Introduction

System to recognize accidental falls reliably. The procedure, based on accelerometric data to which characteristic motor activity indices are applied, eliminates false alarms and only signals events that are not attributable to normal daily activities. Low consumption and simplicity allow it to be implemented in non-invasive wearable devices for everyday use.

Technical features

The system includes sensors for detecting the movement of a user’s body and a fall recognition module, capable of eliminating false alarms due to normal daily activities. The system detects anomalous movement data (e.g., acceleration) and branches them to a series of decision modules. Each module evaluates whether the input data can be interpreted as characteristic of a normal activity (false alarm) or not. If no decision module reports a false alarm, the device signals the detection of a user fall. The system is modular and configurable. In a particular configuration, the device is able to recognize as false alarms the act of sitting/lying down quickly on a soft surface, sitting on a hard surface, running, jumping. The system operates on the basis of thresholds and indices representative of daily activities, applied to the acceleration module.

Possible Applications

• Wearable systems for continuous monitoring, indoor and outdoor, of subjects exposed professionally at risk of falling;
• Integration in remote assistance services of elderly people;
• Development of personal healthcare applications for smart devices.

Advantages

• High sensitivity (ability to detect all real falls) and specificity (ability to exclude all false falls);
• Possibility to work with only one accelerometer;
• Simple algorithm implemented on-board (does not need to communicate data externally);
• Low energy consumption.