Q-MedTriage India – 10 Important Problem-Based Questions and Answers

1. What problem are we trying to solve?

Healthcare facilities face high patient volumes, manual data collection, language barriers and delays in organizing patient information for medical review.

2. Who are we solving this problem for?

Government hospitals, PHCs, doctors, nurses, frontline health workers, public health camps, industrial clinics and campus health centres across India.

3. Why is the current patient intake process challenging?

Healthcare workers often need to manually collect symptoms, review medical reports and understand patient histories, which can consume valuable time.

4. How can patients communicate their symptoms easily?

Through text and voice input, with support for English, Hindi and regional languages.

5. How can we reduce manual medical report analysis?

By using OCR to extract relevant information, such as laboratory values, test results and reference ranges from uploaded reports.

6. How can we ensure important patient information is collected?

By analyzing available symptoms and medical history, identifying missing information and generating relevant follow-up questions.

7. How can we help healthcare workers prioritize patients?

By highlighting potential urgency signals using clinician-reviewed safety rules and supporting patient queue prioritization.

8. How does the system help doctors?

It prepares structured patient summaries containing symptoms, medical history, vital signs, report information and follow-up responses.

9. How will we ensure patient safety and privacy?

Through human verification, emergency escalation, patient consent, anonymization, secure data handling and role-based access.

10. What is the expected outcome of this project?

To make preliminary healthcare assessment more organized, accessible and efficient, helping healthcare professionals review patients faster without replacing their clinical judgment.

Our central objective:

"To bridge the gap between patients and healthcare professionals by using AI to collect, understand, organize and prioritize patient information, enabling faster and more informed medical review."

The approach

Our approach is to develop Q-MedTriage India, a multimodal, AI-assisted healthcare triage system that collects patient information through text and voice, extracts details from medical reports using OCR, and supports multiple Indian languages. It integrates patient history, symptoms, vital signs and laboratory results to create structured patient summaries. The system uses TabTransformer, CNN, Bi-LSTM, WAV2LLM and a 12-qubit Variational Quantum Circuit (VQC) to explore multimodal data analysis and risk prioritization. It identifies missing information, asks relevant follow-up questions and highlights potential urgency signals. A healthcare professional reviews all outputs and makes final decisions. The system supports clinical workflows but does not independently diagnose or prescribe treatment.

The result

The expected result of Q-MedTriage India is a user-friendly, multimodal healthcare triage prototype that collects and organizes patient information from text, voice and medical reports. It will generate structured patient summaries, identify missing information, ask relevant follow-up questions and highlight potential urgency levels to support healthcare staff in prioritizing patients. The system will provide a reviewer dashboard, organized medical history and referral-ready triage notes. Its AI models will be evaluated using training, validation and testing datasets to measure performance. The final outcome will be a human-supervised decision-support tool that aims to improve preliminary assessment efficiency, information accessibility and patient prioritization without replacing medical professionals or independently diagnosing patients.

Reflection

Key Differences Between Existing Models and Our Proposed Q-MedTriage India

1. Multimodal Input: Supports patient information through text, voice and uploaded medical reports rather than relying mainly on manual text entry.

2. Regional Language Support: Designed to support English, Hindi and regional Indian languages.

3. OCR Integration: Extracts relevant information from medical reports automatically, reducing manual data entry.

4. Hybrid AI Architecture: Combines TabTransformer, CNN and Bi-LSTM to process structured patient data and sequential information.

5. WAV2LLM Audio Processing: Integrates an audio-processing approach to convert patient speech into useful information.

6. 12-Qubit Quantum Model: Introduces a 12-qubit Variational Quantum Circuit (VQC) using PennyLane alongside classical machine-learning models.

7. Primary and Secondary Disease Classification: Handles primary disease classification and secondary multi-label disease identification as separate tasks.

8. Adaptive Follow-up Questions: Identifies missing patient information and generates relevant follow-up questions.

9. Patient History Integration: Combines symptoms, age, BMI, blood pressure, heart rate, temperature, laboratory results and medical history.

10. AI-Assisted Patient Prioritization: Highlights potential urgency signals to help healthcare workers organize patient queues.

11. Human-in-the-Loop Review: Keeps doctors and qualified healthcare professionals responsible for reviewing information and making final decisions.

12. India-Focused Healthcare Design: Targets government hospitals, PHCs, public health camps, company clinics and campus health centres.

Main innovation: Combining multimodal AI, hybrid classical–quantum learning, regional language support and human-supervised patient triage in one healthcare platform.

Note: These are proposed technical differences. Their actual advantages must be demonstrated through experiments and comparison with existing models.

Submitted to The Multimodal Triage Assistant

Your idea in one line

Q-MedTriage India leverages Quantum Machine Learning (QML) with a four-qubit Variational Quantum Circuit (VQC), multimodal AI, and EHR quality-control pipelines to standardize complex medical data, handle missing information, and generate structured, human-reviewed triage insights while experimentally evaluating quantum–classical performance against classical AI.

What will be working by the finale?

The final outcome of Q-MedTriage India is a multilingual healthcare triage-support system powered by a multimodal hybrid 12-qubit Quantum Machine Learning (QML) model, integrating text, voice, medical reports, and images. It standardizes EHR data, handles missing information, and combines classical AI with Variational Quantum Circuits (VQCs) to analyze patient information and experimentally evaluate predictive performance against classical models. The system generates structured triage notes, highlights potential warning signs, and identifies incomplete information to support efficient clinical review, while qualified healthcare professionals retain responsibility for all final clinical decisions.

Tech stack and hardware you plan to use

The Q-MedTriage India project uses React.js, Vite, and Tailwind CSS for frontend development, Python and FastAPI for backend services, and PostgreSQL for database management. EHR data quality and standardization are handled using Pandas, Great Expectations, OHDSI ATHENA, ACHILLES, and OMOP CDM workflows, with dedicated preprocessing logic for missing data. Hugging Face Transformers, spaCy, and TabTransformer support text processing; Whisper and CNN-BiLSTM handle voice processing; Tesseract OCR or EasyOCR extract medical report data; and OpenCV with EfficientNet-B0 processes images. PyTorch and scikit-learn support classical machine learning, feature preparation, and evaluation, while PennyLane or Qiskit implements the four-qubit Variational Quantum Circuit (VQC). PyTorch and NumPy enable multimodal feature fusion, while Python-based safety rules, Pydantic, JWT authentication, role-based access control, and audit logging support validation, security, and human review. Finally, Docker, Uvicorn, and GitHub support deployment and version control. The system will be evaluated using synthetic or public datasets, with qualified healthcare professionals retaining responsibility for final clinical decisions.

What have you built or tested so far?

For the time being we have successfully tested a hybrid model for T2DMS using Tab -Transformer and a 4 bit VQC and the audio to text model is working properly for 8 Indian language and currently working on visual and medical report input models.

Pitch deck (PDF)
Pitch deck (PDF)PDF
What's working now?

In my current model, the speech to text and text to speech is working with analysis of medical records using ocr ,appointment with doctors ,blood bank inventory and proper triage and emergency system for flagging the diseases and symptoms

Deck or report (PDF)
Deck or report (PDF)PDF
A photo or screenshot
A photo or screenshot: Q-MED TRIAGE INDIA : QUANTUM POWERED MULTIMODAL HEALTHCARE INTELLIGENCE

Gallery

3