Panasonic develops care AI '予兆ソリューション' with 89.4% recall

Panasonic Holdings framed the "予兆ソリューション" as a response to two pressures named in its announcement: an aging population and a shortage of care workers, which it says make it a challenge to maintain and improve care quality with limited staff. The system combines a health-change analysis AI, which reads biometric and daily-life data from monitoring sensors against each person's normal state, with a report-generation AI that lays out the grounds for a detected change and the points staff should observe, displayed as graphs and text for sharing across care and nursing roles.

The accuracy figures come from an evaluation of about 2,000 people. The analysis used daily statistical analysis plus the differences from the previous day and the day before as features, which the company says made it possible to extract information that could serve as a trigger for checking a resident's condition, including changes that are hard to catch by eye or through self-reporting. The solution is also positioned as a common platform that does not depend on a particular monitoring service or sensor, normalizing and extracting features from sensor signals that differ by device.

Going forward, Panasonic plans to use what it learns from the trial to improve its analysis methods and report content, and to expand collaboration with outside partners including monitoring service providers, aiming toward data-based care that does not rely only on intuition and experience. Whether that wider use materializes appears likely to hinge on how the October field trial with ユーアイ二十一 judges usefulness and operability in a live care setting.