MEDICONNECT

Framing the Problem

What is the problem you are trying to solve? Who does it affect?

Modern healthcare systems face increasing pressure from rising patient volumes, extended ER wait times, and inefficient triage processes that often delay care. Patients frequently struggle to determine the seriousness of their symptoms like whether to seek emergency attention, book an urgent care appointment, or schedule a routine checkup. Without clear guidance, this uncertainty results in unnecessary ER visits, delayed interventions, and anxiety for patients seeking timely help.

On the other hand, doctors and healthcare facilities experience challenges in managing patient flow and optimizing their schedules. Many healthcare providers lack visibility into real-time patient demand or opportunities to fill appointment cancellations quickly. This disconnect between patients seeking help and providers with available capacity leads to systemic inefficiency across both sides.

MEDICONNECT bridges this gap. The platform serves both patients and healthcare professionals, providing a unified, intelligent interface that mirrors the real-world triage experience.

On the patient side, the app allows users to input symptoms, insurance, and location details before being connected to Eva, a virtual AI nurse. Eva performs an interactive, conversational assessment asking brief, relevant follow-up questions to understand symptom severity and urgency. Once complete, Eva classifies the condition as:

- Emergency – Advises immediate ER visit or 911 call.

- Urgent – Directs to nearby urgent care centers.

- Routine – Offers the choice to Find a Specialist or Enter the Smart Queue.

The Find a Specialist feature displays available doctors filtered by specialty, distance, and insurance network. If no slot is immediately available, the Smart Queue allows patients to consent to share anonymized health data with all relevant doctors in the region. Doctors with last-minute availability can view these cases and claim them, ensuring faster access to care.

On the doctor side, MediConnect provides an organized dashboard with two sections - Direct Requests and Smart Queue. Each patient card includes symptom descriptions, AI Nurse Eva’s assessment, condition keywords, confidence scores, and urgency classification. This data allows physicians to make quick, informed decisions about accepting cases that fit their specialty or schedule.

By connecting both users through an AI-assisted triage interface, MediConnect transforms scattered patient experiences into a streamlined, intelligent, and proactive healthcare ecosystem.

Idea Explanation

What is your idea? How does it fix the problem?

MediConnect is an AI-driven healthcare platform designed to simplify triage, accelerate access to care, and create an intelligent bridge between patients and doctors.

The system centers around Eva, a conversational AI nurse powered by the Gemini API, capable of understanding natural speech and typed input. Patients can describe their symptoms verbally or through text, and Eva engages in a short dialogue of follow-up questions to refine understanding like mimicking how nurses gather information in real clinical settings. After evaluating the responses, Eva classifies the case into one of three categories—Emergency, Urgent, or Routine—and provides recommendations accordingly. For non-critical cases, patients can:

- Use Find a Specialist to view doctors filtered by ZIP code, distance, specialization, insurance coverage, and appointment slots.

- Join the Smart Queue, where their anonymized triage card is shared with qualified specialists citywide, enabling doctors to “claim” open cases in real time.

Doctors access an intuitive dual-column dashboard displaying both Direct Requests (patients who selected them) and Smart Queue cases (AI-recommended anonymized profiles). Each record includes a structured evaluation—symptom keywords, AI verdict, confidence score, and insurance information—helping doctors prioritize and manage time efficiently.

To ensure privacy and compliance, all data in the current system is mock data stored locally, adhering to HIPAA guidelines for secure handling of health information.

Through conversational triage, intelligent filtering, and doctor-matching algorithms, MediConnect delivers a near real-world healthcare experience using Gemini-powered intelligence, voice interaction, and ethical data simulation.

Implementation

How do all the pieces fit together? Does your frontend make requests to your backend? Where does your database fit in?

The MediConnect architecture is designed to be modular, scalable, and HIPAA-conscious. It integrates AI-driven reasoning, user-friendly interaction, and efficient data management—all operating through a client-side React application.

Frontend

- Built entirely with React.js, ensuring modularity and dynamic updates.

- Includes separate modules for:

1. AI Chat Interface (Eva) – handles user inputs and conversational logic.

2. Voice Input Panel – allows patients to communicate via voice or text. [Voice module powered by browsers Speech recognition API]

3. Smart Queue & Doctor Cards – displays doctor availability and filters results by specialty, location, and insurance.

4. Doctor Dashboard – dual-pane interface showing Direct Requests and Smart Queue cases with full AI reports.

AI Layer

- Powered by Google’s Gemini API, Eva interprets symptoms, follows up with adaptive questioning, and outputs a triage verdict.

- The AI uses contextual reasoning to remember previous user responses, avoiding redundant questions.

- Final outputs include:

1. Urgency level classification (Emergency / Urgent / Routine)

2. Condition keywords and confidence scores

3. Recommendation for next action (ER, urgent care, or specialist)

Data Layer

- All data used is mock data, stored locally in .ds or .json files to ensure HIPAA compliance.

- Includes:

1. Doctor datasets (specialties, insurance, urgent slots)

2. Urgent care centers (mock geolocations)

3. Anonymized patient cards (for Smart Queue sharing)

- Data filtering and ranking are handled entirely on the client side, without requiring backend servers.

System Workflow

1. Patient Interaction – User enters or speaks symptoms, insurance, and ZIP code.

2. AI Consultation – Eva asks follow-up questions and classifies urgency.

3. Triage Verdict – System recommends action (call 911, visit urgent care, or proceed with specialist options).

4. Doctor Matching – Patients view or send anonymized cards via the Smart Queue.

5. Doctor Review – Doctors view cards, AI summaries, and confidence levels before claiming cases.

Scalability & Future Integration

The frontend-only architecture is designed for easy transition to a cloud-based backend (Node.js or Firebase). Future versions can add authentication, real-time syncing, and external API integration without major redesigns.

Challenges

What did you struggle with? How did you overcome it?

Building MediConnect was as much about teamwork as it was about technology. While we faced a variety of design and development hurdles like from integrating AI responses to designing a fluid interface. The biggest challenge was ensuring that all the moving parts worked seamlessly together.

Since the system had multiple components such as AI conversation, mock data management, patient and doctor dashboards, and voice interaction; it required precise coordination across our disciplines. Each team member brought a different technical strength, and through continuous testing and iteration, we aligned the workflows into one unified system.

For instance, when the AI logic and user interface initially didn’t synchronize well, our team collectively redesigned the data flow to make Eva’s triage responses appear smoother and more human. Similarly, when linking doctor cards and AI evaluations, we worked together to refine the JSON structure so that every module communicated efficiently.

In short, every challenge we encountered became an opportunity for collaborative problem-solving. By blending our expertise in Data Science, CSE, ECE, and Biomedical Informatics, we learned how interdisciplinary thinking could turn a complex concept into a working, intelligent healthcare platform.

Accomplishments

What did you learn? What did you accomplish?

Technical Milestones

- Developed a Gemini-powered conversational triage assistant (Eva) capable of contextual reasoning and adaptive questioning.

- Built a React-based two-sided system for patients and doctors with live data filtering.

- Implemented Smart Queue, an intelligent matching feature connecting patients with doctors who have sudden slot availability.

- Integrated voice input to enable natural speech interactions.

- Created a doctor dashboard showcasing AI evaluation summaries, confidence scores, and anonymized patient data.

- Achieved seamless mock data handling without a backend, simulating real-world hospital and insurance networks.

Cross-Domain Collaboration

The multidisciplinary collaboration between Data Science, CSE, ECE, and Biomedical Informatics members made the project technically robust and medically credible.

- Data Science optimized the AI and triage logic.

- CSE/ECE handled frontend state management and voice integration.

- Biomedical Informatics ensured medical relevance and ethical data handling.

Learning & Insights

Through iterative development, the team learned that user experience in healthcare AI depends as much on clarity and empathy as on accuracy. Short, relevant AI follow-ups fostered trust, while structured output summaries enhanced transparency for both users and doctors.

Next Steps

What are the next steps for your project? How can you improve it?

1. Real Data & Secure API Integration

Future iterations will move beyond mock datasets to integrate:

- Real-time hospital and urgent care APIs for verified availability.

- Mapping and geolocation APIs to calculate accurate travel distances.

- Insurance verification APIs for real coverage validation.
All integrations will be governed by HIPAA-compliant frameworks to ensure patient privacy.

2. Analytics and Data Visualization

The next milestone involves building an AI and performance analytics dashboard, including:

- Doctor responsiveness, case acceptance, and average consultation delay.

- Real-time triage accuracy tracking and system-wide efficiency metrics.

- Heat maps identifying common symptom trends by region or season.

3. Personalization and Multi-Language Support

- Future updates will add multi-language dialogue, enabling Eva to assist patients in multiple languages.

- The app will also store consent-based patient histories, allowing for personalized care suggestions and improved continuity.

4. EHR Integration

MediConnect will connect directly with Electronic Health Record (EHR) systems, enabling automatic transfer of triage reports and AI evaluations into clinical workflows, streamlining doctor preparation and patient intake.

5. Pilot Testing and Deployment

- The long-term plan is to collaborate with UF-affiliated medical centers to pilot the system in a controlled environment.

- This real-world deployment will allow the team to evaluate usability, scalability, and model performance bringing MediConnect one step closer to real clinical implementation.