NANIcept

Nurse Dashboard (Ward)
Bedside Dashboard (Ward)
Patient Tracker for Family Members

🌟 Tagline

Because pain shouldn't have to be spoken to be heard.

🗣 Elevator Pitch

Ultra-short (10s):
A multi-sensor system that detects pain in non-verbal patients — giving caregivers a new sense: the ability to hear what cannot be spoken.

Standard (20–25s):
NANIcept is a speculative multi-sensor system designed to detect pain in patients who cannot communicate — such as stroke survivors, dementia patients, or sedated ICU patients. The system combines facial micro-expression analysis, physiological signals like heart rate variability, and movement patterns from smart bed sensors to compute a real-time Pain Presence Score. When pain is detected, caregivers are alerted instantly and families receive clear, human-friendly updates. Our goal is to give a voice to patients whose pain would otherwise remain invisible.

Inspiration

Pain is one of the most private human experiences. Yet in hospitals, thousands of patients live through it completely alone — not by choice, but because they cannot speak.

A friend on our team experienced this firsthand:

Their grandmother(NANI) had a stroke. She lost the ability to speak, gesture, or point to where she hurt. During her hospital stay, she suffered a hairline femur fracture that went undetected for hours because she couldn’t communicate her pain.

She wasn’t neglected — nurses cared deeply. The system just wasn’t built to hear her.

What if caregivers could perceive pain even when a patient cannot express it?

Many patients fall into this category:

🧠 Stroke survivors
🧓 Dementia patients
🏥 Intubated ICU patients
🧬 Individuals with severe neurological conditions
👶 Infants

Their pain becomes invisible. We built NANIcept to change that.

The name comes from the Hindi word Nani — meaning grandmother and Nociception (Sense of Pain), a reminder that behind every patient chart is someone deeply loved.

What it does

NANIcept gives caregivers a new sensory capability: the ability to perceive pain when a patient cannot express it.

Instead of relying on periodic observation or self-reported pain scales, NANIcept continuously monitors the body's involuntary signals, the ones that tell the truth even when words can't.

Four signal streams:

Facial micro-expressions → Brow tension, eye tightening, cheek raising
Muscle guarding & movement resistance → Smart bed-integrated sensors detect involuntary flinching and asymmetry
Heart Rate Variability (HRV) & electrodermal activity → Wristband sensor captures autonomic stress
Neural activity proxies (fNIRS) → Near-infrared spectroscopy signals associated with nociceptive processing

These signals are fused to create a Pain Presence Score (PPS) in real-time:

PPS = f(Ef, Mg, HRV, Nb)
Ef  = facial expression signals  
Mg  = muscle guarding patterns  
HRV = autonomic stress indicators  
Nb  = neural activity proxies

When pain is detected, NANIcept:

🚨 Alerts caregivers immediately
💌 Notifies family members in human-friendly language
🛌 Initiates gentle environmental responses
📈 Logs a Pain Timeline for medical review

Over time, it learns each patient’s baseline to improve accuracy.

How we built it

Smart Bed Sensor Layer
- Detects muscle guarding, flinching, and asymmetrical movement
- Avoids wearables that interfere with IV lines or equipment
Vision-Based Expression Detection
- Facial micro-expressions analyzed on-device
- No video is stored; only scores leave the device
Sensor Fusion Engine
- Correlates multi-stream signals over time
- Detects pain-specific cascades rather than single spikes
- Differentiates pain from fear, exertion, or startle responses
Multi-Surface Interface
- Caregiver Dashboard (Tablet): PPS gauge, body-map localization, Pain Timeline, one-tap alert acknowledgment, messaging
- Family Mobile App: Emotional updates, color-coded history, nurse messaging, no raw numbers
- Admin Dashboard (Desktop): Full Pain Timeline, exportable reports, cross-patient analysis
- Patient Wristband: LED + haptic pulse indicating current PPS

Challenges we ran into

False Positives → HRV spikes from anxiety, movement from exertion, or facial changes from fever mimic pain. Solved with multi-stream correlation requiring ≥2 signals to trigger alert.
Ethics & Privacy → Consent management, on-device processing, family sees summaries only, nurse override, auto-escalation, data deletion after discharge.
Two User Types → Nurses need clinical dashboards; families need emotionally supportive updates.
Hardware Feasibility → Wearables impractical for ICU patients, so sensors were integrated into the bed itself.

Accomplishments that we're proud of

✅ Built a working multi-sensor prototype demonstrating real-time PPS
✅ Designed dual-layer communication for caregivers and families
✅ Implemented ethical safeguards from day one (consent, privacy, transparency)
✅ Conceptually aligned with state-of-the-art research in multimodal pain detection and affective computing
✅ Created a human-centered workflow that research prototypes often ignore

What we learned

Healthcare technology is human-first, technical-second
Trust is harder to engineer than sensors
Data alone is not enough; information must reduce anxiety
Designing for vulnerable patients requires careful consideration of consent, autonomy, and power dynamics
Speculative design uncovers invisible gaps in real-world healthcare systems

Pain should never go unheard simply because someone cannot speak.

What's next for NANIcept

Expand signal modalities (EEG, advanced fNIRS) for even more accurate pain inference
Conduct real-world usability studies with clinicians and families
Explore integration with hospital EMRs for seamless workflows
Research predictive analytics to anticipate pain events before they occur
Refine the family communication app to support personalized, context-aware updates