Deepfens AI Multimodal Edge AI Fall Detection Runs on Avalue's Medical-Grade Platform

A privacy-first answer to falls in aged care

As Taiwan enters a super-aged society — over 20% of the population aged 65+ — hospitals and long-term care institutions face growing staff-shortage and safety-management pressure, with falls among the most common elderly accidents. Bathrooms and restrooms are the highest-risk areas yet the hardest to monitor with traditional cameras due to privacy. AI startup Deepfens AI has launched what it describes as the market's first multimodal edge AI fall-detection system for exactly these environments, partnering with Avalue Technology and its medical-grade edge computing platform to deliver a complete smart-healthcare solution.

How the system works

Instead of camera surveillance or wearables, the system uses millimetre-wave radar non-contact sensing to read human posture and motion in real time — detecting falls or abnormal behaviour without cameras, preserving privacy. It adds voice interaction: on detecting an anomaly it can speak to the user to confirm their condition and reduce false alarms. Combining multiple AI modalities — body movement, positional change and environmental information — improves accuracy. An Edge AI architecture runs inference locally on the device, keeping sensitive data off the cloud for privacy and low-latency response, and over time analyses behavioural patterns (daily activity, nighttime mobility, prolonged inactivity) to give care teams useful insight.

Avalue's medical-grade hardware

Avalue integrates the AI sensing system with medical-grade embedded computers and edge platforms so facilities can deploy quickly — a one-stop path from sensing devices and AI analytics to care-management platforms, covering fall detection in patient rooms and bathrooms, in-hospital safety, post-discharge home monitoring and chronic-patient activity tracking. The showcase features the compact AI Box PC MAB-T660D — Intel 14th Gen Raptor Lake Refresh on the Intel H610E platform, up to 64GB DDR5 and PCIe Gen4 x16 for AI accelerator cards — alongside the HPS-GNRU4A high-performance server, powered by Intel Xeon 6, supporting four dual-width GPUs with PCIe Gen5 for flexible, high-speed computing expansion and centralised management.

Where it's headed

Deepfens AI expects smart-care solutions to expand into home healthcare and smart-city applications, using AI sensing and behavioural analytics to safeguard elderly and patient safety without disrupting daily life while giving healthcare systems richer behavioural-health data. The solution was showcased at the 2026 Smart City Summit & Expo (March 17–20, AI Zone Pavilion, Booth Q1023).

Availability

TSL Automation Solutions supplies Avalue worldwide. The MAB-T660D AI Box PC and HPS-GNRU4A server are listed on our site. Contact our team to scope a medical-grade edge-AI platform for smart healthcare or long-term care.