Researchers at Nationwide Institute of Technology Rourkela have developed an evolved multi-sensor, AI-enabled system able to precisely detecting human sleep postures, offering a non-intrusive solution for patient monitoring whereas conserving privacy.
The detect, printed within the IEEE Sensors Journal, was as soon as performed by Saptarshi Chatterjee, Assistant Professor within the Department of Electronics and Dialog Engineering, alongside with B.Tech scholar Shiladitya Mondal and Dr. Debangshu Dey of Jadavpur College.
Addresses Health Risks of Sad Sleep Posture
Scientific research display disguise that snide sound asleep posture can lead to serious prolonged-interval of time health concerns, including continual musculoskeletal anguish, spinal degeneration, obstructive sleep apnea, and nerve-connected complications. For bedridden patients, prolonged unhappy posture can outcome in tension ulcers or bedsores.
Unusual monitoring ideas, much like handbook enlighten, wearable devices, or camera-based mostly completely systems, veritably face obstacles including inconsistency, discomfort, high fees, and privacy concerns.
Multi-Sensor Manner Ensures Privateness
To conquer these challenges, the research group developed a system that integrates three somewhat about a sensing applied sciences:
- Lengthy-wave infrared sensors to detect body warmth with out capturing visible photos, even below blankets
- Depth sensors to attract body posture and shape
- Power sensors to analyse weight distribution on the mattress
The records from these sensors is processed the whisper of a generative AI mannequin combined with a graph-based mostly completely neural community to precisely title body posture and joint positions.
“Our system uses sensor fusion and avoids RGB imaging, enabling apt posture detection even in low light or when the body is roofed,” Prof. Chatterjee talked about.
Excessive Accuracy with Proper-World Doable
Laboratory tests have confirmed that the system can lift out an accuracy of around 98 per cent, making it appropriate for excellent deployment. Its non-contact nature lets in continuous monitoring with out anxious patients, whereas additionally cutting again the workload on caregivers.
The abilities would possibly perchance perchance well be embedded into sanatorium beds to video display patients, aged folks, and these tormented by sleep-connected concerns much like sleep apnea.
Furthermore Be taught: NIT Rourkela researchers get patents for evolved system to detect intruders in restricted zones
Sensible and Scalable Resolution
The estimated payment of the system is around ₹30,000, with attainable for reduction by strategy of broad-scale production. Researchers think the abilities would possibly perchance perchance well be built-in into multi-modal healthcare monitoring systems and prolonged for identifying posture-connected health instances.
The group is now working on refining the system for broader capabilities, including home-based mostly completely care, which would possibly perchance perchance a good deal give a enhance to patient monitoring beyond sanatorium settings.
