2026 Galleries

2026 Galleries // 2026 Spark: Spring Student // PuffO2

PuffO2

Winner - Silver

Competition: Spark: Spring Student
Designer: ChiaYi Liu
Design Type:
Company / Organization / School: National Taipei University of Education
Team Members: Chia-Yi Liu & Yen-Chun Chen; Prof. Kai-Chu Li

PuffO₂ is a smart smoke inhalation emergency mask that provides AI-guided first response for fire victims during the critical window between rescue and hospital arrival. Inhalation injuries can cause complete airway obstruction within 10–20 minutes, and mortality increases 24-fold when burns combine with inhalation damage. In Taiwan alone, over 3,000 high-risk locations experience average rescue delays of 5–10 minutes — a gap that existing simple oxygen masks cannot address.
The device consists of two main modules: a Protective Face Mask featuring a transparent display glass panel, air cushion seal, optical fiber triage indicator, and electronic metal frame; and an Oxygen & Medication Delivery Unit housing a bronchodilator port, pressurization motor, control module, battery, ultrasonic sensor, and oxygen canister. Materials are selected for extreme environments: medical-grade silicone for the seal, tempered glass for the face panel (non-porous, disinfectant-resistant, thermally stable), and medical-grade stainless steel for the delivery unit.
The AI workflow operates in four stages. First, dual sensors — an ultrasonic transducer detecting airway tissue vibration and deformation, and a microphone detecting respiratory acoustic waves — capture continuous physiological data. Second, ANC (active noise cancellation) filters fire-scene environmental noise, and the signal is digitized and amplified. Third, an onboard CNN model analyzes the combined waveform data and classifies the condition: Scenario A (upper airway edema with turbulent flow pattern) triggers positive pressure ventilation to mechanically stent the swollen airway; Scenario B (lower bronchial spasm with resonance pattern) triggers bronchodilator nebulization to relieve constriction; Scenario C handles mixed obstruction. Fourth, the paramedic reviews the transparent display — which shows vital alerts such as "CO HIGH / HR128 / RR32" or "Lower Airway Spasm" — and physically slides to administer treatment, maintaining full human authority over execution. An IoT cloud module logs usage history and supports continuous AI model improvement.
The triage fiber optic ring communicates status to the entire rescue team without verbal communication: red (Immediate), yellow (Emergency/transport), green (Urgent/stable). This allows a single paramedic to manage dual-patient scenarios — focusing on the critical patient while the second patient receives continuous automated monitoring under green status.
The modular disassembly system allows the portable oxygen canister to be swapped for a direct connection to the ambulance's onboard oxygen supply upon loading, confirmed on-screen as "VEHICLE O₂ LOCKED," eliminating supply interruption during handoff.
Feasibility was validated through expert interviews with the New Taipei City Fire Department (Second Brigade, Taishan Fire and Rescue Station), confirming that the monitoring and decision-support system would be valuable for EMT-P personnel, particularly in rural areas and narrow urban roadways where response delays compound risk.