HUMAN–AI COLLABORATION IN HEALTHCARE

NovaCare
Symptoms First.
Faster Care.

Patients submit symptoms through NovaCare before they call 811 or visit the hospital. By the time a healthcare practitioner connects with them, the intake is already done — the conversation is about follow-up, not data collection.

Pre-Visit
Symptom Intake
< 1 mo
Idea to Prototype
6
Team Members
10+
Deliverables

From Challenge NS to the Classroom

This project began as a Challenge Nova Scotia 2025 submission (Team 59, November 14) on the theme of Human–AI Collaboration in Healthcare. We explored four AI-in-healthcare ideas, then scoped down to one actionable concept and built it as a full class project in under a month.

💡

Challenge NS Infographic

Our original submission explored four Human–AI collaboration ideas for Nova Scotia healthcare: robotic surgery, telemedicine, predictive care, and early disease detection — positioning the province as a leader in modern care delivery.

🎯

Scoped to One Idea

We narrowed to a single concept: a pre-visit symptom intake tool that sends structured patient data to healthcare practitioners — so calls and visits become about follow-up, not repetitive intake.

👥

Team 59

Six NSCC IT Campus students — spanning Business Intelligence & Analytics and Information Technology — brought together skills in BA, BI, development, and UX design.

Juliana Ayide Chibuzor Opara Morgan Garnier Ann Mwangi Edward Murray Michael Okafor

Built on Lovable

With no dedicated web developer on the team, we used Lovable.dev to prototype the app rapidly. The PRD, decision trees, and use cases drove the build — Human–AI collaboration in practice.

A Healthcare System That Repeats Itself

Problem Statement

Nova Scotia's healthcare system wastes time at every point of contact. When a patient calls 811, the first 10–15 minutes are spent collecting basic information — name, age, symptoms, duration, severity. When that same patient walks into an ER, the intake process starts over from scratch. The practitioner has no advance context. Every interaction begins at zero.

Meanwhile, the system is already strained. ER wait times average 3.5 hours and exceed 10 hours in congested areas. The 811 tele-nurse service reports callback delays of over 4 hours. Over 87,000 Nova Scotians (8.2%) sit on a waitlist for a family doctor with no primary care pathway at all. Non-urgent patients fill emergency departments alongside urgent cases because there's no mechanism to sort before arrival.

The core inefficiency is simple: symptom intake happens live, in real time, consuming practitioner minutes that should be spent on clinical follow-up and care decisions. NovaCare addresses this by moving symptom collection upstream — patients submit their information through the app before the call or visit, so when a practitioner connects, the conversation starts at follow-up, not "tell me what's wrong."

3.5 hrs
Average ER wait time — up to 10+ hours in congested areas
4+ hrs
Reported 811 tele-nurse callback times
87,000+
Nova Scotians on the waitlist for a family doctor (8.2%)
39 days
Wait time for the Ambulatory Care Clinic
🔁

Repeated Intake

Patients describe the same symptoms to 811, then again at triage, then again to the doctor. Every handoff restarts at zero.

Practitioner Time Wasted

Nurses and doctors spend the first minutes of every interaction on data collection instead of clinical assessment and care.

🏥

Blind Arrivals

Hospitals have no advance notice of who's coming or what they need. Every patient is a surprise — making resource planning impossible.

Without NovaCare

  • Patient calls 811 — spends 10–15 min describing symptoms
  • Gets callback hours later, repeats everything
  • Goes to ER — intake starts from scratch again
  • Doctor has zero context before walking in
  • Non-urgent and urgent patients mixed blindly
  • System has no data on incoming demand

With NovaCare

  • Patient inputs symptoms through the app at home
  • Structured data sent to 811 or the hospital ahead of time
  • 811 callback jumps straight to follow-up questions
  • Doctor knows who's coming and what to expect
  • Patients pre-sorted by urgency before arrival
  • System gets advance demand signal for resource planning

Symptom Intake, Before the Visit

NovaCare collects, structures, and forwards patient symptom data to healthcare practitioners — turning every 811 call and hospital visit from a cold start into an informed follow-up.

📋

Pre-Visit Symptom Submission

Patients enter demographics, temperature, symptoms, severity, and duration through a guided flow. AI-assisted follow-up questions refine the data — so the practitioner receives a complete picture, not a vague complaint.

📤

Data Forwarded to Practitioners

Structured symptom data is sent to the receiving end — whether that's an 811 tele-nurse, a clinic, or a hospital triage desk. The call or visit starts at follow-up, not intake.

🏥

Hospital & Pharmacy Locator

Find nearby hospitals with estimated wait times and pharmacies with hours and contact info. Geolocation-based search with map view and clickable phone numbers.

🚨

Red-Flag Emergency Detection

Critical symptoms — chest pain, severe bleeding, loss of consciousness — immediately bypass the intake flow and display emergency guidance with a direct 911 dial link. Safety first, always.

Human–AI Collaboration, Not Replacement

NovaCare is built on the principle that AI should augment human judgment — never replace it. The AI handles symptom parsing, follow-up question logic, and data structuring. The healthcare practitioner makes every clinical decision. The patient stays in control throughout, and every recommendation includes clear disclaimers and escalation pathways.

🧑‍💻

Patient Inputs

Symptoms, demographics, temperature, severity, duration — at their own pace, from home

🤖

AI Structures

Parses input, asks follow-up questions, flags red flags, packages data for the practitioner

🧑‍⚕️

Practitioner Acts

Receives pre-structured intake, skips to clinical follow-up, makes care decisions with full context

How It Works

Three stages — patient, AI, practitioner. Each does what they do best.

Patient Side
📱

Submit Symptoms

Open app → consent → enter demographics, temperature, and symptoms. AI asks 5–7 follow-up questions to refine the picture.

AI Layer
⚙️

Structure & Route

AI parses input, detects red flags, structures data into a clinical-ready summary, and routes it to the right care channel.

Practitioner Side
🩺

Follow Up, Not Intake

811 nurse or hospital doctor receives the structured summary. The conversation starts with "I see you're experiencing…" — not "tell me what's wrong."

01
Open App
02
Consent
03
Demographics
04
Temperature
05
Symptoms
06
AI Follow-Up
07
Data Sent to Practitioner

MoSCoW Requirements

Requirements scoped and prioritized to drive the initial proof of concept, with clear boundaries on what's in and out for V1.

Must Have

  • Digital symptom intake questionnaire
  • Structured data output for practitioners
  • Red-flag emergency detection
  • Secure data handling & encryption
  • Mobile-friendly interface
  • PHIA-compliant data handling
  • Medical disclaimers throughout

Should Have

  • Practitioner-side summary dashboard
  • Error handling & fallback
  • Exportable symptom summaries
  • Admin interface for questions

Could Have

  • Multi-language support
  • Dark mode interface
  • Optional user accounts
  • Booking system integration
  • Progress indicators

Won't Have

  • Full EMR integration
  • AI medical diagnosis
  • Offline functionality
  • Wearable device integration

Deliverables

A full BA lifecycle — from charter and stakeholder analysis through to a working prototype, gap analysis, and solution evaluation.

Project Charter & Scope

Goals, scope boundaries, data sources, and out-of-scope items. Web/app-based AI tool for pre-visit symptom intake — no direct diagnosis.

Stakeholder Register & RACI

Seven stakeholders mapped with power/interest levels. RACI matrix defines roles across a fictional implementation organization.

MoSCoW Prioritization

Requirements scoped into Must/Should/Could/Won't with pivot table for readability. Drove the proof of concept priorities.

Business Model Canvas

Key partnerships (NS Health, ER departments, 811), value propositions, customer segments, revenue streams, and cost structures mapped.

Process Flow — User Journey

End-to-end flow from app launch to data delivery to practitioners. Branching paths by symptom severity with red-flag emergency routing.

Use Cases (6 total)

Fully described interactions: Enter Symptoms, App Launch, Hospital Information, Pharmacy Information, and more — with preconditions, flows, and exceptions.

Product Requirements Document

17-section PRD covering personas, functional requirements, technical architecture, compliance (PIPEDA, PHIA), risk matrix, and phased roadmap.

Lovable Proof of Concept

Working web app built on Lovable.dev with React frontend and Supabase backend. Symptom intake flows, AI follow-up logic, and facility locators for Halifax/HRM.

Gap Analysis

Identified missing features across Pharmacies tab, Hospitals tab, and Pre-Screen flow — including broken directions, inaccurate wait times, and missing inputs.

Solution Evaluation

Assessed alignment with original charter. Most backend resources, APIs, and practitioner-side integration not yet implemented — but highly functional for one month's work.

Gaps & Next Steps

We documented what works, what doesn't, and where to take it next.

Known Gaps

  • No practitioner-side dashboard to receive symptom data
  • No integration with 811 systems for data handoff
  • Location locked to Halifax — no postal code or geolocation input
  • Directions buttons not functional on pharmacy & hospital tabs
  • Open/closed status and wait time data not live
  • No back-end database of ailments and symptoms
  • No data collection, reporting, or analytics pipeline

Future Roadmap

  • Practitioner dashboard for receiving and reviewing intake data
  • Integration with 811 and hospital triage systems
  • AI agent for smarter follow-up questions & symptom structuring
  • Power BI dashboard for operational analytics
  • Hospital & pharmacy coverage beyond HRM
  • Google Maps integration for live directions
  • Multi-language support (French, Arabic, Mi'kmaq)

Human–AI Collaboration, Built for Nova Scotia

From a Challenge Nova Scotia infographic to a working pre-visit symptom intake app — in under one month. When patients arrive informed and practitioners start at follow-up, the whole system moves faster.

Try NovaCare ↗